resolveCommandConfig used execSync with a 10s timeout and only cached non-empty successful results, so empty/failing/timeout !command values in models.yml re-ran on every API key lookup and blocked the event loop repeatedly. It now probes the cache with has(), stores empty stdout and caught failures as undefined, and returns the cached undefined on later lookups. Extension-registered fetchDynamicModels is now wrapped with a 15s Promise.race timeout so a stuck extension cannot hang model refresh indefinitely. Verified by Bun transpiler syntax check of packages/coding-agent/src/config/model-registry.ts and the model-registry test suites.
Closes#4237
getOrCreateSnapshot in shell-snapshot.ts pre-creates snapshotPath as an empty temp file and returns null on spawn failure, timeout, or nonzero exit without deleting it, leaving stale files in os.tmpdir()/omp-shell-snapshots/. Track whether snapshot creation succeeded and remove the pre-created file in a finally block on every failure path using fs.rmSync(snapshotPath, { force: true }), with best-effort error suppression so cleanup failures never propagate. Verified with `bunx tsc --noEmit -p packages/coding-agent/tsconfig.json` and `bun test packages/coding-agent/test/shell-snapshot.test.ts` (22 pass).
Closes#4236
Replaces the bespoke batch-scoped process.exit interceptor with the
withExitGuard convention main established for extension/hook/plugin
loaders (500c39aa2): guard the module import and factory invocation so
a synchronous process.exit()/process.reallyExit() from a custom tool
becomes an ExtensionExitError handled as a recoverable load error,
while host exit paths stay untouched outside the guarded windows.
Tests cover the import-time exit (issue #1704 repro) and factory-time
exit; both would kill the test process without the guard.
main independently absorbed batch-1 (incremental grapheme slice 718c7cea2, markdown
stream-prefix cache 705426548 + 3822a83b4) and the pathTo/patch.ts items; restore
main's refined versions wholesale. Port the model-resolver optimization onto main's
resolver shape: hoist per-candidate case folds in matchModel and build the
preference context once per role resolution (matchPatternWithContext) instead of
per fallback pattern. Rewrite changelog entries to the surviving items only.
Byte-identical copy of the withExitGuard/ExtensionExitError block from
main (500c39aa2) so the custom-tool loader can reuse the established
guard convention; merges as an identical change against main.
Narrow acronym restoration so plain all-caps English words such as FIX
and WORK do not get restored when the model naturally capitalizes the
first title word. Restorable all-caps source tokens now need a stronger
acronym signal: a common technical acronym allowlist, digits, or a
consonant-only shape.
This keeps CNPG, ETL, JWT, SQL, and API restoration while preserving the
anti-shout behavior for single emphatic words.
Fixes#4220
- Added an enhanced speech pipeline that utilizes small models to rewrite text for natural language synthesis.
- Implemented `SpeechEnhancer` and `BlockAccumulator` to manage fence-aware text streaming and paragraph splitting.
- Configured a new `speech.enhanced` setting to toggle between mechanical and enhanced vocalization modes.
- Resolved `EPIPE` rejections during speech playback by ensuring stream flushes and suppressing stop-related errors.
reconcileTitleCasing now maps ALL-CAPS source tokens (CNPG, API, ETL,
JWT) into an acronyms table and restores them when the model produces a
plain title-cased artifact (Cnpg). Restoration is disabled when the
source is shouty (>=2 consecutive multi-letter ALL-CAPS tokens like
"FIX the BUG NOW" or "ALL ERROR HANDLING"), and lowercase model output
is left alone so isolated single-word emphasis (WORK -> work) is never
re-shouted.
The three title system prompts (title-system.md, title-system-marker.md,
tiny-title-system.md) also gained an explicit instruction to preserve
ALL-CAPS acronyms verbatim, so a competent model short-circuits via the
verbatim set before post-processing kicks in.
Fixes#4220
Removed the live-context guard that let default model selection persist a new role without changing the active session model. The next prompt's compaction path now owns oversized-context recovery after a switch.
Fixes#4219
- Added `SpeakableStream` to strip markdown noise, silence code blocks and tables, normalize links and paths, and emit sentence/clause segments.
