feat: standardized elision markers and improve transcript viewer robustness

- Standardized elision markers across all tool outputs and filters to use cohesive `[...N [type] elided...]`, `[...Nln elided...]`, and `[...xB elided...]` syntax.
- Updated documentation, prompts, and test expectations to reflect the unified elision format.
- Improved transcript viewer robustness by preventing content aliasing through path-inclusive signature hashing.
- Added logic to clear stale transcript content when associated session files are deleted, accompanied by verifying test cases.

crates/pi-shell/src/minimizer/engine.rs

@@ -1097,10 +1097,15 @@ strip_lines_matching = [".*"]
 		}
 		let out = apply("rails db:migrate", &input, 0, &cfg);
 		assert!(out.changed);
-		// The def's max_lines=40 produces ONE truncation marker.
-		// filter_rake's head_tail marker ("lines omitted") must NOT appear.
-		assert!(!out.text.contains("lines omitted"), "double truncation detected: {:#?}", out.text);
-		assert!(out.text.contains("lines truncated"));
+		// Only the def's max_lines=40 cap should fire; filter_rake's head_tail must
+		// NOT also condense. Both now emit `[…Nln elided…]`, so a single marker
+		// proves no double truncation (two would mean both stages fired).
+		assert_eq!(
+			out.text.matches("ln elided…]").count(),
+			1,
+			"double truncation detected: {:#?}",
+			out.text
+		);
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/filters/binary_tools.rs

@@ -73,7 +73,7 @@ mod tests {
 		// 50 head + 20 tail + 1 marker = 71 lines
 		let line_count = out.text.lines().count();
 		assert_eq!(line_count, HEAD_LINES + TAIL_LINES + 1, "got {line_count} lines: {out:?}");
-		assert!(out.text.contains("lines omitted"));
+		assert!(out.text.contains("ln elided…]"));
 		// Head anchor preserved.
 		assert!(out.text.starts_with("00000000: 00000000"));
 	}
@@ -95,7 +95,7 @@ mod tests {
 		let context = ctx("strings", "strings /bin/ls", &cfg);
 		let out = filter(&context, &input, 0);
 		assert!(out.changed);
-		assert!(out.text.contains("lines omitted"));
+		assert!(out.text.contains("ln elided…]"));
 	}
 
 	#[test]
@@ -105,7 +105,7 @@ mod tests {
 		let context = ctx("od", "od -c /bin/ls", &cfg);
 		let out = filter(&context, &input, 0);
 		assert!(out.changed);
-		assert!(out.text.contains("lines omitted"));
+		assert!(out.text.contains("ln elided…]"));
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/filters/bun.rs

@@ -253,9 +253,9 @@ fn compact_bun_check_output(ctx: &MinimizerCtx<'_>, input: &str, exit_code: i32)
 	}
 	let omitted = nonzero_exits.len() + diagnostics.len();
 	if omitted > 40 {
-		out.push_str("… ");
+		out.push_str("[…");
 		out.push_str(&(omitted - 40).to_string());
-		out.push_str(" diagnostic lines omitted\n");
+		out.push_str(" diagnostic lines elided…]\n");
 	}
 	Some(out)
 }

crates/pi-shell/src/minimizer/filters/cloud.rs

@@ -474,7 +474,7 @@ fn compact_aws_generic(root: &Value) -> Option<String> {
 		let mut out =
 			compact_named_rows(&columns.iter().map(String::as_str).collect::<Vec<_>>(), &values);
 		if rows.len() > MAX_AWS_ROWS {
-			let _ = writeln!(out, "... +{} more {name}", rows.len() - MAX_AWS_ROWS);
+			let _ = writeln!(out, "[…{} {name} elided…]", rows.len() - MAX_AWS_ROWS);
 		}
 		return Some(out);
 	}
@@ -559,7 +559,7 @@ fn compact_named_rows(headers: &[&str], rows: &[Vec<String>]) -> String {
 		out.push('\n');
 	}
 	if rows.len() > MAX_AWS_ROWS {
-		let _ = writeln!(out, "... +{} more rows", rows.len() - MAX_AWS_ROWS);
+		let _ = writeln!(out, "[…{} rows elided…]", rows.len() - MAX_AWS_ROWS);
 	}
 	out
 }
@@ -761,9 +761,7 @@ fn compact_aws_dynamodb_items(items: &[&serde_json::Map<String, Value>]) -> Stri
 		out.push('\n');
 	}
 	if items.len() > 40 {
-		out.push_str("… ");
-		out.push_str(&(items.len() - 40).to_string());
-		out.push_str(" item(s) omitted …\n");
+		let _ = writeln!(out, "[…{} items elided…]", items.len() - 40);
 	}
 	let _ = writeln!(out, "{} item(s)", items.len());
 	out
@@ -1010,7 +1008,7 @@ fn compact_line(line: &str, max_chars: usize) -> String {
 	let edge = max_chars / 2;
 	let start: String = chars.iter().take(edge).collect();
 	let end: String = chars.iter().skip(chars.len() - edge).collect();
-	format!("{start} … {} chars omitted … {end}", chars.len() - edge * 2)
+	format!("{start}[…{}ch elided…]{end}", chars.len() - edge * 2)
 }
 
 fn looks_like_table(input: &str) -> bool {
@@ -1073,7 +1071,7 @@ fn compact_delimited_table(input: &str, max_rows: usize) -> String {
 		}
 	}
 	if data_rows > max_rows {
-		out.push(format!("… {} more rows", data_rows - max_rows));
+		out.push(format!("[…{} rows elided…]", data_rows - max_rows));
 	}
 	join_lines(out)
 }
@@ -1114,7 +1112,7 @@ fn compact_psql_table(input: &str) -> String {
 	}
 
 	if data_rows > MAX_PSQL_ROWS {
-		out.push(format!("... +{} more rows", data_rows - MAX_PSQL_ROWS));
+		out.push(format!("[…{} rows elided…]", data_rows - MAX_PSQL_ROWS));
 	}
 	out.extend(row_count_lines);
 	join_lines(out)
@@ -1152,7 +1150,7 @@ fn compact_psql_expanded(input: &str) -> String {
 	}
 	flush_record(&mut out, &mut current, records);
 	if records > MAX_PSQL_ROWS {
-		out.push(format!("... +{} more records", records - MAX_PSQL_ROWS));
+		out.push(format!("[…{} records elided…]", records - MAX_PSQL_ROWS));
 	}
 	out.extend(row_count_lines);
 	join_lines(out)

crates/pi-shell/src/minimizer/filters/docker.rs

@@ -745,9 +745,9 @@ fn compact_table(input: &str, visible_rows: usize) -> String {
 		out.push_str(line.trim_end());
 		out.push('\n');
 	}
-	out.push_str("… ");
+	out.push_str("[…");
 	out.push_str(&(lines.len() - 1 - visible_rows).to_string());
-	out.push_str(" more rows\n");
+	out.push_str(" rows elided…]\n");
 	out
 }
 
@@ -908,7 +908,7 @@ mod tests {
 		let out = filter(&compose_ctx, &input, 0).text;
 		assert!(out.contains("20 rows"));
 		assert!(out.contains("svc-0"));
-		assert!(out.contains("… 8 more rows"));
+		assert!(out.contains("[…8 rows elided…]"));
 	}
 
 	#[test]
@@ -988,7 +988,7 @@ mod tests {
 		assert!(out.contains("api-1  | request 0 complete"));
 		assert!(out.contains("api-1  | WARN cache miss"));
 		assert!(out.contains("worker | failed to process job"));
-		assert!(out.contains("omitted"));
+		assert!(out.contains("elided"));
 	}
 
 	#[test]
@@ -1000,7 +1000,7 @@ mod tests {
 		}
 		let out = compact_table(&input, 10);
 		assert!(out.contains("25 rows"));
-		assert!(out.contains("… 15 more rows"));
+		assert!(out.contains("[…15 rows elided…]"));
 	}
 
 	fn ctx<'a>(
@@ -1071,7 +1071,7 @@ mod tests {
 
 		assert!(out.contains("25 rows"), "docker --format table output should still compact: {out}");
 		assert!(
-			out.contains("… 13 more rows"),
+			out.contains("[…13 rows elided…]"),
 			"docker --format table should keep table omission: {out}"
 		);
 	}
@@ -1390,7 +1390,7 @@ mod tests {
 		}
 		let out = filter(&kubectl_ctx, &input, 0).text;
 		assert!(
-			out.contains("lines omitted") || out.contains("lines truncated"),
+			out.contains("ln elided…]"),
 			"kubectl apply output should be head/tail condensed: {out}"
 		);
 	}

crates/pi-shell/src/minimizer/filters/dotnet.rs

@@ -184,9 +184,9 @@ fn compact_format_json(input: &str) -> Option<String> {
 		out.push('\n');
 	}
 	if rows.len() > 40 {
-		out.push_str("… ");
+		out.push_str("[…");
 		out.push_str(&(rows.len() - 40).to_string());
-		out.push_str(" more diagnostics\n");
+		out.push_str(" diagnostics elided…]\n");
 	}
 	Some(out)
 }

crates/pi-shell/src/minimizer/filters/gh.rs

@@ -319,7 +319,7 @@ mod tests {
 		// First failures stay verbatim with their actionable URLs.
 		assert!(out.text.contains("X\tcheck0\t1s\thttps://ci.test/0"));
 		// The output is capped, not 501 lines of raw rows.
-		assert!(out.text.contains("lines omitted"));
+		assert!(out.text.contains("ln elided…]"));
 		assert!(out.text.lines().count() < 60);
 		assert!(!out.text.contains("https://ci.test/499"));
 	}

crates/pi-shell/src/minimizer/filters/git.rs

@@ -209,7 +209,7 @@ fn compact_diff_listing(input: &str, mode: DiffListingMode) -> String {
 		out.push_str(entry);
 		out.push('\n');
 	}
-	let _ = writeln!(out, "… {} files omitted …", entries.len() - DIFF_LISTING_LIMIT);
+	let _ = writeln!(out, "[…{} files elided…]", entries.len() - DIFF_LISTING_LIMIT);
 	out
 }
 
@@ -512,9 +512,7 @@ fn format_status_summary(summary: &StatusSummary) -> String {
 		out.push('\n');
 	}
 	if summary.paths.len() > 40 {
-		out.push_str("… ");
-		out.push_str(&(summary.paths.len() - 40).to_string());
-		out.push_str(" paths omitted\n");
+		let _ = writeln!(out, "[…{} paths elided…]", summary.paths.len() - 40);
 	}
 	out
 }
@@ -531,9 +529,7 @@ fn condense_log(input: &str, head: usize, tail: usize) -> String {
 			for entry in entries.iter().take(head) {
 				push_log_entry(&mut out, entry);
 			}
-			out.push_str("… ");
-			out.push_str(&(entries.len() - head - tail).to_string());
-			out.push_str(" commits omitted …\n");
+			let _ = writeln!(out, "[…{} commits elided…]", entries.len() - head - tail);
 			for entry in entries.iter().skip(entries.len() - tail) {
 				push_log_entry(&mut out, entry);
 			}
@@ -566,8 +562,8 @@ struct LogEntry {
 	subject: String,
 	body:    Vec<String>,
 	/// Body lines dropped past the per-commit body cap, surfaced as an explicit
-	/// `[+N lines omitted]` marker instead of being silently lost.
-	omitted: usize,
+	/// `[…Nln elided…]` marker instead of being silently lost.
+	elided:  usize,
 }
 
 /// Soft cap on rendered subject/body line width. Long commit subjects and body
@@ -587,10 +583,8 @@ fn push_log_entry(out: &mut String, entry: &LogEntry) {
 		out.push_str(&primitives::truncate_line(line, LOG_LINE_WIDTH));
 		out.push('\n');
 	}
-	if entry.omitted > 0 {
-		out.push_str("  [+");
-		out.push_str(&entry.omitted.to_string());
-		out.push_str(" lines omitted]\n");
+	if entry.elided > 0 {
+		let _ = writeln!(out, "  […{}ln elided…]", entry.elided);
 	}
 }
 
