- Removed the `F` sed flag and `literal` field, keeping only `pat`, `rep`, and `g`, and dropped the obsolete `F` test.
- Updated sed parsing, serialization, and error text to remove `splice_block` exclusivity and explain regex escaping.
- Reworked atom prompt format to `<ops>/<splice>/<sed>`, with shared `{path, edits}` calls and strict one-loc/one-verb rules.
- Updated docs (`read.md` and `CHANGELOG.md`) to remove `F:true` anti-patterns and add current `line+hash` and sed behavior guidance.
- Replaced `AgentTool` `nointent` and `deriveIntent` with a unified `intent` option that supports `omit`, `optional`, `require`, or a callback function.
- Updated agent tool schema injection and execution paths to inject `_i` as required/optional/omitted and to derive intent through the new `intent` callback.
- Aligned atom-edit tests with the updated default sed behavior (`g` now off by default and explicit `g: true` for global replacement).
- Replaced `atom.sed` string syntax with `{pat,rep,g?,F?,i?}` objects in schema, prompts, and tests.
- Changed sed validation defaults so `pat` is regex, `g` defaults to global replacement, and `F:true` forces literal matching.
- Fixed `sed` edge cases by allowing sequential same-anchor edits, handling zero-length matches, and preserving chaining.
- Updated native build flow to local `--profile` defaults, refactored enum generation, and switched totals to `total_ms`.
Drop three pieces of well-intentioned-but-fragile string mangling that
were quietly mutating model edit content:
- `maybeAutocorrectEscapedTabIndentation` in both hashline and atom edit
modes (plus the `PI_HASHLINE_AUTOCORRECT_ESCAPED_TABS` env toggle and
`isEscapedTabAutocorrectEnabled` helper). Literal `\\t` in edit content
is now preserved verbatim.
- `maybeWarnSuspiciousUnicodeEscapePlaceholder` and the
`runHashlinePreflightSanitizers` wrapper. The warning was misleading:
it flagged `\\uDDDD` as suspicious without distinguishing intentional
literal text from a likely-malformed escape, and it was the only
remaining caller.
- The hand-rolled JSON-unescape fallback in
`decodePartialJsonStringFragment` (renderer.ts). The fallback used
`String.fromCharCode` (mishandling lone surrogates) and only covered a
subset of JSON escapes, so the same fragment decoded one way mid-stream
and a different way after `JSON.parse` succeeded. Now we trim a
trailing partial `\\u`/odd-backslash escape and surface the raw
fragment if `JSON.parse` still fails.
- Added a guard in applySedToLine that probes matches and rejects sed patterns that match empty strings.
- Updated atom edit error handling to label invalid sed expressions as rejected rather than failed to compile.
- Adjusted sed tests to expect the new rejection wording and added coverage for zero-length pattern `s/()/X/`.
- Updated `read` selector parsing for file and URL reads to accept optional leading `L` and `+` count-style ranges.
- Adjusted truncation notices and schema/help text to emit and suggest `sel` offsets without the `L` prefix, including continuation and suggestion messages.
- Updated hashline/output parsing and related tests to recognize the new `sel` formatting in truncation notices.
- Renamed atom tool schema fields and internal `AtomEdit` operation from `set` to `splice` in edit parsing and application.
- Adjusted resolver and conflict checks to use `splice` semantics, including file-loc rejection, blank-line deletion handling, and sed precedence.
- Updated atom edit documentation and tests to describe and verify the new `splice`-based API.
- Updated `formatMatchLine` to emit `*` for matched lines, a leading space for context, and a `|` anchor/content separator.
- Revised grep/hashline mismatch messages and prompts to describe the new marker and separator format.
- Aligned affected atom and hashline tests with the updated match-line prefixes and separators.
- Updated atom edit location parsing to treat "$" as a file-wide target for prepend, append, and sed, and rejected "^" as a supported locator.
- Added a new sed_file operation that runs line-wise substitutions across the full file, preserving ordering and trailing-newline semantics while failing when no lines match.