- Reworked `Vocalizer` to segment assistant deltas in the parent process, lazily open TTS streams, idle-flush partial thoughts, and chain playback sessions.
- Added gapless streaming playback with ffmpeg/sox backends, ducking-aware pacing, fallback file playback, and immediate stop handling.
- Added IPC `sendAndFlush` support and used it in the TTS worker so audio chunks drain before blocking ONNX inference resumes.
- Added speakable-stream coverage for markdown filtering, segmentation latency, idle flushing, and forced long-segment splits.
- Replaced `grep`, `glob`, and `ast_grep` `paths` inputs with optional single `path` strings while preserving default workspace-root behavior.
- Added shared `toPathList` normalization for legacy arrays and JSON-encoded arrays across tool execution and TUI renderers.
- Updated prompts, fixtures, shims, transcript summaries, and tests to send and display the new `path` argument.
- Updated collab-web search tool cards to read `path` while falling back to legacy `paths` for historical transcripts.
- Recorded the contiguous coding-agent changelog run for the tool-path breaking change and adjacent TTS entries.
Loaded cached runtime extension provider catalogs before deferred model resolution so dynamic-only providers can satisfy cold-start --model and session resume selection from models.db.\n\nFixes #4216
Some filesystems (NFS, FUSE, older SMB) return UV_DIRENT_UNKNOWN, making every Dirent.isX() return false. The prior early-return dropped real directories on those mounts. Now only reject when a known non-directory type (file/device/fifo/socket) is reported; symlink and unknown-type entries fall back to fs.statSync.
`discoverExtensionPaths` loaded the extension-module capability across all
registered providers (native, claude, codex, gemini, opencode) and then
discarded every item whose `_source.provider !== "native"`. Four foreign
directory walks ran on every session startup only for their results to be
dropped — worst on Windows.
Scope the load to the native provider via the existing `LoadOptions.providers`
filter: `loadCapability(extensionModuleCapability.id, { ...loadOptions,
providers: ["native"] })`. The hook-capability load in the same function is
unchanged (hooks legitimately span providers). Output is identical.
Regression test in `extensions-discovery.test.ts` spies on each extension-module
provider's `load()` and asserts only "native" is invoked.
Fixes#4198
collectExtensionModules already reads every own-source module's text to scan imports; the onLoad rewrite hook then re-read each file. Return a Map<path,source> from the collector and serve it consume-once in onLoad (delete on first hit, disk fallback on miss), so transitive modules are read once and the entry still re-reads on its ?mtime re-import. Adds a regression test asserting exactly one read per graph module.
Closes#4196
The overlay's directory cache stored names only, so listChildDirectories/searchDirectories called fs.statSync on every cached entry per keystroke to filter directories — a stat storm scaling with the target directory size and particularly costly on Windows.
Cache fs.Dirent[] via readdirSync({ withFileTypes: true }) instead of names, classify with entry.isDirectory(), and fall back to fs.statSync only for entry.isSymbolicLink() so a symlinked directory still counts as one. Adds a regression test that spies on fs.statSync and asserts an O(1) stat budget per keystroke.
Fixes#4199
Codex flagged (PR #2735): the placeholder conflict check only
consulted the internal deobfuscation map, not the full set of
configured secret literals. If secret A's minted placeholder happened
to equal secret B's raw configured value, obfuscating text containing
A leaked B's literal value verbatim: B's own plain-secret redaction
pass (sorted by length, once per obfuscate() call) already completed
before A's placeholder existed in the text, so the newly-minted
placeholder was never caught and redacted as B's secret.
Adds a #configuredSecretValues set populated with every configured
plain-secret literal (both obfuscate and replace mode, plus the
placeholder key) before any placeholder is minted. #placeholderConflicts
now also rejects a candidate (and its friendly-name-unprefixed alias)
if it equals any OTHER configured secret's value, forcing the
fallback-base retry loop to pick a different token.
Adds a regression test covering both entry orderings.