@@ -611,7 +605,7 @@ fn parse_log_entries(input: &str) -> Vec<LogEntry> {
 				hash:    short_hash(hash),
 				subject: subject.to_string(),
 				body:    Vec::new(),
-				omitted: 0,
+				elided:  0,
 			});
 			continue;
 		}
@@ -627,12 +621,12 @@ fn parse_log_entries(input: &str) -> Vec<LogEntry> {
 			entry.subject = trimmed.to_string();
 		} else if !is_git_trailer(trimmed) {
 			// Real (non-trailer) body lines past the 3-line cap are tallied so
-			// `push_log_entry` can emit an explicit `[+N lines omitted]` marker
+			// `push_log_entry` can emit an explicit `[…Nln elided…]` marker
 			// rather than dropping them silently.
 			if entry.body.len() < 3 {
 				entry.body.push(trimmed.to_string());
 			} else {
-				entry.omitted += 1;
+				entry.elided += 1;
 			}
 		}
 	}
@@ -930,9 +924,7 @@ pub(crate) fn compact_diff_output(input: &str) -> String {
 		out.push('\n');
 	}
 	if files.len() > 20 {
-		out.push_str("… ");
-		out.push_str(&(files.len() - 20).to_string());
-		out.push_str(" files omitted from stat\n");
+		let _ = writeln!(out, "[…{} files from stat elided…]", files.len() - 20);
 	}
 	out.push_str(&format_file_count(files.len()));
 	out.push_str(" changed, ");
@@ -956,21 +948,15 @@ pub(crate) fn compact_diff_output(input: &str) -> String {
 				out.push('\n');
 			}
 			if hunk.lines.len() > 6 {
-				out.push_str("  … ");
-				out.push_str(&(hunk.lines.len() - 6).to_string());
-				out.push_str(" changed lines omitted\n");
+				let _ = writeln!(out, "  […{} changed lines elided…]", hunk.lines.len() - 6);
 			}
 		}
 		if file.hunks.len() > 8 {
-			out.push_str("  … ");
-			out.push_str(&(file.hunks.len() - 8).to_string());
-			out.push_str(" hunks omitted\n");
+			let _ = writeln!(out, "  […{} hunks elided…]", file.hunks.len() - 8);
 		}
 	}
 	if files.len() > 12 {
-		out.push_str("\n… ");
-		out.push_str(&(files.len() - 12).to_string());
-		out.push_str(" files omitted from changes\n");
+		let _ = writeln!(out, "\n[…{} files from changes elided…]", files.len() - 12);
 	}
 	out
 }
@@ -1311,7 +1297,7 @@ fn condense_diff_stat(input: &str) -> String {
 		out.push('\n');
 	}
 	if entries.len() > 20 {
-		let _ = writeln!(out, "… {} files omitted …", entries.len() - 20);
+		let _ = writeln!(out, "[…{} files elided…]", entries.len() - 20);
 	}
 	out
 }
@@ -1528,7 +1514,7 @@ fn condense_worktree(input: &str) -> String {
 ///
 /// Listing-shaped lines (`<abs-path> <hash> [<branch>]`, plus the `(bare)` and
 /// `(detached HEAD)` variants) get a leading `$HOME` abbreviated to `~` and are
-/// capped with an omitted-count marker. Any non-listing output (`add`'s
+/// capped with an elided-count marker. Any non-listing output (`add`'s
 /// "Preparing worktree…"/"HEAD is now at …" confirmations, errors) is left to
 /// `condense_noisy_output`/passthrough so its meaning is preserved.
 fn condense_worktree_with_home(input: &str, home: &str) -> String {
@@ -1554,9 +1540,7 @@ fn condense_worktree_with_home(input: &str, home: &str) -> String {
 		out.push('\n');
 	}
 	if entries.len() > WORKTREE_LIMIT {
-		out.push_str("… ");
-		out.push_str(&(entries.len() - WORKTREE_LIMIT).to_string());
-		out.push_str(" worktrees omitted …\n");
+		let _ = writeln!(out, "[…{} worktrees elided…]", entries.len() - WORKTREE_LIMIT);
 	}
 	out
 }
@@ -1920,7 +1904,7 @@ mod tests {
 			input.push('\n');
 		}
 		let out = filter(&ctx, &input, 0);
-		assert!(out.text.contains("… 22 commits omitted …"));
+		assert!(out.text.contains("[…22 commits elided…]"));
 		assert!(out.text.contains("abcdef1 message 0"));
 		assert!(!out.text.contains("message 47"));
 		assert!(out.text.contains("abcdef1 message 69"));
@@ -1983,7 +1967,7 @@ mod tests {
 	}
 
 	#[test]
-	fn log_body_over_cap_shows_omitted_marker() {
+	fn log_body_over_cap_shows_elided_marker() {
 		let cfg = MinimizerConfig { enabled: true, ..Default::default() };
 		let ctx = test_ctx(Some("log"), "git log", &cfg);
 		// Subject + 5 body lines: 3 kept, 2 past the cap surfaced as a marker.
@@ -1995,7 +1979,7 @@ mod tests {
 		assert!(out.text.contains("body line one"));
 		assert!(out.text.contains("body line three"));
 		assert!(!out.text.contains("body line four"), "{:?}", out.text);
-		assert!(out.text.contains("[+2 lines omitted]"), "{:?}", out.text);
+		assert!(out.text.contains("[…2ln elided…]"), "{:?}", out.text);
 	}
 
 	#[test]
@@ -2062,7 +2046,7 @@ mod tests {
 		assert!(out.text.contains("src/file-0.rs\n"));
 		assert!(out.text.contains("src/file-19.rs\n"));
 		assert!(!out.text.contains("src/file-20.rs\n"));
-		assert!(out.text.contains("… 6 files omitted …"));
+		assert!(out.text.contains("[…6 files elided…]"));
 	}
 
 	#[test]
@@ -2091,7 +2075,7 @@ mod tests {
 		assert!(out.text.contains("R100\told-0.rs\tnew-0.rs\n"));
 		assert!(out.text.contains("M\tpath-1.rs\n"));
 		assert!(!out.text.contains("path-20.rs\n"));
-		assert!(out.text.contains("… 4 files omitted …"));
+		assert!(out.text.contains("[…4 files elided…]"));
 	}
 
 	#[test]
@@ -2139,7 +2123,7 @@ mod tests {
 		assert!(out.text.contains("1\t0\tsrc/file-0.rs\n"));
 		assert!(out.text.contains("20\t5\tsrc/file-19.rs\n"));
 		assert!(!out.text.contains("src/file-20.rs\n"));
-		assert!(out.text.contains("… 2 files omitted …"));
+		assert!(out.text.contains("[…2 files elided…]"));
 	}
 
 	#[test]
@@ -2695,13 +2679,13 @@ hint: See the 'Note about fast-forwards' in 'git push --help' for details.
 
 		assert!(out.text.contains("fatal:"), "error header must survive: {:?}", out.text);
 		assert!(out.text.contains("badref"), "offending ref must survive: {:?}", out.text);
-		assert!(!out.text.contains("commits omitted"), "must not fabricate commit listing on error");
+		assert!(!out.text.contains("commits elided"), "must not fabricate commit listing on error");
 	}
 
 	#[test]
 	fn log_oneline_short_run_emits_all_entries() {
 		// A short log that fits within head+tail should emit all entries
-		// without any "omitted" line, and each entry should carry the
+		// without any "elided" line, and each entry should carry the
 		// 7-char short hash followed by the subject.
 		let cfg = MinimizerConfig { enabled: true, ..Default::default() };
 		let ctx = test_ctx(Some("log"), "git log -5", &cfg);
@@ -2712,7 +2696,7 @@ commit abcdef1234567890\nAuthor: A <a@x.com>\nDate: today\n    feat: first\ncomm
 
 		let out = filter(&ctx, input, 0);
 
-		assert!(!out.text.contains("commits omitted"));
+		assert!(!out.text.contains("commits elided"));
 		assert!(out.text.contains("abcdef1 feat: first"), "{:?}", out.text);
 		assert!(out.text.contains("1111111 fix: second"), "{:?}", out.text);
 		assert!(out.text.contains("2222222 chore: third"), "{:?}", out.text);
@@ -2780,8 +2764,8 @@ error: could not apply abc1234... fix: something\nhint: Resolve all conflicts ma
 		assert!(out.contains("(bare)"));
 		assert!(out.contains("(detached HEAD)"));
 		assert!(!out.contains("/home/alice"), "no absolute $HOME paths should survive: {out:?}");
-		// 22 entries → capped at 20 with a 2-omitted marker.
-		assert!(out.contains("… 2 worktrees omitted …"), "{out:?}");
+		// 22 entries → capped at 20 with a 2-elided marker.
+		assert!(out.contains("[…2 worktrees elided…]"), "{out:?}");
 		assert_eq!(out.lines().filter(|l| l.starts_with('~')).count(), 20);
 	}
 
@@ -2799,7 +2783,7 @@ error: could not apply abc1234... fix: something\nhint: Resolve all conflicts ma
 	#[test]
 	fn worktree_porcelain_passthrough() {
 		// `git worktree list --porcelain` emits machine-readable records;
-		// the minimizer must not insert "… N lines omitted …" markers.
+		// the minimizer must not insert "[…Nln elided…]" markers.
 		let cfg = MinimizerConfig { enabled: true, ..Default::default() };
 		let ctx = test_ctx(Some("worktree"), "git worktree list --porcelain", &cfg);
 		let input = "worktree /home/user/project\nHEAD abc1234def\nbranch \
@@ -2825,7 +2809,7 @@ error: could not apply abc1234... fix: something\nhint: Resolve all conflicts ma
 	fn worktree_non_porcelain_still_condensed() {
 		// Normal (non-porcelain) listing should still go through the condenser.
 		// Build 22 listing-shaped entries so the cap (20) is exceeded; the condenser
-		// would insert an "… N worktrees omitted …" marker that a passthrough would
+		// would insert an "[…N worktrees elided…]" marker that a passthrough would
 		// never emit.
 		let cfg = MinimizerConfig { enabled: true, ..Default::default() };
 		let ctx = test_ctx(Some("worktree"), "git worktree list", &cfg);
@@ -2835,8 +2819,8 @@ error: could not apply abc1234... fix: something\nhint: Resolve all conflicts ma
 		}
 		let out = filter(&ctx, &input, 0);
 		assert!(
-			out.text.contains("worktrees omitted"),
-			"non-porcelain output over the cap must show omitted marker; got: {:?}",
+			out.text.contains("worktrees elided"),
+			"non-porcelain output over the cap must show elided marker; got: {:?}",
 			out.text
 		);
 	}

crates/pi-shell/src/minimizer/filters/glab.rs

@@ -211,11 +211,11 @@ fn filter_release_list(input: &str) -> Option<String> {
 	// Append omission marker when there are more releases than shown.
 	let omitted = total.map_or(0, |t| t.saturating_sub(count));
 	if omitted > 0 {
-		let _ = writeln!(filtered, "… {omitted} releases omitted …");
+		let _ = writeln!(filtered, "[…{omitted} releases elided…]");
 	} else if has_more {
 		// Total not parsed from preamble but a 21st row was observed; signal
 		// truncation.
-		filtered.push_str("… releases omitted (showing first 20) …\n");
+		filtered.push_str("[…releases elided…]\n");
 	}
 