- Removed per-entry path overrides from atom resolution and made read selector parsing return "none" for unrecognized selectors to defer handling to specialized readers.
- Added `sed` atom editing with `g`, `i`, and `F` flags, path/anchor parsing, and conflict handling updates.
- Changed hashline and grep/read output to `LINE+ID|content` with `>` match prefixes and `:` context prefixes.
- Fixed atom anchor parsing for path-qualified locs, hyphenated single anchors, and content hints after `|` or `:`.
- Updated prompts, changelog, and tests to document and validate the new hashline and `sed` formats.
- Removed the atom tool's `sub` verb from its schema, parser, conflict checks, and edit application.
- Updated atom documentation, examples, and tests to reject `sub`-based replacements in favor of `set` replacements.
- Adjusted Anthropic strict-tool handling by dropping `write` from the allowlist and updating alignment expectations.
- Added auto-rebasing for stale atom and hashline anchors within ±2 lines, with warning diagnostics.
- Removed atom range-locator support and dropped `between` ops, updating docs/tests for `sub` over `set` nudges.
- Changed no-op handling to track unchanged hashline edits and emit contextual hints for unchanged ranges.
- Expanded Anthropic strict error handling to retry on schema-too-complex and compiled-grammar-too-large errors.
- Added anchor-retargeting warning checks and updated edit tests to expect locator and range-locator rejections.
- Updated python tool-call fixtures by adding `title` to executed-cell payloads and adjusting related test expectations.
- Adjusted hashline tests to expect colon-prefixed anchors in format, diff preview, and parse assertions.
- Updated ast-edit test regexes and split logic to parse colon-delimited hashline prefixes.
- Disabled strict schema enforcement on AskTool to relax request validation.
- Removed the optional `cwd` field from Python tool parameters and used `session.cwd` for warmup and execution context.
- Updated atom edit test fixtures to pass `set` as arrays instead of scalar strings.
- Added locator-based atom edits with required `loc`, including inline `file:line`, range, and `^`/`$` selectors.
- Added path resolution via top-level path, entry path, or `path:loc` selectors and flattened multi-target atom execution.
- Removed `del`; required `set/pre/post` to use line arrays and mapped `set: []` to deletion semantics.
- Changed `sub` payloads to required `[find, replace]` tuples and tightened parsing to accept null optional verbs and array-or-null lines.
- Added `CodeFrameMarker` and `formatCodeFrameLine()` to centralize code-frame gutter formatting.
- Extended diff rendering to preserve `|` and `│` separators, aligning gutter markers and line numbers.
- Reworked AST, grep, hashline, Vim, and diff renderers to use shared line formatting with computed `lineNumberWidth`.
- Updated atom editing flow and tests, including `resolveAtomEntryPaths` migration and new loc-based/edge-case coverage.
- Adjusted benchmark runner early-stop configuration by passing `buildEarlyStop` through prompt collection.
- Changed hashline/atom parsing to require full `line+suffix` anchors and emit full-anchor guidance on failures.
- Updated atom/hashline prompt docs, `atom.test.ts`, and changelog entries to require exact full anchors like `160sr`.
- Added preformatted `displayContent` for read/grep/ast tools and switched renderers to prefer it in TUI output.
- Updated mismatch and grep/ast renderers to show context with `*` markers and gutter lines, removing legacy helpers.
- Expanded hashline and chunk bigram tables to 647 entries and moved chunk checksums to a 40-item namespace.
- Changed hashline anchors from `LINE#ID`/`:` to concatenated `LINEID`\t forms across parsing and tool outputs.
- Removed line-number padding and routed diff/read/grep/renderer output through raw numbers, tabs, and `toDisplayLine` formatting.
- Renamed atom ops to `pre`/`post`, removed `ins`, and updated schemas, prompts, and tests for new insertion behavior.
- Added `examples` support to `StringEnum` schemas and propagated it to tool metadata.