Codex flagged (PR #2735): when a replace-mode regex match fully
overlapped a placeholder freshly created earlier in the SAME
obfuscate() call, the redaction call site blanket-tagged the entire
redacted span's origin as 'I' (prior-call), including the preserved
placeholder bytes. A later regex entry over the same result then
wrongly treated that fresh placeholder as prior-call, applying the
greedy-spillover fixed-point skip and leaving genuinely new adjacent
content (e.g. SECRET in ABCDEFGHSECRET) provider-visible.
Threads the origin tag string through redactRegexMatchOutsidePlaceholders
and redactWithFixedReplacementOutsidePlaceholders (both now return
{text, origin} via a new shared transformOutsidePlaceholdersTracked
helper), so a placeholder preserved inside a redacted span keeps its
own existing origin tag instead of being overwritten.
Adds a regression test reproducing the exact reviewer scenario in a
single obfuscate() call across three chained entries.
- Persist signed message blocks (`text`, `thinking`, `toolCall`) and encrypted reasoning payloads verbatim during session serialization instead of clearing or truncating them.
- Preserve signature keys instead of replacing them with empty strings when they exceed persistence size limits.
- Exempt official first-party OpenAI and Anthropic API endpoints from the leaked-thinking stream healing wrapper to prevent misfires on legitimate visible text fences.
Codex flagged (PR #2735): when re-obfuscating text with a prior-call
placeholder, the outside-chunk independence check
(outsidePlaceholderRangesAnyIndependentlyMatch) tested only the literal
'#...#' placeholder token text. A regex like
'ABCDEFGHSECRET|(?<=ABCDEFGH)SECRET|ABCDEFGH' combined with a prior
placeholder for ABCDEFGH left a trailing SECRET provider-visible,
because the lookbehind (?<=ABCDEFGH) never matches the literal
placeholder token — only the expanded secret value satisfies it.
Fix tests each outside chunk in BOTH the literal placeholder-token
context (needed when the token's own non-word boundary completes a
boundary-sensitive pattern like \b[A-Z]{8}\b) and the expanded scan
context (needed when a lookbehind/lookahead only resolves once the
neighboring placeholder is deobfuscated), unioning the result.
Adds a regression test covering the scan-context-only case, distinct
from the three existing adjacent tests which are all satisfied by
literal-context alone.
- Excluded signed `thinking` blocks and `redactedThinking` blobs from size-based persistence truncation.
- Preserved signature-bound reasoning verbatim to prevent provider validation failures on session replay.
- Maintained normal truncation behavior for unsigned thinking and standard text blocks.
The branch-scan rehydrator only rebuilt `#lastCompletedRewind` and wiped
`#checkpointState` unconditionally at entry — so a branch whose latest
checkpoint had not yet been rewound came back with neither an active
checkpoint nor completed-rewind guidance. Reloading such a session (or
`switchSession()` on the same file) made the next `rewind` fail with
"No active checkpoint" even though the checkpoint entry was still the
branch leaf.
Extended the walker to also track the last unresolved checkpoint entry
and, when the branch ends without a rewind-report, seed `#checkpointState`
from that entry (id, `details.startedAt`) so `rewind` can complete
normally. Renamed the method to `#rehydrateCheckpointRewindState` to
reflect the widened responsibility and added a regression test that
truncates the branch to the checkpoint entry, resumes into a fresh
`AgentSession`, and calls `rewind` end-to-end.
Fixes#4187
- Refactored `renderSubagentHudLines` to use `renderTreeList` with dim connectors and a single-space indentation shift.
- Adjusted budget limits to account for the new layout wrapping and tree-list padding.
Cleared checkpoint rewind runtime state when starting new sessions or creating branch sessions so stale completed-rewind guidance cannot leak into unrelated contexts.
Added regression coverage for /new and branch reset paths.
Fixes#4187
Reconstructed the completed rewind marker from the active branch so resumed sessions keep repeat-rewind recovery guidance.
Covered resume rehydration with the checkpoint rewind branch regression test.
Fixes#4187
llama-server in router/preset mode advertises each preset via /v1/models,
but meta.n_ctx / n_ctx_train are only merged in after the preset's child
instance loads. The router-level /props returns a dummy n_ctx: 0. As a
result every unloaded preset fell through to DISCOVERY_DEFAULT_CONTEXT_WINDOW
(128000), and picking a preset from /model kept surfacing 128k in the
status bar regardless of the configured --ctx-size — a restart didn't
help because discovery repopulated the cache from the same broken chain.