 	Some(filtered)
@@ -661,8 +661,8 @@ section_end:1711234600:build_script[0K
 		let out = filter(&ctx, &input, 0);
 
 		assert!(out.changed);
-		// Output must signal that releases were omitted.
-		assert!(out.text.contains("omitted"), "expected omission marker, got: {}", out.text);
+		// Output must signal that releases were elided.
+		assert!(out.text.contains("elided"), "expected omission marker, got: {}", out.text);
 		// First release present, 21st not shown verbatim in the list.
 		assert!(out.text.contains("Release 1"));
 		assert!(!out.text.contains("Release 21 ["));

crates/pi-shell/src/minimizer/filters/go.rs

@@ -341,9 +341,9 @@ fn summarize_golangci_json(line: &str) -> Option<String> {
 		out.push_str(")\n");
 	}
 	if issues.len() > 40 {
-		out.push_str("… ");
+		out.push_str("[…");
 		out.push_str(&(issues.len() - 40).to_string());
-		out.push_str(" more issues\n");
+		out.push_str(" issues elided…]\n");
 	}
 	Some(out)
 }
@@ -506,6 +506,22 @@ mod tests {
 		let out = filter_golangci_lint(input);
 		assert!(out.contains("golangci-lint: 1 issues"));
 		assert!(out.contains("main.go:7:2: unreachable code (govet)"));
+
+		// Match up-to-40 limits, testing elison formatting
+		let mut many_issues = r#"{"Issues":["#.to_string();
+		for i in 0..42 {
+			if i > 0 {
+				many_issues.push(',');
+			}
+			let _ = write!(
+				many_issues,
+				r#"{{"FromLinter":"govet","Text":"err {i}","Pos":{{"Filename":"main.go","Line":{i},"Column":2}}}}"#
+			);
+		}
+		many_issues.push_str("]}");
+		let out_many = filter_golangci_lint(&many_issues);
+		assert!(out_many.contains("golangci-lint: 42 issues"));
+		assert!(out_many.contains("[…2 issues elided…]"));
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/filters/gt.rs

@@ -91,9 +91,7 @@ fn compact_log(input: &str) -> String {
 	}
 
 	if omitted_entries > 0 {
-		out.push_str("… ");
-		out.push_str(&omitted_entries.to_string());
-		out.push_str(" entries omitted …\n");
+		let _ = writeln!(out, "[…{omitted_entries} entries elided…]");
 	}
 
 	primitives::head_tail_lines(&out, 80, 24)
@@ -172,13 +170,13 @@ fn dense_sync_summary(input: &str, exit_code: i32) -> Option<String> {
 		// head_tail_lines bound that compact_noisy_command (the fallback) applies.
 		// Without a cap every name lands on one unbounded line, defeating the
 		// minimizer's bounding guarantee. Show the first DELETED_NAME_CAP names and
-		// summarize the rest as `+N more` (the count above stays exact).
+		// summarize the rest as `[…N names elided…]` (the count above stays exact).
 		const DELETED_NAME_CAP: usize = 20;
 		let shown = deleted_names.len().min(DELETED_NAME_CAP);
 		let names = deleted_names[..shown].join(", ");
 		let _ = write!(summary, " ({names}");
 		if deleted_names.len() > DELETED_NAME_CAP {
-			let _ = write!(summary, ", +{} more", deleted_names.len() - DELETED_NAME_CAP);
+			let _ = write!(summary, " […{} names elided…]", deleted_names.len() - DELETED_NAME_CAP);
 		}
 		summary.push(')');
 	}
@@ -344,7 +342,7 @@ mod tests {
 
 		assert!(out.changed);
 		assert!(out.text.contains("abc1230"));
-		assert!(out.text.contains("entries omitted"));
+		assert!(out.text.contains("entries elided"));
 		assert!(!out.text.contains("user@example.com"));
 	}
 
@@ -439,7 +437,8 @@ mod tests {
 	fn sync_dense_summary_caps_deleted_name_list() {
 		// A long-lived stack cleanup can delete hundreds of merged branches at
 		// once. The dense summary must stay bounded: cap the inline name list and
-		// summarize the remainder as `+N more`, never emit one unbounded line.
+		// summarize the remainder as `[…N names elided…]`, never emit one unbounded
+		// line.
 		let cfg = MinimizerConfig { enabled: true, ..Default::default() };
 		let ctx = test_ctx(Some("sync"), &cfg);
 		let mut input = String::from("Synced with remote\n");
@@ -452,9 +451,9 @@ mod tests {
 		assert!(out.changed);
 		// Exact deleted count is preserved.
 		assert!(out.text.contains("ok sync: 1 synced, 500 deleted"));
-		// First names stay visible; the rest collapse to a `+N more` marker.
+		// First names stay visible; the rest collapse to a `[…N names elided…]` marker.
 		assert!(out.text.contains("feat/merged-0"));
-		assert!(out.text.contains("+480 more"));
+		assert!(out.text.contains("[…480 names elided…]"));
 		// Bounded: a single short line, not 500 names concatenated.
 		assert!(!out.text.contains("feat/merged-499"));
 		let longest = out.text.lines().map(str::len).max().unwrap_or(0);

crates/pi-shell/src/minimizer/filters/js_tools.rs

@@ -219,9 +219,9 @@ fn filter_prettier(input: &str, exit_code: i32) -> String {
 			push_line(&mut out, error);
 		}
 		if errors.len() > 20 {
-			out.push_str("... +");
+			out.push_str("[…");
 			out.push_str(&(errors.len() - 20).to_string());
-			out.push_str(" more errors\n");
+			out.push_str(" errors elided…]\n");
 		}
 		return out;
 	}
@@ -391,9 +391,9 @@ fn push_file_list(out: &mut String, files: &[String], limit: usize) {
 		push_line(out, file);
 	}
 	if files.len() > limit {
-		out.push_str("... +");
+		out.push_str("[…");
 		out.push_str(&(files.len() - limit).to_string());
-		out.push_str(" more files\n");
+		out.push_str(" files elided…]\n");
 	}
 }
 
@@ -406,6 +406,7 @@ fn push_line(out: &mut String, line: &str) {
 mod tests {
 	use super::*;
 	use crate::minimizer::MinimizerConfig;
+	use std::fmt::Write as _;
 
 	fn ctx<'a>(
 		program: &'a str,
@@ -532,4 +533,29 @@ mod tests {
 		assert!(out.contains("P3009"));
 		assert!(!out.contains("Datasource"));
 	}
+
+	#[test]
+	fn prettier_errors_truncation_is_canonical() {
+		let mut input = String::new();
+		for i in 1..=25 {
+			let _ = writeln!(input, "error in file{i}.js: line {i}");
+		}
+		let out = filter_prettier(&input, 1);
+		assert!(out.contains("[…5 errors elided…]\n"));
+		assert!(!out.contains("more errors"));
+		assert!(!out.contains("... +"));
+	}
+
+	#[test]
+	fn file_list_truncation_is_canonical() {
+		let mut files = Vec::new();
+		for i in 1..=35 {
+			files.push(format!("src/file{i}.tsx"));
+		}
+		let mut out = String::new();
+		push_file_list(&mut out, &files, 30);
+		assert!(out.contains("[…5 files elided…]\n"));
+		assert!(!out.contains("more files"));
+		assert!(!out.contains("... +"));
+	}
 }

crates/pi-shell/src/minimizer/filters/jvm.rs

@@ -655,7 +655,7 @@ impl FailuresSummaryCap {
 			return;
 		}
 		if self.dropped > 0 {
-			let _ = write!(out, "\n… +{} more failures\n", self.dropped);
+			let _ = write!(out, "\n[…{} failures elided…]\n", self.dropped);
 		}
 		self.in_summary = false;
 		self.emitted = 0;
@@ -665,7 +665,7 @@ impl FailuresSummaryCap {
 	/// End-of-stream tail emission for cases where the AGG line never arrives.
 	fn finish(&self, out: &mut String) {
 		if self.in_summary && self.dropped > 0 {
-			let _ = write!(out, "\n… +{} more failures\n", self.dropped);
+			let _ = write!(out, "\n[…{} failures elided…]\n", self.dropped);
 		}
 	}
 }
@@ -744,7 +744,7 @@ fn filter_surefire_with_cap(raw: &str, cap: usize) -> String {
 	block.finish(&mut out);
 	summary.finish(&mut out);
 	if dropped_failing > 0 {
-		let _ = write!(out, "\n… +{dropped_failing} more failing test classes\n");
+		let _ = write!(out, "\n[…{dropped_failing} failing test classes elided…]\n");
 	}
 	out
 }
@@ -903,7 +903,7 @@ fn filter_package_with_cap(raw: &str, cap: usize) -> String {
 	block.finish(&mut out);
 	summary.finish(&mut out);
 	if dropped_failing > 0 {
-		let _ = write!(out, "\n… +{dropped_failing} more failing test classes\n");
+		let _ = write!(out, "\n[…{dropped_failing} failing test classes elided…]\n");
 	}
 	out
 }
@@ -1602,7 +1602,7 @@ fn filter_gradle_dependencies(input: &str) -> String {
 			let _ = writeln!(result, "  {dep}");
 		}
 		if deps.len() > cap {
-			let _ = writeln!(result, "  ... +{} more", deps.len() - cap);
+			let _ = writeln!(result, "  […{} dependencies elided…]", deps.len() - cap);
 		}
 	}
 
@@ -1994,7 +1994,7 @@ mod tests {
 			"capped class second per-test block dropped (re-arm inherits drop); got:\n{o}"
 		);
 		assert!(
-			o.contains("… +1 more failing test classes"),
+			o.contains("[…1 failing test classes elided…]"),
 			"tail counts one class, not one per failure; got:\n{o}"
 		);
 	}
@@ -2214,7 +2214,7 @@ mod tests {
 				"Fail{n} exception dropped; got:\n{o}"
 			);
 		}
-		assert!(o.contains("… +2 more failing test classes"), "tail emitted; got:\n{o}");
+		assert!(o.contains("[…2 failing test classes elided…]"), "tail emitted; got:\n{o}");
 	}
 
 	#[test]
@@ -2235,7 +2235,7 @@ mod tests {
 				"Fail{n} dropped under cap=0; got:\n{o}"
 			);
 		}
-		assert!(o.contains("+5 more failing test classes"), "tail counts all 5; got:\n{o}");
+		assert!(o.contains("[…5 failing test classes elided…]"), "tail counts all 5; got:\n{o}");
 	}
 
 	#[test]
@@ -2256,7 +2256,7 @@ mod tests {
 		for n in 4..=5 {
 			assert!(!o.contains(&format!("ClassA.test{n}:25")), "entry {n} dropped; got:\n{o}");
 		}
-		let tail_idx = o.find("… +2 more failures").expect("tail must appear");
+		let tail_idx = o.find("[…2 failures elided…]").expect("tail must appear");
 		let agg_idx = o
 			.find("[ERROR] Tests run: 100")
 			.expect("aggregate must appear");

crates/pi-shell/src/minimizer/filters/lint.rs

@@ -540,9 +540,9 @@ pub fn group_diagnostics(input: &str) -> String {
 			out.push('\n');
 		}
 		if entries.len() > 12 {
-			out.push_str("  … ");
+			out.push_str("  […");
 			out.push_str(&(entries.len() - 12).to_string());
-			out.push_str(" more\n");
+			out.push_str(" diagnostics elided…]\n");
 		}
 	}
 
@@ -551,9 +551,9 @@ pub fn group_diagnostics(input: &str) -> String {
 		out.push('\n');
 	}
 	if ungrouped.len() > 40 {
-		out.push_str("… ");
+		out.push_str("[…");
 		out.push_str(&(ungrouped.len() - 40).to_string());
-		out.push_str(" ungrouped lines omitted\n");
+		out.push_str(" ungrouped lines elided…]\n");
 	}
 	out
 }
@@ -701,7 +701,7 @@ mod tests {
 		}
 		let out = group_diagnostics(&input);
 		assert!(out.contains("src/main.rs (20 diagnostics)"));
-		assert!(out.contains("… 8 more"));
+		assert!(out.contains("  […8 diagnostics elided…]"));
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/filters/listing.rs