- Added concise descriptions and example values across coding-agent tool schemas for clearer guidance.
- Documented the new StringEnum examples capability in `packages/ai/CHANGELOG.md`.
- Cast `real` to `HASHLINE_BIGRAMS` elements in `staleBigramFor` test setup.
- Updated the atom edit schema and tool prompt docs to replace the standalone `between` form with a 2-tuple `set` form for exclusive block-body replacement.
- Adjusted `resolveAtomToolEdit` so `set` accepts either a single anchor or `[open, close]`, translates tuple input to an internal between-style op, and rejects tuple values that are not anchor strings.
- Expanded tests to verify tuple-based `set` resolution and to enforce validation errors for invalid `set` array inputs.
- Added `between` atom edits with `after`/`before` anchors and logged them in changelog docs.
- Added `between` docs to prompt tooling, including two-anchor exception and anchor-order rules.
- Added `between` validation to reject stale anchors, invalid ranges, overlap, and nested interior edits.
- Reworked `applyAtomEdits` to apply `between` operations as bottom-up interior splices and adjacent-line inserts.
- Updated `file_path` schema description in review tool to "Path to the file".
- Added `AtomEdit` tests for `between` parsing and edge-case success/failure coverage.
- Added raw read output propagation so read/archive commands bypass anchors, line numbers, and chunk formatting.
- Fixed atom/hashline editing by tightening hashline prefixes and applying grouped anchor edits in stable order.
- Hardened chunk parsing and path handling by using `bigram_end` checks and resolving edit paths via shared `args.path` fallback.
- Updated path-related behavior for chunk, replace, patch, and hashline previews to honor optional edit paths.
- Removed mode-level param validators and replaced them with runtime handling, then removed obsolete validation tests.
- Added `atom` edit mode to `EditTool` with single-point hashline-style dispatch and `executeAtomSingle`.
- Added `AtomToolEdit`/`AtomParams` schema and `atom` prompt docs to require one op per edit entry.
- Changed `EditTool` entry processing to inherit per-entry `path` from a top-level request `path` fallback.
- Added top-level `path` defaults to patch and replace schemas and now reject missing entry paths.
- Added an `atom` streaming strategy and registered it in `EDIT_MODE_STRATEGIES` for live mode support.
- Enabled hashline read anchors for `atom` mode via `file-display-mode.ts` path-display logic.
- Added `atom` unit tests covering schema checks, anchor conflicts, and hash mismatch diagnostics.
- Aligned hashline and chunk token handling to `HASHLINE_BIGRAMS` in `computeLineHash` and parsing regexes.
- Removed hashline file-level `move`/`delete` options from schemas and execution so edits are in-place updates only.
- Hardened hashline prefix parsing to reject non-bigram IDs, preventing false stripping of `#` comment lines.
- Updated `read()`/prompt wording so raw reads skip prefixes and examples now show anchors like `123#th`.
- Updated hashline chunk tests for new `HASHLINE_BIGRAMS` IDs and replaced stale hash constants.
The previous mismatch text ('1 line has changed since last read. Use
the updated LINE#ID references shown below.') reads like a successful
edit followed by an informational note. Providers whose tool-result
plumbing does not surface `isError: true` to the model (e.g. Qwen via
the OpenAI-completions shim, where mitmproxy traces show the model only
receives the content text) treat the response as a success and proceed
with stale state.
Reword the message to start with 'Edit rejected:' and explicitly state
'The edit was NOT applied' so the failure is unambiguous in plaintext.
The structural payload (>>> markers, updated LINE#IDs) is unchanged.
Fixes#742
Slots a new "apply_patch" variant alongside the existing edit modes
(replace, patch, hashline, chunk, vim). The mode accepts a single input
string containing a Codex *** Begin Patch / *** End Patch envelope,
parses it with a new lenient parser (heredoc-tolerant), and fans each
file-op out to the existing executePatchSingle so LSP writethrough,
plan-mode guards, fs-cache invalidation and diagnostics are shared
with the patch mode.