Parse each entry's status.args (rendered CLI vector) for --ctx-size or
-c, and fall back to ctx-size = N in status.preset (INI). Positive values
slot between runtimeContextWindow and serverMetadata in the resolution
chain so a running child's live n_ctx still wins; --ctx-size 0 ("loaded
from model") is correctly skipped so we don't publish 0.
The same fallback wires through discoverLlamaCppModelRuntimeMetadata so
the refresh triggered by /model uses the configured window even before
the child spawns.
Fixes#4190
Reconstructed the completed rewind marker from the active branch so resumed sessions keep repeat-rewind recovery guidance.
Covered resume rehydration with the checkpoint rewind branch regression test.
Fixes#4187
- Refactored the mid-run todo nudge to trigger on mutating tools (bash, eval, edit, write, ast_edit) rather than overall tool turns.
- Simplified the nudge prompt template to a concise, non-escalating reminder.
- Migrated nudge messages from "developer" role with public events to a hidden "custom" role that is excluded from the TUI and transcript.
- Introduced a separate per-cycle reminder cap of 2 to decouple mid-run hints from the user-visible stop-time escalation budget.
- Avoided triggering the todo nudge when read-only exploration tools (e.g. grep, read, glob, lsp) or errored results are returned.
Wrapped retained rewind reports with completion guidance so the post-rewind turn knows the checkpoint is closed.
Added repeat-rewind recovery errors and regression coverage for both the retained context and no-active-checkpoint path.
Fixes#4187
- Reverted to flushing only on the last file write or explicitly on early failure paths within `apply_patch` multi-file operations.
- Refactored error counting logic within single path entries to use clean booleans instead of numeric counters.
- Replaced custom preview capping logic in task progress rendering with `capPreviewLines` and added an option to hide the expand hint.
- Instructed the tester agent to never write assertions or tests for default values, configurations, or fallback properties.
- Allowed the tester agent to skip writing tests entirely if the changes are trivial, already covered, or if any new tests would be worthless.
- Required the deletion of existing default-testing assertions or entire tests when modifying code that currently contains them.
- Introduced the `CollabGuestUiResult` type to distinguish between answers and unavailable states during guest UI requests.
- Updated the remote dialog race logic to ignore unavailable guest results and fallback to the local host dialog.
- Centralized the `FakeWebSocket` and `InMemoryRelay` test infrastructure into a shared test helpers file.
- Cleaned up duplicate mock implementations and updated existing test suites to utilize the shared in-memory relay helpers.
- Projected full tool-call arguments down to a compact summary containing only `command` and `path`.
- Truncated summarized argument fields to 200 characters to prevent inflating session log sizes.
- Replaced routine clean session disposal warnings with debug logs to reduce noise.
- Streamlined debug context in assistant message removal and agent continuation skip paths.
- Extracted duplicate user-facing compaction warning strings into a helper function.
- Consolidated duplicated inline thinking level comparisons into a unified `concreteThinkingLevel` helper.
- Enhanced legacy tool shims to respect isolated session settings and support legacy options.
- Cleaned up redundant UI render requests and extra status-line updates.
- Refactored `grep` tool shim to configure context dynamically via isolated settings.
- Disabled platform-incompatible shell shim tests on Windows environments.
- Introduced `RpcShutdownCoordinator` to track background tasks and manage deferred shutdowns safely.
- Guaranteed all background bash task response frames are fully written before the process exits.
- Re-checked shutdown requests automatically as each tracked background task settles.
- Latched the shutdown sequence to prevent concurrent execution from duplicate triggers.
- Updated `mock-rpc-agent` to consume stdin via an async iterator to match standard behavior.
- Added comprehensive unit tests in `rpc-input-frame.test.ts` covering background task coordination.
- Introduced JSON depth tracking to ensure string keys are only extracted at the top level.
- Prevented nested keys matching designated streaming keys from being prematurely or incorrectly captured.
- Added comprehensive unit tests validating nested key exclusion and correct top-level extraction order.
metaphorics