@@ -173,16 +173,16 @@ fn compact_grep_grouped(
 			out.push('\n');
 		}
 		if matches.len() > 4 {
-			out.push_str("  … ");
+			out.push_str("  […");
 			out.push_str(&(matches.len() - 4).to_string());
-			out.push_str(" more in file\n");
+			out.push_str(" matches in file elided…]\n");
 		}
 	}
 
 	let omitted_files = grouped.len().saturating_sub(shown_files);
 	let omitted_matches = match_count.saturating_sub(shown_matches);
 	if omitted_files > 0 || omitted_matches > 0 {
-		out.push_str("\n… ");
+		out.push_str("\n[…");
 		out.push_str(&omitted_matches.to_string());
 		out.push_str(" matches");
 		if omitted_files > 0 {
@@ -190,7 +190,7 @@ fn compact_grep_grouped(
 			out.push_str(&omitted_files.to_string());
 			out.push_str(" files");
 		}
-		out.push_str(" omitted\n");
+		out.push_str(" elided…]\n");
 	}
 	for line in ungrouped {
 		out.push_str(line);
@@ -365,14 +365,14 @@ fn compact_find_output_inner(input: &str, paths: &[&str]) -> String {
 		push_wrapped_names(&mut out, names, 4, 24);
 	}
 	if grouped.len() > 16 {
-		out.push_str("\n… ");
+		out.push_str("\n[…");
 		out.push_str(&(grouped.len() - 16).to_string());
-		out.push_str(" dirs omitted\n");
+		out.push_str(" dirs elided…]\n");
 	}
 	if skipped_noise > 0 {
-		out.push_str("… ");
+		out.push_str("[…");
 		out.push_str(&skipped_noise.to_string());
-		out.push_str(" noisy paths omitted\n");
+		out.push_str(" noisy paths elided…]\n");
 	}
 	out
 }
@@ -407,9 +407,9 @@ fn push_wrapped_names(out: &mut String, names: &[String], per_line: usize, max_n
 	}
 	out.push('\n');
 	if names.len() > max_names {
-		out.push_str("  … ");
+		out.push_str("  […");
 		out.push_str(&(names.len() - max_names).to_string());
-		out.push_str(" more\n");
+		out.push_str(" names elided…]\n");
 	}
 }
 
@@ -453,9 +453,9 @@ fn compact_ls_output(input: &str) -> Option<String> {
 	}
 	let shown = dir_count.min(12) + file_count.min(36);
 	if entries.len() > shown {
-		out.push_str("… ");
+		out.push_str("[…");
 		out.push_str(&(entries.len() - shown).to_string());
-		out.push_str(" entries omitted\n");
+		out.push_str(" entries elided…]\n");
 	}
 	out.push('\n');
 	out.push_str(&file_count.to_string());
@@ -622,9 +622,9 @@ fn summarize_cargo_toml(input: &str) -> Option<String> {
 		out.push('\n');
 	}
 	if dependencies.len() > 15 {
-		out.push_str("  … ");
+		out.push_str("  […");
 		out.push_str(&(dependencies.len() - 15).to_string());
-		out.push_str(" more\n");
+		out.push_str(" dependencies elided…]\n");
 	}
 	Some(out)
 }
@@ -705,9 +705,9 @@ fn summarize_package_json(input: &str) -> Option<String> {
 		out.push('\n');
 	}
 	if deps.len() > 15 {
-		out.push_str("  … ");
+		out.push_str("  […");
 		out.push_str(&(deps.len() - 15).to_string());
-		out.push_str(" more\n");
+		out.push_str(" dependencies elided…]\n");
 	}
 	Some(out)
 }
@@ -752,9 +752,9 @@ fn summarize_go_mod(input: &str) -> Option<String> {
 		out.push('\n');
 	}
 	if deps.len() > 15 {
-		out.push_str("  … ");
+		out.push_str("  […");
 		out.push_str(&(deps.len() - 15).to_string());
-		out.push_str(" more\n");
+		out.push_str(" dependencies elided…]\n");
 	}
 	Some(out)
 }
@@ -1265,7 +1265,7 @@ mod tests {
 		let out = filter(&ctx, &input, 0);
 		assert!(out.text.starts_with("grep: 20 matches in 20 files"));
 		assert!(out.text.contains("17: pub fn run(...) -> Result<()> {"));
-		assert!(out.text.contains("matches in 8 files omitted"));
+		assert!(out.text.contains("matches in 8 files elided…]"));
 	}
 
 	#[test]
@@ -1447,7 +1447,7 @@ mod tests {
 		let out = filter(&ctx, &input, 0);
 		assert!(out.text.starts_with("find: 31 paths in 3 dirs"));
 		assert!(out.text.contains("src/module0/ file0.rs file1.rs"));
-		assert!(out.text.contains("1 noisy paths omitted"));
+		assert!(out.text.contains("[…1 noisy paths elided…]"));
 		assert!(!out.text.contains("target/debug"));
 	}
 
@@ -1488,7 +1488,7 @@ mod tests {
 			out.text
 				.contains("Filesystem 1K-blocks Used Available Use% Mounted on")
 		);
-		assert!(out.text.contains("… 13 lines omitted …"));
+		assert!(out.text.contains("[…13ln elided…]"));
 		assert!(out.text.contains("/dev/disk35"));
 	}
 

crates/pi-shell/src/minimizer/filters/pkg.rs

@@ -251,7 +251,7 @@ fn compact_package_tree_output(input: &str) -> String {
 		out.push_str(line);
 		out.push('\n');
 	}
-	let _ = writeln!(out, "… {} package entries omitted …", lines.len() - PACKAGE_TREE_HEAD_LINES);
+	let _ = writeln!(out, "[…{} package entries elided…]", lines.len() - PACKAGE_TREE_HEAD_LINES);
 	out
 }
 
@@ -292,7 +292,7 @@ fn summarize_package_rows(rows: Vec<String>) -> Option<String> {
 		out.push('\n');
 	}
 	if rows.len() > PACKAGE_TREE_HEAD_LINES {
-		let _ = writeln!(out, "… {} package entries omitted …", rows.len() - PACKAGE_TREE_HEAD_LINES);
+		let _ = writeln!(out, "[…{} package entries elided…]", rows.len() - PACKAGE_TREE_HEAD_LINES);
 	}
 	Some(out)
 }
@@ -793,7 +793,7 @@ mod tests {
 		assert!(out.text.contains("├── dep000@1.0.0"));
 		assert!(out.text.contains("├── dep078@1.0.0"));
 		assert!(!out.text.contains("├── dep089@1.0.0"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -808,7 +808,7 @@ mod tests {
 		let out = filter(&context, &input, 0);
 		assert!(out.text.starts_with("package tree/list: 91 entries\n"));
 		assert!(out.text.contains("dep000"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -825,7 +825,7 @@ mod tests {
 		assert!(out.text.starts_with("package tree/list: 92 entries\n"));
 		assert!(out.text.contains("react 19.0.0"));
 		assert!(out.text.contains("└─ dependent000"));
-		assert!(out.text.contains("… 12 package entries omitted …"));
+		assert!(out.text.contains("[…12 package entries elided…]"));
 	}
 
 	#[test]
@@ -889,7 +889,7 @@ mod tests {
 		assert!(out.text.contains("pkg000 1.0.0"));
 		assert!(out.text.contains("pkg078 1.0.78"));
 		assert!(!out.text.contains("pkg089 1.0.89"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -905,7 +905,7 @@ mod tests {
 		assert!(out.text.starts_with("package tree/list: 91 entries\n"));
 		assert!(out.text.contains("project v1.0.0"));
 		assert!(out.text.contains("pkg000"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -920,7 +920,7 @@ mod tests {
 		let out = filter(&context, &input, 0);
 		assert!(out.text.starts_with("package tree/list: 91 entries\n"));
 		assert!(out.text.contains("requests 2.32.0"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -951,7 +951,7 @@ mod tests {
 		let out = filter(&context, &input, 0);
 		assert!(out.text.starts_with("package tree/list: 91 entries\n"));
 		assert!(out.text.contains("pkg000==1.0.0"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -966,7 +966,7 @@ mod tests {
 		let out = filter(&context, &input, 0);
 		assert!(out.text.starts_with("package tree/list: 91 entries\n"));
 		assert!(out.text.contains("dep000==2.0.0"));
-		assert!(out.text.contains("… 11 package entries omitted …"));
+		assert!(out.text.contains("[…11 package entries elided…]"));
 	}
 
 	#[test]
@@ -996,7 +996,7 @@ mod tests {
 		let out = filter(&context, &input, 0);
 		assert!(out.text.starts_with("package tree/list: 92 entries\n"));
 		assert!(out.text.contains("pkg000"));
-		assert!(out.text.contains("… 12 package entries omitted …"));
+		assert!(out.text.contains("[…12 package entries elided…]"));
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/filters/python.rs

@@ -172,15 +172,15 @@ fn filter_pytest(input: &str, exit_code: i32) -> String {
 	// overflow markers so the compact output reads top-to-bottom.
 	let error_overflow = error_blocks.saturating_sub(MAX_PYTEST_FAILURES);
 	if error_overflow > 0 {
-		out.push_str("… +");
+		out.push_str("[…");
 		out.push_str(&error_overflow.to_string());
-		out.push_str(" more errors\n");
+		out.push_str(" errors elided…]\n");
 	}
 	let overflow = failure_blocks.saturating_sub(MAX_PYTEST_FAILURES);
 	if overflow > 0 {
-		out.push_str("… +");
+		out.push_str("[…");
 		out.push_str(&overflow.to_string());
-		out.push_str(" more failures\n");
+		out.push_str(" failures elided…]\n");
 	}
 
 	if has_content(&out) {
@@ -575,7 +575,7 @@ mod tests {
 		assert!(out.contains("test_case_0"), "got: {out}");
 		assert!(out.contains("test_case_9"), "got: {out}");
 		// The overflow marker accounts for the remaining 2 verbose blocks.
-		assert!(out.contains("… +2 more failures"), "got: {out}");
+		assert!(out.contains("[…2 failures elided…]"), "got: {out}");
 		// The traceback bodies for the capped blocks are gone, but the compact
 		// short-summary one-liners for every test survive uncapped.
 		let assert_false_lines = out.matches("E       assert False").count();
@@ -664,10 +664,10 @@ mod tests {
 		let assert_lines = out.matches("E   assert 0 == 100").count();
 		assert_eq!(assert_lines, 5, "all 5 assertion bodies must survive: {out}");
 		// 5 < cap, so there is NO failure overflow stealing real tracebacks.
-		assert!(!out.contains("more failures"), "no failures should be capped: {out}");
+		assert!(!out.contains("failures elided"), "no failures should be capped: {out}");
 		// Collection errors are still surfaced (under their own cap of 10).
 		assert!(out.contains("ERROR collecting tests/test_imp_0.py"), "got: {out}");
-		assert!(!out.contains("more errors"), "9 errors < cap, no error overflow: {out}");
+		assert!(!out.contains("errors elided"), "9 errors < cap, no error overflow: {out}");
 		assert!(out.contains("pytest: 5 failed, 9 errors"), "got: {out}");
 	}
 
@@ -706,9 +706,9 @@ mod tests {
 		let out = filter_pytest(&input, 1);
 