Exposes both tool shapes from the spec: the JSON function-tool variant
(§1.2, {input: string}) and the OpenAI custom-tool / Lark-grammar
"freeform" variant (§1.1, raw patch string). The edit tool advertises
a Lark grammar via customFormat and a wire name via customWireName;
openai-responses emits it as a grammar-constrained custom tool when a
model opts in with applyPatchToolType: "freeform" in models.json.
custom_tool_call / custom_tool_call_output are plumbed end-to-end
through the shared responses code (emission, streaming, history
replay), and the agent-loop dispatcher matches tool calls by either
name or customWireName so returned calls route correctly.
Also threads preview/diff rendering for apply_patch through the TUI
(tool-execution + edit renderer) so streaming patches show per-file
diffs like the other edit modes.
Default edit mode is unchanged (hashline); opt in via edit.mode or
PI_EDIT_VARIANT=apply_patch.
- Standardized missing-file read errors and now return `File not found: <path>` for absent edit targets.
- Centralized AI provider, usage, and OAuth helpers into shared modules to remove duplicated logic.
- Migrated OAuth/API-key login flows to shared factory helpers and removed inline prompt/token-exchange code.
- Reused shared tools and formatter utilities for discovery, stream tails, LSP batching, and source formatting.
- Consolidated repeated test helpers and fixtures into shared modules, replacing inline helper duplicates.
The hashline prefix stripping logic uses an "all-or-nothing" heuristic:
it only strips N#XX: anchors when every non-empty line carries one.
When the Read tool truncation notice (e.g. [Showing lines 1-300 of
332. Use sel=L301 to continue]) is present in content passed to the
Edit or Write tool, that marker line breaks the heuristic, causing ALL
anchors to be written to disk verbatim. On subsequent reads, those
persisted anchors get double-prefixed (1#NX:1#BQ:---), compounding
the corruption.
This can happen when the AI model pastes Read tool output (including
the truncation marker) back into an Edit/Write call. While this is
arguably a model-level issue, the toolchain should be resilient against
it rather than silently corrupting files.
Changes:
- Add READ_TRUNCATION_NOTICE_RE to detect Read tool truncation markers
- Update collectLinePrefixStats to exclude truncation marker lines from
the non-empty line count so they no longer break the stripping heuristic
- Add stripLeadingHashlinePrefixes helper for recursive stripping of
nested/double-prefixed anchors (handles already-corrupted content)
- Add filterTruncationNotices helper to remove truncation marker lines
- Update both stripNewLinePrefixes and stripHashlinePrefixes to filter
truncation markers and use recursive anchor stripping
- Add 5 new test cases covering truncation markers and nested anchors
- Added `vim` as a supported edit mode and resolved mode-specific active tool names.
- Migrated chunk-edit requests from legacy `op`/`content` to `path` plus `write`, `replace`, and `insert`.
- Replaced chunk-edit internals with `put`/`replace`/`delete` operations and ID-based selection.
- Updated chunk prompts/output docs to document ID-based selectors and `write: null` deletions.
- Switched chunk and grep render output to `@`/dash-style anchors and updated file-count behavior for multi-file edits.
- Updated multi-file result rendering with distinct file counting and pending-file progress indicator.
- Refreshed chunk-mode and chunk-tree tests for the new contract and anchor style.
- Added hierarchical chunk grep output grouped by directory, file, and chunk with aligned chunk anchors.
- Updated chunk rendering to show clipped head/tail context with explicit truncation and expand markers.
- Changed chunk edit and streaming validation to use chunkToolEditSchema-based checks for safe edit detection.
- Enhanced grep rendering to use structured ChunkedGrepMatch data with displayPath and chunk checksum metadata.
- Updated chunk-mode and core chunk tests to use abbreviated chunk selectors introduced by path truncation.
- Adjusted expected chunk-path and error-path strings for markdown, Rust, and TLA+ fixtures to match the new naming scheme.