 		// Error overflow marker fires for the 2 capped collection banners…
-		assert!(out.contains("… +2 more errors"), "got: {out}");
+		assert!(out.contains("[…2 errors elided…]"), "got: {out}");
 		// …but NOT a failure overflow — both real tracebacks render in full.
-		assert!(!out.contains("more failures"), "failures must be uncapped here: {out}");
+		assert!(!out.contains("failures elided"), "failures must be uncapped here: {out}");
 		assert!(out.contains("test_real_0"), "got: {out}");
 		assert!(out.contains("test_real_1"), "got: {out}");
 		let assert_false = out.matches("E   assert False").count();

crates/pi-shell/src/minimizer/filters/ruby.rs

@@ -75,7 +75,7 @@ fn filter_rspec(input: &str, exit_code: i32) -> String {
 
 	let mut out = String::new();
 	// Each rendered failure block is buffered separately so the total can be
-	// capped at MAX_RENDERED_FAILURES with a `+N more failures` marker. The
+	// capped at MAX_RENDERED_FAILURES with a `[…N failures elided…]` marker. The
 	// buffered blocks are rendered into `out` the moment the `Failures:` section
 	// ends — at the summary line or the `Failed examples:` boundary — so real
 	// rspec ordering (`Failures:` details, then summary, then `Failed examples:`)
@@ -139,8 +139,8 @@ fn filter_rspec(input: &str, exit_code: i32) -> String {
 }
 
 /// Render the buffered `Failures:` detail blocks into `out`, capped at
-/// `MAX_RENDERED_FAILURES` with a `+N more failures` marker. Runs at most once
-/// — guarded by `rendered` — so the section is emitted exactly where the
+/// `MAX_RENDERED_FAILURES` with a `[…N failures elided…]` marker. Runs at most
+/// once — guarded by `rendered` — so the section is emitted exactly where the
 /// `Failures:` block ends (the summary line or `Failed examples:` boundary),
 /// preserving real rspec section ordering.
 fn render_rspec_blocks(out: &mut String, blocks: &mut Vec<String>, rendered: &mut bool) {
@@ -157,14 +157,14 @@ fn render_rspec_blocks(out: &mut String, blocks: &mut Vec<String>, rendered: &mu
 		out.push_str(block);
 	}
 	if total > MAX_RENDERED_FAILURES {
-		push_line(out, &format!("+{} more failures", total - MAX_RENDERED_FAILURES));
+		push_line(out, &format!("[…{} failures elided…]", total - MAX_RENDERED_FAILURES));
 	}
 	blocks.clear();
 }
 
 /// Cap on rendered failure blocks in non-JSON rspec output. rtk uses the same
 /// limit (`MAX_RSPEC_FAILURES = 5`) — failure blocks carry full context, so a
-/// handful is enough before collapsing to a `+N more failures` marker.
+/// handful is enough before collapsing to a `[…N failures elided…]` marker.
 const MAX_RENDERED_FAILURES: usize = 5;
 
 fn flush_rspec_block(blocks: &mut Vec<String>, current: &mut String) {
@@ -896,7 +896,7 @@ mod tests {
 		assert!(out.contains("Example number 1 fails"));
 		assert!(out.contains("Example number 5 fails"));
 		assert!(!out.contains("Example number 6 fails"));
-		assert!(out.contains("+2 more failures"));
+		assert!(out.contains("[…2 failures elided…]"));
 		assert!(out.contains("7 examples, 7 failures"));
 	}
 

crates/pi-shell/src/minimizer/filters/rust_tools.rs

@@ -79,7 +79,7 @@ fn condense_rustfmt(input: &str, exit_code: i32) -> String {
 		out.push('\n');
 	}
 	if total > 3 {
-		let _ = writeln!(out, "  … {} more", total - 3);
+		let _ = writeln!(out, "  […{} files elided…]", total - 3);
 	}
 	out
 }
@@ -125,7 +125,7 @@ mod tests {
 		assert!(out.changed);
 		assert!(out.text.contains("50 files reformatted"));
 		assert!(out.text.contains("src/file_0.rs"));
-		assert!(out.text.contains("… 47 more"));
+		assert!(out.text.contains("[…47 files elided…]"));
 		// Diff bodies must be elided.
 		assert!(!out.text.contains("old line"));
 		// Savings ratio ≥ 0.7
@@ -154,7 +154,7 @@ mod tests {
 		let out = filter(&context, &input, 1);
 		assert!(out.changed);
 		// head_tail_lines(input, 80, 40) keeps 120 lines + marker.
-		assert!(out.text.contains("lines omitted"));
+		assert!(out.text.contains("ln elided…]"));
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/filters/system.rs

@@ -1,6 +1,7 @@
 //! Conservative text filters for system-style commands.
 
 use std::collections::HashMap;
+use std::fmt::Write as _;
 
 use super::git;
 use crate::minimizer::{MinimizerCtx, MinimizerOutput, primitives};
@@ -358,9 +359,7 @@ fn render_counted_lines(lines: &[LogLine], head: usize, tail: usize) -> String {
 	for line in lines.iter().take(head) {
 		push_counted_line(&mut out, &line.original, line.count);
 	}
-	out.push_str("… ");
-	out.push_str(&(lines.len() - head - tail).to_string());
-	out.push_str(" unique lines omitted …\n");
+	let _ = writeln!(out, "[…{} unique lines elided…]", lines.len() - head - tail);
 	for line in lines.iter().skip(lines.len() - tail) {
 		push_counted_line(&mut out, &line.original, line.count);
 	}
@@ -750,9 +749,7 @@ fn compact_format_output(input: &str) -> String {
 			out.push('\n');
 		}
 		if files.len() > 50 {
-			out.push_str("… ");
-			out.push_str(&(files.len() - 50).to_string());
-			out.push_str(" more files\n");
+			let _ = writeln!(out, "[…{} files elided…]", files.len() - 50);
 		}
 	}
 	if errors.is_empty() && summary.is_empty() && files.is_empty() {
@@ -1073,7 +1070,7 @@ Time        0.42s
 		assert!(out.text.contains("errors:"));
 		assert!(out.text.contains("failed to parse src/bad.py"));
 		assert!(out.text.contains("files:"));
-		assert!(out.text.contains("more files"));
+		assert!(out.text.contains("[…50 files elided…]"));
 	}
 
 	#[test]

crates/pi-shell/src/minimizer/primitives.rs

@@ -93,9 +93,9 @@ pub fn head_tail_lines(input: &str, head: usize, tail: usize) -> String {
 		out.push_str(line);
 		out.push('\n');
 	}
-	out.push_str("… ");
+	out.push_str("[…");
 	out.push_str(&omitted.to_string());
-	out.push_str(" lines omitted …\n");
+	out.push_str("ln elided…]\n");
 	for line in lines.iter().skip(lines.len() - tail) {
 		out.push_str(line);
 		out.push('\n');
@@ -292,9 +292,9 @@ pub fn head_lines_only(input: &str, head: usize) -> String {
 		out.push_str(line);
 		out.push('\n');
 	}
-	out.push_str("… ");
+	out.push_str("[…");
 	out.push_str(&omitted.to_string());
-	out.push_str(" lines omitted …\n");
+	out.push_str("ln elided…]\n");
 	out
 }
 
@@ -307,9 +307,9 @@ pub fn tail_lines_only(input: &str, tail: usize) -> String {
 	}
 	let omitted = lines.len() - tail;
 	let mut out = String::new();
-	out.push_str("… ");
+	out.push_str("[…");
 	out.push_str(&omitted.to_string());
-	out.push_str(" lines omitted …\n");
+	out.push_str("ln elided…]\n");
 	for line in lines.iter().skip(omitted) {
 		out.push_str(line);
 		out.push('\n');
@@ -330,9 +330,9 @@ pub fn max_lines(input: &str, max: usize) -> String {
 		out.push_str(line);
 		out.push('\n');
 	}
-	out.push_str("… ");
+	out.push_str("[…");
 	out.push_str(&dropped.to_string());
-	out.push_str(" lines truncated …\n");
+	out.push_str("ln elided…]\n");
 	out
 }
 
@@ -380,7 +380,7 @@ mod tests {
 	#[test]
 	fn head_tail_marks_omitted_lines() {
 		let out = head_tail_lines("1\n2\n3\n4\n5\n", 2, 1);
-		assert_eq!(out, "1\n2\n… 2 lines omitted …\n5\n");
+		assert_eq!(out, "1\n2\n[…2ln elided…]\n5\n");
 	}
 
 	#[test]
@@ -397,7 +397,7 @@ mod tests {
 			.collect::<Vec<_>>()
 			.join("\n");
 		let out = head_tail_cap(&input, CapClass::List);
-		assert!(out.contains("lines omitted"));
+		assert!(out.contains("ln elided…]"));
 		assert!(out.lines().count() <= CapClass::List.lines() + 1);
 	}
 

docs/compaction.md

@@ -263,7 +263,7 @@ Cumulative behavior:
 - In split turns, includes turn-prefix file ops too.
 - `details.readFiles` excludes files also modified; `details.modifiedFiles` carries the rest (persisted shape is unchanged).
 