- Updated SIXEL sanitization assertions to check for the new "visible columns omitted" truncation message.
fixed retained-kernel restart and owner cleanup edge cases during recovery and disposal
tracked async user_python hooks during disposal-sensitive execution paths and hardened startup warmup tracking
strengthened cleanup and kernel lifecycle regressions to remove deadlocks, false positives, and timing flakes
scoped retained-kernel ownership to agent sessions and cleaned it up on session disposal
cleaned up warmed python owners on session startup failure and rejected new direct and tool-based python starts during disposal, including async hook, preflight, and warmup races
- Added language-aware chunking for Svelte, Vue, Astro, and HTML script/style tags using `lang` aliases.
- Added markdown/html markup handling for fenced code and HTML blocks with detected injected languages.
- Added injected-subtree chunk infrastructure and build-time merge logic in `pi-natives/chunk`.
- Added tests for markdown and script-chunking embedded-language behavior in chunk-tree and chunk-mode suites.
- Refactored chunk rendering to compute prologue and body line counts separately via new `chunk_head_body_lines()` function.
- Updated ellipsis marker format from line count to line range selection syntax (e.g., 'to expand above/below').
- Modified `render()` method signature to accept `head_lines` and `body_lines` parameters for precise line count display.
- Simplified no-changes response in chunk edit mode to omit redundant text when content matches.
- Expanded chunk-edit documentation with compound statement examples and region suffix behavior for non-code formats.
- Simplified chunk region notation from @head/@body/@tail/@decl to ^/~ suffix format across all modules.
- Removed ChunkRegion::Tail and ChunkRegion::Decl enum variants and updated region matching logic accordingly.
- Added call-with-callback pattern detection via try_promote_call_with_callback() and trailing_callback_body() functions.
- Updated documentation and test fixtures to reflect new region selector syntax throughout codebase.
- Extracted `raceWithAbort()` utility to pi-utils package for reuse across projects.
- Replaced local `raceAbort()` implementation with imported `raceWithAbort()` from pi-utils.
- Refactored cleanup timer to use async iterator pattern with `timers.setInterval()` instead of callback-based `setInterval()`.
- Converted `#cleanupIdleSessions()` and `#disposeSession()` from async to synchronous methods with async cleanup loop delegated to `#runCleanupLoop()`.
- Migrated language classifiers from imperative methods to declarative semantic rule tables with ClassifierTables and StructuralOverrides.
- Extracted node shape analysis into dedicated shape module with field priority constants and helper functions for AST traversal.
- Added schema module with language-aware node metadata and thread-local language context management.
- Centralized environment variable parsing across codebase using $flag(), $envpos(), and isBunTestRuntime() utilities.
- Added PI_CHUNK_AUTOINDENT configuration to control indentation normalization in chunk read/edit operations.
- Enhanced system prompt with instruction priority, output contract, tool persistence, and completeness guidelines.
- Corrected markdown section spacing logic to preserve blank lines between inserted sections and following headings.
- Added markdown-aware spacing detection to handle FirstChild, LastChild, Before, and After insert positions correctly.
- Added two test cases validating blank line preservation in markdown After and Append operations.
- Reformatted doc comments in RenderParams and ReadRenderParams for line length compliance.
- Extracted chunk tree formatting logic into dedicated `format_selector_tree()` function for reuse across error handling.
- Refactored error messages to display chunk children as hierarchical tree structure with line numbers and signatures instead of comma-separated lists.
- Removed `format_chunk_ref()` helper and consolidated formatting logic into tree-based error hint generation.
- Updated test assertions to validate tree-formatted output with regex patterns matching branch symbols and structure.
- Fixed file not found error handling in auto-generated marker detection to gracefully return undefined instead of propagating errors.
- Made assertEditableFileContent function synchronous to simplify API and improve error handling consistency.
- Updated test assertions to match synchronous function behavior using toThrow instead of rejects patterns.
- Removed obsolete test case for doc comment preservation during replace operations.
can1357