-Summary text gets one `<files>` tag appended via prompt template: a grouped, prefix-folded directory tree (find-tool shape) with a per-file access marker — `(Read)` for read-only files, `(Write)` for modified files never read, `(RW)` for modified files also present in the cumulative read set. Capped at 20 files with an `… (N more files omitted)` line.
+Summary text gets one `<files>` tag appended via prompt template: a grouped, prefix-folded directory tree (find-tool shape) with a per-file access marker — `(Read)` for read-only files, `(Write)` for modified files never read, `(RW)` for modified files also present in the cumulative read set. Capped at 20 files with an `[…N files elided…]` line.
 
 ```xml
 <files>

docs/tools/read.md

@@ -97,7 +97,7 @@ URL selectors are parsed separately in `packages/coding-agent/src/tools/fetch.ts
 ### Local text files
 - No selector: if summarization is enabled and the file is small enough, `#trySummarize()` calls `summarizeCode()`.
   - Guards: file size `<= 2 MiB` (`MAX_SUMMARY_BYTES`), line count `<= 20_000` (`MAX_SUMMARY_LINES`).
-  - Summary output keeps selected declarations and replaces elided spans with `...` or merged brace-pair lines containing `..`. When at least one span is elided, the text content ends with a footer like `[NN lines elided; re-read needed ranges, e.g. <path>:5-16,40-80]` using concrete ranges from the actual elisions.
+  - Summary output keeps selected declarations and replaces elided spans with `…` or merged brace-pair lines containing `{ … }`. When at least one span is elided, the text content ends with a footer like `[…NNln elided; re-read needed ranges, e.g. <path>:5-16,40-80]` using concrete ranges from the actual elisions.
   - When an elided block sits between matching brace lines, `#renderSummary()` may merge them into one anchored line rather than emitting separate opener/closer lines.
 - Explicit selector or summarization miss: streamed text read.
   - Default open-ended limit is `min(session setting read.defaultLimit, DEFAULT_MAX_LINES)`.

packages/agent/CHANGELOG.md

@@ -1,6 +1,9 @@
 # Changelog
 
 ## [Unreleased]
+### Changed
+
+- Updated the display format for truncated file operation summaries
 
 ## [16.0.8] - 2026-06-18
 
@@ -869,4 +872,4 @@ Initial release under @oh-my-pi scope. See previous releases at [badlogic/pi-mon
 ### Changed
 
 - `Agent` constructor now has all options optional (empty options use defaults).
-- `queueMessage()` is now synchronous (no longer returns a Promise).
+- `queueMessage()` is now synchronous (no longer returns a Promise).

packages/agent/src/compaction/utils.ts

@@ -173,7 +173,7 @@ export function formatFileOperations(
 	const all = [...mode.keys()].sort();
 	let files = formatGroupedPaths(all.slice(0, FILE_OPERATION_SUMMARY_LIMIT), path => ` (${mode.get(path)})`);
 	if (all.length > FILE_OPERATION_SUMMARY_LIMIT) {
-		files += `\n… (${all.length - FILE_OPERATION_SUMMARY_LIMIT} more files omitted)`;
+		files += `\n[…${all.length - FILE_OPERATION_SUMMARY_LIMIT} files elided…]`;
 	}
 	return prompt.render(fileOperationsTemplate, { files });
 }

packages/coding-agent/CHANGELOG.md

@@ -1,6 +1,7 @@
 # Changelog
 
 ## [Unreleased]
+
 ### Added
 
 - Added `__advisor.jsonl` transcript persistence for advisor model usage attribution and visibility in the Agent Hub
@@ -12,6 +13,8 @@
 
 ### Changed
 
+- Standardized file elision markers to a compact `[…Nln elided…]` format
+- Updated transcript viewer to clear cached content if the session file is deleted while opened
 - Refactored agent hub to use a fullscreen overlay for agent transcripts instead of inline chat
 - Improved transcript rendering stability by using a non-incremental builder for viewer replays
 - Advisor transcripts are now excluded from agent-facing surfaces like `irc`, `history://`, and peer rosters
@@ -21,6 +24,7 @@
 
 ### Fixed
 
+- Fixed memory-leaking stale transcripts in the agent viewer when underlying files are deleted
 - Fixed the Agent Hub transcript viewer rendering the transcript body one column right of the "Agent Hub" title (and the title appearing to shift when scrolled to the top): the fullscreen viewer added its own outer gutter on top of the transcript rows, which already carry a 1-column left pad, so the header and body no longer shared a gutter. The viewer now renders the scroll body at full width without the extra gutter, and the file-mention row carries the same 1-column pad as every other row.
 - Fixed the bash tool failing with `pi-natives:command: syntax error at end of input` on a valid `&&`/`;` chain whose later pipeline stage is a compound command, e.g. `echo x && git log | while read h; do …; done | head`. The minimizer's segmented-chain runner rebuilds each chain segment from the brush AST via `pipeline.to_string()` and re-executes that string, but `simple_segment` only validated the *first* pipeline stage — so a compound later stage (`while`/`for`/`if`/subshell) was re-serialized without its terminator and re-run as broken shell. Every stage is now required to be a Display-safe simple command, and — as a general guard against the recurring class of brush `Display` round-trip divergences (previously: quoted here-doc close tags, multi-byte char/byte offsets) — each reconstructed segment is now re-parsed and must match the original pipeline shape before the chain runner executes it; any divergence runs the command whole, unsegmented, instead of corrupting it.
 - Fixed `Ctrl+T` (toggle thinking blocks) and the `/settings` "Hide Thinking Blocks" toggle only collapsing/expanding thinking in the live region: blocks that had scrolled into committed native scrollback on ED3-risk terminals kept their pre-toggle snapshot, so scrolling up showed the old thinking state. Both paths now `resetDisplay()` after flipping each block's flag, forcing a full clear + replay of the whole transcript (matching the tool-output expansion toggle) so every block above the fold re-renders at its new height.

packages/coding-agent/src/hindsight/mental-models.ts

@@ -387,7 +387,7 @@ export function diffMentalModelContent(previous: string | null, current: string,
 
 	if (out.length > maxLines) {
 		const dropped = out.length - maxLines;
-		return `${out.slice(0, maxLines).join("\n")}\n… ${dropped} more line${dropped === 1 ? "" : "s"} elided`;
+		return `${out.slice(0, maxLines).join("\n")}\n[…${dropped}ln elided…]`;
 	}
 	return out.join("\n");
 }

packages/coding-agent/src/lsp/index.ts

@@ -2035,7 +2035,7 @@ export class LspTool implements AgentTool<typeof lspSchema, LspToolDetails, Them
 			const lines = limitedSymbols.map(s => formatSymbolInformation(s, this.session.cwd));
 			const truncationLine =
 				dedupedSymbols.length > WORKSPACE_SYMBOL_LIMIT
-					? `\n... ${dedupedSymbols.length - WORKSPACE_SYMBOL_LIMIT} additional symbol(s) omitted`
+					? `\n[…${dedupedSymbols.length - WORKSPACE_SYMBOL_LIMIT} symbols elided…]`
 					: "";
 			return {
 				content: [

packages/coding-agent/src/modes/components/agent-transcript-viewer.ts

@@ -161,19 +161,28 @@ export class AgentTranscriptViewer implements Component {
 		let signature: string;
 		try {
 			const stat = fs.statSync(sessionFile);
-			signature = `${stat.size}:${stat.mtimeMs}`;
+			// Include the path: a different file with the same size/mtime must not alias.
+			signature = `${sessionFile}:${stat.size}:${stat.mtimeMs}`;
 		} catch {
+			// File deleted/rotated while open (e.g. the owning session was dropped):
+			// clear stale content once instead of freezing on it forever.
+			if (this.#lastSignature !== "missing") {
+				this.#lastSignature = "missing";
+				this.#model = undefined;
+				this.#rebuild([]);
+			}
 			return;
 		}
 		if (signature === this.#lastSignature) return;
-		this.#lastSignature = signature;
 		let text: string;
 		try {
 			text = fs.readFileSync(sessionFile, "utf-8");
 		} catch (err) {
+			// Leave #lastSignature unchanged so a transient read error retries next poll.
 			logger.debug("transcript viewer: read failed", { err: String(err) });
 			return;
 		}
+		this.#lastSignature = signature;
 		this.#model = undefined;
 		this.#rebuild(this.#extractMessages(parseSessionEntries(text)));
 	}

packages/coding-agent/src/prompts/tools/read.md

@@ -33,7 +33,7 @@ Append `:<sel>` to `path`; bare path = default mode.
 - File with explicit selector → lines prefixed with numbers: `41|def alpha():`.
 {{/if}}
 {{/if}}
-- Parseable code without selector → **structural summary**: declarations kept, bodies collapsed to `..` (merged brace pair) or `…` (standalone). The footer shows the recovery selector: `[NN lines elided; re-read needed ranges, e.g. <path>:5-16,40-80]`. Re-issue ONLY the ranges you need via the multi-range selector. `..`/`…` carry no content — NEVER guess what's inside; NEVER re-read the whole file or `:raw` when ranges suffice.
+- Parseable code without selector → **structural summary**: declarations kept, body elided with `…`. The footer shows the recovery selector. Re-issue ONLY the ranges you need via the multi-range selector.
 
 # Documents & Notebooks
 

packages/coding-agent/src/session/streaming-output.ts

@@ -495,8 +495,10 @@ export function truncateTail(content: string, options: TruncationOptions = {}):
  * Returned without surrounding newlines so callers can position it freely.
  */
 export function formatMiddleElisionMarker(elidedLines: number, elidedBytes: number): string {
-	const linesPart = `${elidedLines.toLocaleString()} line${elidedLines === 1 ? "" : "s"}`;
-	return `[… ${linesPart} elided (${formatBytes(elidedBytes)}) …]`;
+	// A 0/1-line elision (e.g. one giant single line) would read as
+	// "[…0ln elided…]"; fall back to a byte count there.
+	if (elidedLines <= 1) return `[…${elidedBytes}B elided…]`;
+	return `[…${elidedLines}ln elided…]`;
 }
 
 /**
@@ -578,8 +580,6 @@ export function truncateMiddle(content: string, options: TruncationOptions = {})
 export interface InlineByteCapOptions {
 	/** Inline byte budget. Defaults to {@link DEFAULT_MAX_BYTES}. */
 	maxBytes?: number;
-	/** What the text is, for the elision marker (e.g. "bash output"). */
-	label: string;
 	/**
 	 * Persist the full text as a session artifact. When an artifact id is
 	 * returned, a `[raw output: artifact://<id>]` footer is appended so the
@@ -619,7 +619,7 @@ export async function enforceInlineByteCap(text: string, options: InlineByteCapO
 	const head = trimHeadToLineBoundary(truncateHeadBytes(text, Math.floor(maxBytes * 0.6)).text);
 	const tail = trimTailToLineBoundary(truncateTailBytes(text, Math.floor(maxBytes * 0.25)).text);
 	const elidedBytes = Math.max(0, totalBytes - Buffer.byteLength(head, "utf-8") - Buffer.byteLength(tail, "utf-8"));
-	const marker = `[… elided ${elidedBytes} bytes of ${options.label} …]`;
+	const marker = `[…${elidedBytes}B elided…]`;
 	let composed = `${head}\n${marker}\n${tail}`;
 
 	const artifactId = await options.saveArtifact?.(text);

packages/coding-agent/src/tools/bash.ts

@@ -485,7 +485,6 @@ export class BashTool implements AgentTool<typeof bashSchemaBase | typeof bashSc
 		// head-retention spill, minimizer miss) may emit more than
 		// ~DEFAULT_MAX_BYTES inline. No-op for already-bounded output.
 		const cappedOutputText = await enforceInlineByteCap(outputText, {
-			label: "bash output",
 			saveArtifact: full => saveBashOriginalArtifact(this.session, full),
 		});
 

packages/coding-agent/src/tools/browser.ts

@@ -332,7 +332,6 @@ export class BrowserTool implements AgentTool<typeof browserSchema, BrowserToolD
 		// text inline; the full text stays recoverable via the artifact footer
 		// when allocation succeeds.
 		const cappedText = await enforceInlineByteCap(textOnly, {
-			label: "browser output",
 			saveArtifact: full => saveBrowserOutputArtifact(this.session, full),
 		});
 		details.result = cappedText;

packages/coding-agent/src/tools/read.ts

@@ -275,7 +275,7 @@ function formatMergedBraceLine(
 	shouldAddHashLines: boolean,
 	shouldAddLineNumbers: boolean,
 ): { model: string; display: string } {
-	const merged = `${headText.trimEnd()} .. ${tailText.trim()}`;
+	const merged = `${headText.trimEnd()} … ${tailText.trim()}`;
 	if (shouldAddHashLines) {
 		return { model: `${startLine}-${endLine}:${merged}`, display: merged };
 	}
@@ -315,7 +315,7 @@ const FOOTER_RANGE_SAMPLES = 2;
 
 /**
  * Footer appended to summarized reads telling the model how to recover the
- * elided body. Without this hint, agents either ignore the `...`/`{ .. }`
+ * elided body. Without this hint, agents either ignore the `…`/`{ … }`
  * markers or burn a turn guessing the right selector (see issue #1046). The
  * footer demonstrates the multi-range selector syntax with concrete sample
  * ranges drawn from the actual elision so the model re-reads only what it
@@ -327,7 +327,6 @@ function formatSummaryElisionFooter(
 	elidedLines: number,
 ): string {
 	if (elidedRanges.length === 0) return "";
-	const lineWord = elidedLines === 1 ? "line" : "lines";
 	const sampleCount = Math.min(elidedRanges.length, FOOTER_RANGE_SAMPLES);
 	const selector = elidedRanges
 		.slice(0, sampleCount)
@@ -335,7 +334,7 @@ function formatSummaryElisionFooter(
 		.join(",");
 	const example = `${readPath}:${selector}`;
 	const tail = elidedRanges.length > sampleCount ? `, e.g. ${example}` : ` with ${example}`;
-	return `[${elidedLines} ${lineWord} elided; re-read needed ranges${tail}]`;
+	return `[…${elidedLines}ln elided; re-read needed ranges${tail}]`;
 }
 const READ_CHUNK_SIZE = 8 * 1024;
 
@@ -1904,8 +1903,8 @@ export class ReadTool implements AgentTool<typeof readSchema, ReadToolDetails> {
 		let elidedLines = 0;
 		for (const unit of units) {
 			if (unit.kind === "elided") {
-				modelParts.push("...");
-				displayParts.push("...");
+				modelParts.push("…");
+				displayParts.push("…");
 				elidedRanges.push({ start: unit.startLine, end: unit.endLine });
 				elidedLines += unit.endLine - unit.startLine + 1;
 				continue;

packages/coding-agent/src/workspace-tree.ts

@@ -298,7 +298,7 @@ function applyLineCap(
 	const removed = new Set(removable.map(item => item.index));
 	const kept = lines.filter((_, index) => !removed.has(index));
 	kept.push({
-		label: `… (${removable.length} lines elided beyond depth/cap)`,
+		label: `[…${removable.length}ln elided…]`,
 		depth: 0,
 		isRoot: false,
 	});

packages/coding-agent/test/agent-hub-advisor-scroll.test.ts

@@ -140,4 +140,32 @@ describe("AgentTranscriptViewer", () => {
 			expect(atBottom).not.toContain("PROMPTMARKER");
 		});
 	});
+
+	it("clears stale content when the transcript file is deleted while open", async () => {
+		const dir = fs.mkdtempSync(path.join(os.tmpdir(), "adv-view-"));
+		const file = path.join(dir, "__advisor.jsonl");
+		fs.writeFileSync(file, buildJsonl());
+		const viewer = makeViewer(file);
+		const body = () =>
+			viewer
+				.render(80)
+				.map(l => Bun.stripANSI(l))
+				.join("\n");
+		try {
+			viewer.render(80);
+			viewer.handleInput("g");
+			expect(body()).toContain("PROMPTMARKER");
+
+			fs.rmSync(file);
+			// Poll until the viewer's own poll timer re-stats and clears (deadline-bounded).
+			const deadline = Date.now() + 5000;
+			while (body().includes("PROMPTMARKER") && Date.now() < deadline) {
+				await Bun.sleep(50);
+			}
+			expect(body()).not.toContain("PROMPTMARKER");
+		} finally {
+			viewer.dispose();
+			fs.rmSync(dir, { recursive: true, force: true });
+		}
+	});
 });

packages/coding-agent/test/edit/file-snapshot-store.test.ts

@@ -72,17 +72,15 @@ describe("snapshot store fusion via canonical keys", () => {
 
 describe("parseSeenLinesFromHashlineBody", () => {
 	it("collects single NN: line numbers and skips the header and footer rows", () => {
-		const body = ["[src/x.ts#1A2B]", "300:function f() {", "301:\treturn 1;", "302:}", "[2 lines elided; …]"].join(
-			"\n",
-		);
+		const body = ["[src/x.ts#1A2B]", "300:function f() {", "301:\treturn 1;", "302:}", "[…2ln elided; …]"].join("\n");
 		expect(parseSeenLinesFromHashlineBody(body)).toEqual([300, 301, 302]);
 	});
 
 	it("adds only the boundary lines of a collapsed NN-MM: summary row, never the interior", () => {
 		const body = [
-			"30-39:export interface Snapshot { .. }",
+			"30-39:export interface Snapshot { … }",
 			"40:",
-			"46-61:export abstract class SnapshotStore { .. }",
+			"46-61:export abstract class SnapshotStore { … }",
 		].join("\n");
 		expect(parseSeenLinesFromHashlineBody(body)).toEqual([30, 39, 40, 46, 61]);
 	});
@@ -92,7 +90,7 @@ describe("parseSeenLinesFromHashlineBody", () => {
 	});
 
 	it("tolerates grep `*`/space match markers before the line number (search/ast-grep output)", () => {
-		const body = ["*73:matched line", " 74:context line", "...", " 75:more context"].join("\n");
+		const body = ["*73:matched line", " 74:context line", "…", " 75:more context"].join("\n");
 		expect(parseSeenLinesFromHashlineBody(body)).toEqual([73, 74, 75]);
 	});
 });

packages/coding-agent/test/hindsight-mental-models.test.ts

@@ -319,7 +319,7 @@ describe("diffMentalModelContent", () => {
 		const out = diffMentalModelContent(null, big, 50);
 		const lines = out.split("\n");
 		expect(lines.length).toBe(51); // 50 diff lines + 1 elision marker
-		expect(lines[lines.length - 1]).toMatch(/more lines? elided$/);
+		expect(lines[lines.length - 1]).toMatch(/\[…\d+ln elided…\]$/);
 	});
 
 	it("caps LCS input lines so a huge curated model cannot hang the diff", () => {

packages/coding-agent/test/read-summary.test.ts

@@ -72,8 +72,8 @@ describe("read summary", () => {
 		const result = await tool.execute("read-summary-ts", { path: fixture });
 		const text = textOutput(result);
 
-		expect(text).toContain("export function alpha(value: string): string { .. }");
-		expect(text).toContain("export function beta(): number { .. }");
+		expect(text).toContain("export function alpha(value: string): string { … }");
+		expect(text).toContain("export function beta(): number { … }");
 		expect(text).not.toContain("const clean = value.trim()");
 		expect(result.details?.summary?.elidedSpans).toBe(2);
 	});
@@ -124,7 +124,7 @@ describe("read summary", () => {
 		const text = textOutput(result);
 
 		expect(text).toContain("const clean = 'alpha';");
-		expect(text).not.toContain("...");
+		expect(text).not.toContain("…");
 		expect(result.details?.summary).toBeUndefined();
 	});
 
@@ -197,10 +197,10 @@ describe("read summary", () => {
 		expect(text).toContain("name: Ada");
 	});
 
-	it("renders brace-pair elisions as a single numbered line with `..`", async () => {
+	it("renders brace-pair elisions as a single numbered line with `…`", async () => {
 		// Regression for the read-tool format request: collapse the head /
 		// elided / closing-brace sandwich into one numbered line of the form
-		// `START-END:head { .. }` instead of three separate lines.
+		// `START-END:head { … }` instead of three separate lines.
 		const fixture = path.join(tmpDir, "merge.ts");
 		await fs.writeFile(
 			fixture,
@@ -211,9 +211,9 @@ describe("read summary", () => {
 		const result = await tool.execute("read-summary-merge", { path: fixture });
 		const text = textOutput(result);
 
-		expect(text).toContain("export function stripNewLinePrefixes(lines: string[]): string[] { .. }");
-		// The plain `...` ellipsis line must NOT appear once the merge fires.
-		expect(text).not.toContain("\n...\n");
+		expect(text).toContain("export function stripNewLinePrefixes(lines: string[]): string[] { … }");
+		// The plain `…` ellipsis line must NOT appear once the merge fires.
+		expect(text).not.toContain("\n…\n");
 		// The merged line must use the numbered range shape.
 		expect(text).toMatch(/\b1-7:export function stripNewLinePrefixes/);
 		expect(result.details?.summary?.elidedSpans).toBe(1);
@@ -229,8 +229,8 @@ describe("read summary", () => {
 		const result = await tool.execute("read-summary-merge-trailing", { path: fixture });
 		const text = textOutput(result);
 
-		expect(text).toContain("export const config = { .. };");
-		expect(text).not.toContain("\n...\n");
+		expect(text).toContain("export const config = { … };");
+		expect(text).not.toContain("\n…\n");
 	});
 
 	it("does not merge when the closing line is not a bare brace", async () => {
@@ -248,15 +248,15 @@ describe("read summary", () => {
 		const text = textOutput(result);
 
 		expect(text).toContain("def greet(name: str) -> str:");
-		// Python's body elision keeps first/last body lines, so plain `...`
+		// Python's body elision keeps first/last body lines, so plain `…`
 		// must remain as the elided segment.
-		expect(text).toContain("\n...\n");
-		expect(text).not.toContain(" .. ");
+		expect(text).toContain("\n…\n");
+		expect(text).not.toContain(" … ");
 	});
 
 	it("appends an elision footer that names targeted recovery ranges", async () => {
 		// Regression for issue #1046: summarized reads must tell the model how
-		// to recover the elided body so it does not stall on `...` / `{ .. }`
+		// to recover the elided body so it does not stall on `…` / `{ … }`
 		// markers and burn a turn guessing the selector.
 		const fixture = path.join(tmpDir, "footer.ts");
 		await fs.writeFile(
@@ -270,7 +270,7 @@ describe("read summary", () => {
 
 		expect(result.details?.summary?.elidedSpans).toBe(2);
 		expect(result.details?.summary?.elidedLines).toBeGreaterThan(0);
-		expect(text).toContain("lines elided");
+		expect(text).toContain("ln elided");
 		expect(text).toContain(`${fixture}:1-5,7-11`);
 		expect(text).not.toContain(`${fixture}:raw`);
 		expect(text).not.toContain(`${fixture}:1-9999`);

packages/coding-agent/test/silent-abort-overlay-render.test.ts

@@ -54,7 +54,7 @@ function makeViewer(sessionFile: string, observed: ObservableSession[]): AgentTr
 		registry: agents,
 		observers: makeSubagentRegistry(observed),
 		ui,
-		cwd: tmpDir,
+		cwd: path.dirname(sessionFile),
 		expandKeys: ["ctrl+o"],
 		hubKeys: ["ctrl+s"],
 		requestRender: () => {},
@@ -117,7 +117,7 @@ describe("Agent hub silent-abort regression", () => {
 			},
 		]);
 
-		const hub = makeHub(sessionFile, [
+		const viewer = makeViewer(sessionFile, [
 			{
 				id: SESSION_ID,
 				kind: "subagent",
@@ -128,10 +128,8 @@ describe("Agent hub silent-abort regression", () => {
 			},
 		]);
 
-		// Render with a reasonable width — the hub chat view reads the session
-		// file and calls #buildTranscriptLines internally.
-		const rendered = hub.render(120);
-		hub.dispose();
+		const rendered = viewer.render(120);
+		viewer.dispose();
 		const renderedText = rendered.join("\n");
 
 		// The sentinel MUST NOT appear verbatim in any rendered line
@@ -176,7 +174,7 @@ describe("Agent hub silent-abort regression", () => {
 			},
 		]);
 
-		const hub = makeHub(sessionFile, [
+		const viewer = makeViewer(sessionFile, [
 			{
 				id: SESSION_ID,
 				kind: "subagent",
@@ -187,8 +185,8 @@ describe("Agent hub silent-abort regression", () => {
 			},
 		]);
 
-		const rendered = hub.render(120);
-		hub.dispose();
+		const rendered = viewer.render(120);
+		viewer.dispose();
 		const renderedText = rendered.join("\n");
 
 		// AssistantMessageComponent renders the error as "Error: <message>"

packages/coding-agent/test/streaming-output.test.ts

@@ -560,9 +560,10 @@ describe("truncateMiddle", () => {
 		expect(result.content).not.toContain("elided");
 	});
 
-	test("formatMiddleElisionMarker pluralises and formats bytes", () => {
-		expect(formatMiddleElisionMarker(1, 100)).toBe("[… 1 line elided (100B) …]");
-		expect(formatMiddleElisionMarker(123, 4096)).toBe("[… 123 lines elided (4.0KB) …]");
+	test("formatMiddleElisionMarker uses lines, falling back to bytes for <=1 line", () => {
+		expect(formatMiddleElisionMarker(0, 512)).toBe("[…512B elided…]");
+		expect(formatMiddleElisionMarker(1, 100)).toBe("[…100B elided…]");
+		expect(formatMiddleElisionMarker(123, 4096)).toBe("[…123ln elided…]");
 	});
 });
 

packages/coding-agent/test/tools/output-caps.test.ts

@@ -8,7 +8,7 @@
 import { describe, expect, it } from "bun:test";
 import { DEFAULT_MAX_BYTES, enforceInlineByteCap } from "@oh-my-pi/pi-coding-agent/session/streaming-output";
 
-const MARKER_PATTERN = /\[… elided \d+ bytes of test output …\]/;
+const MARKER_PATTERN = /\[…\d+B elided…\]/;
 
 /** Build `count` complete lines of the form `line-00001 <pad>`. */
 function makeLines(count: number, pad = "x".repeat(40)): string {
@@ -22,24 +22,24 @@ function makeLines(count: number, pad = "x".repeat(40)): string {
 describe("enforceInlineByteCap", () => {
 	it("returns sub-cap text unchanged (identity, not just equality)", async () => {
 		const text = makeLines(10);
-		const result = await enforceInlineByteCap(text, { maxBytes: 4096, label: "test output" });
+		const result = await enforceInlineByteCap(text, { maxBytes: 4096 });
 		expect(result).toBe(text);
 	});
 
 	it("returns text exactly at the cap unchanged", async () => {
 		const text = "a".repeat(1000);
 		expect(Buffer.byteLength(text, "utf-8")).toBe(1000);
-		const result = await enforceInlineByteCap(text, { maxBytes: 1000, label: "test output" });
+		const result = await enforceInlineByteCap(text, { maxBytes: 1000 });
 		expect(result).toBe(text);
 	});
 
 	it("uses DEFAULT_MAX_BYTES when maxBytes is omitted", async () => {
 		const under = "a".repeat(DEFAULT_MAX_BYTES - 1);
-		expect(await enforceInlineByteCap(under, { label: "test output" })).toBe(under);
+		expect(await enforceInlineByteCap(under, {})).toBe(under);
 
 		const over = makeLines(2000); // ~94KB, well over the 50KB default
 		expect(Buffer.byteLength(over, "utf-8")).toBeGreaterThan(DEFAULT_MAX_BYTES);
-		const result = await enforceInlineByteCap(over, { label: "test output" });
+		const result = await enforceInlineByteCap(over, {});
 		expect(result).not.toBe(over);
 		expect(result).toMatch(MARKER_PATTERN);
 		expect(Buffer.byteLength(result, "utf-8")).toBeLessThanOrEqual(DEFAULT_MAX_BYTES);
@@ -48,7 +48,7 @@ describe("enforceInlineByteCap", () => {
 	it("elides over-cap text to head + tail on line boundaries within the budget", async () => {
 		const maxBytes = 4096;
 		const text = makeLines(500); // ~23KB
-		const result = await enforceInlineByteCap(text, { maxBytes, label: "test output" });
+		const result = await enforceInlineByteCap(text, { maxBytes });
 
 		expect(result).toMatch(MARKER_PATTERN);
 		// Stays within the cap plus a sliver of slack for the marker line.
@@ -80,7 +80,7 @@ describe("enforceInlineByteCap", () => {
 		// Each line is multi-byte heavy: é (2B), € (3B), 😀 (4B).
 		const text = makeLines(800, "é€😀".repeat(12));
 		const maxBytes = 4096;
-		const result = await enforceInlineByteCap(text, { maxBytes, label: "test output" });
+		const result = await enforceInlineByteCap(text, { maxBytes });
 
 		expect(result).toMatch(MARKER_PATTERN);
 		// Valid UTF-8 round-trip: encode/decode is lossless and introduces no
@@ -102,7 +102,6 @@ describe("enforceInlineByteCap", () => {
 		let saved: string | undefined;
 		const result = await enforceInlineByteCap(text, {
 			maxBytes: 4096,
-			label: "test output",
 			saveArtifact: full => {
 				saved = full;
 				return Promise.resolve("17");
@@ -118,7 +117,6 @@ describe("enforceInlineByteCap", () => {
 		const text = makeLines(500);
 		const result = await enforceInlineByteCap(text, {
 			maxBytes: 4096,
-			label: "test output",
 			saveArtifact: () => undefined,
 		});
 		expect(result).not.toContain("[raw output: artifact://");
@@ -130,7 +128,6 @@ describe("enforceInlineByteCap", () => {
 		const text = "short output";
 		const result = await enforceInlineByteCap(text, {
 			maxBytes: 4096,
-			label: "test output",
 			saveArtifact: () => {
 				called = true;
 				return "99";

packages/coding-agent/test/tools/search-path-lists.test.ts

@@ -7,7 +7,7 @@ import { validateToolArguments } from "@oh-my-pi/pi-ai/utils/validation";
 import { resetSettingsForTest, Settings } from "@oh-my-pi/pi-coding-agent/config/settings";
 import { canonicalSnapshotKey } from "@oh-my-pi/pi-coding-agent/edit/file-snapshot-store";
 import type { RenderResultOptions } from "@oh-my-pi/pi-coding-agent/extensibility/custom-tools/types";
-import { AgentHubOverlayComponent } from "@oh-my-pi/pi-coding-agent/modes/components/agent-hub";
+import { AgentTranscriptViewer } from "@oh-my-pi/pi-coding-agent/modes/components/agent-transcript-viewer";
 import { TreeSelectorComponent } from "@oh-my-pi/pi-coding-agent/modes/components/tree-selector";
 import type {
 	ObservableSession,
@@ -354,16 +354,20 @@ describe("tool path arrays", () => {
 			status: "parked",
 		});
 
-		const hub = new AgentHubOverlayComponent({
-			observers,
-			hubKeys: ["ctrl+s"],
-			onDone: () => {},
-			requestRender: () => {},
+		const viewer = new AgentTranscriptViewer({
+			agentId: "search-overlay-session",
 			registry: agents,
+			observers,
+			ui: { requestRender: () => {}, requestComponentRender: () => {} } as never,
+			cwd: tmp,
+			expandKeys: ["ctrl+o"],
+			hubKeys: ["ctrl+s"],
+			requestRender: () => {},
+			onClose: () => {},
+			onHubClose: () => {},
 		});
-		hub.openChat("search-overlay-session");
-		const rendered = Bun.stripANSI(hub.render(120).join("\n"));
-		hub.dispose();
+		const rendered = Bun.stripANSI(viewer.render(120).join("\n"));
+		viewer.dispose();
 
 		// The hub chat now renders through searchToolRenderer.renderCall; the
 		// single-string `paths` arg shows up as the "in <paths>" scope meta on the

packages/coding-agent/test/tools/strip-output-notice.test.ts

@@ -45,7 +45,7 @@ const limitsOnly: OutputMeta = {
 
 describe("stripOutputNotice", () => {
 	it("removes the exact notice appended by the wrapper for middle elision", () => {
-		const body = "line1\nline2\n[… 4 lines elided (215B) …]\nline7\nline8";
+		const body = "line1\nline2\n[…4ln elided…]\nline7\nline8";
 		const notice = formatOutputNotice(truncation);
 		const combined = body + notice;
 

packages/coding-agent/test/workspace-tree.test.ts

@@ -120,7 +120,7 @@ describe("buildWorkspaceTree", () => {
 		expect(tree.truncated).toBe(true);
 		expect(tree.totalLines).toBeLessThanOrEqual(120);
 		expect(renderedLines.length).toBeLessThanOrEqual(120);
-		expect(tree.rendered).toContain("lines elided beyond depth/cap");
+		expect(tree.rendered).toContain("ln elided…]");
 	});
 
 	it("can keep root entries uncapped while truncating child directories", async () => {

packages/snapcompact/CHANGELOG.md

@@ -1,8 +1,10 @@
 # Changelog
 
 ## [Unreleased]
+
 ### Changed
 
+- Refined elision markers for file operations and truncated text for better display consistency
 - Updated summary text for consistent descriptions of archived tool output
 - Folded a much wider range of Unicode to ASCII in `normalize()` before native rendering: added a per-character Unicode NFKD decomposition fallback (fullwidth forms, super/subscripts, ligatures, circled and math-styled alphanumerics, Roman numerals, vulgar fractions) and expanded the `CHAR_FOLD` punctuation table (more quotes/primes, hyphens, the fraction slash, dot leaders, bullets, and arrows) so undrawable glyphs land on close ASCII equivalents instead of `?`
 

packages/snapcompact/src/snapcompact.ts

@@ -592,7 +592,7 @@ function formatFileOperations(readFiles: string[], modifiedFiles: string[], read
 	const all = [...mode.keys()].sort();
 	let files = formatGroupedPaths(all.slice(0, FILE_OPERATION_SUMMARY_LIMIT), path => ` (${mode.get(path)})`);
 	if (all.length > FILE_OPERATION_SUMMARY_LIMIT) {
-		files += `\n… (${all.length - FILE_OPERATION_SUMMARY_LIMIT} more files omitted)`;
+		files += `\n[…${all.length - FILE_OPERATION_SUMMARY_LIMIT} files elided…]`;
 	}
 	return prompt.render(fileOperationsTemplate, { files });
 }
@@ -657,7 +657,7 @@ function truncateForSummary(text: string, maxChars: number, headRatio: number):
 	const tailChars = maxChars - headChars;
 	const elided = text.length - maxChars;
 	const tail = tailChars > 0 ? text.slice(-tailChars) : "";
-	return `${text.slice(0, headChars)} [… ${elided}ch elided …] ${tail}`;
+	return `${text.slice(0, headChars)} […${elided}ch elided…] ${tail}`;
 }
 
 const DIM_MARKERS = /[\u000e\u000f]/g;

packages/snapcompact/test/snapcompact.test.ts

@@ -489,7 +489,7 @@ describe("serializeConversation", () => {
 		// Default cap 2000 at 0.6 head ratio: 1200 head + 800 tail survive.
 		expect(out).toContain("<out>");
 		expect(out).toContain("HEAD-");
-		expect(out).toContain("[… 3010ch elided …]");
+		expect(out).toContain("[…3010ch elided…]");
 		expect(out.endsWith(`-TAIL${snapcompact.DIM_OFF}\n</out>`)).toBe(true);
 	});
 
@@ -499,7 +499,7 @@ describe("serializeConversation", () => {
 			toolResultMaxChars: 10,
 			truncateHeadRatio: 0.5,
 		});
-		expect(tight).toContain("[… 90ch elided …]");
+		expect(tight).toContain("[…90ch elided…]");
 		const off = snapcompact.serializeConversation([createToolResultMessage(text)], {
 			toolResultMaxChars: Number.POSITIVE_INFINITY,
 		});
@@ -514,7 +514,7 @@ describe("serializeConversation", () => {
 		]);
 		// JSON-encoded content is 3002 chars; per-value cap 500 elides 2502.
 		expect(out).toContain('write(path="a.ts", content=');
-		expect(out).toContain("[… 2502ch elided …]");
+		expect(out).toContain("[…2502ch elided…]");
 	});
 
 	it("caps the whole serialized argument list per call", () => {

packages/tui/CHANGELOG.md

@@ -1,9 +1,9 @@
 # Changelog
 
 ## [Unreleased]
-
 ### Fixed
 
+- Corrected logic in string truncation to prevent improper truncation of short strings
 - Fixed a one-frame transcript flash during a non-multiplexer resize drag: while the drag borrowed the alternate screen and painted only the viewport, any ordinary (non-forced) render from a still-animating block — a tool spinner tick, a streamed token, a cursor blink — fell through to the deferred geometry-rebuild full paint, which left the alternate screen to repaint the whole transcript on the normal screen for a single frame before the next SIGWINCH re-entered the viewport fast path, so a live tool block flashed in and vanished. Ordinary renders mid-drag now stay on the viewport fast path; only forced renders (tool finalization, reset, image reconciliation) still preempt it.
 
 ## [16.0.10] - 2026-06-18
@@ -1730,4 +1730,4 @@ Initial release under @oh-my-pi scope. See previous releases at [badlogic/pi-mon
 
 ### Fixed
 
-- **Readline-style Ctrl+W**: Now skips trailing whitespace before deleting the preceding word, matching standard readline behavior. ([#306](https://github.com/badlogic/pi-mono/pull/306) by [@kim0](https://github.com/kim0))
+- **Readline-style Ctrl+W**: Now skips trailing whitespace before deleting the preceding word, matching standard readline behavior. ([#306](https://github.com/badlogic/pi-mono/pull/306) by [@kim0](https://github.com/kim0))

packages/tui/src/utils.ts

@@ -90,7 +90,7 @@ export function truncateToWidth(
 	const safeWidth = Number.isFinite(maxWidth) ? Math.max(0, Math.trunc(maxWidth)) : 0;
 	// Fast path: every UTF-16 unit is at most 3 cells wide, so a string whose
 	// `length * 3` already fits within `safeWidth` cannot need truncation.
-	if (!pad && text.length * 3<= safeWidth) {
+	if (!pad && text.length * 3 <= safeWidth) {
 		return text;
 	}
 	let resolvedEllipsis: Ellipsis | null | undefined | string = ellipsisKind;