- buildOpenAiNativeHistory now emits custom_tool_call / custom_tool_call_output for blocks with customWireName (apply_patch and other freeform tools), matching the normal Responses replay path; previously demoted to function_call which broke remote-compaction replay or mismatched the original call.
- requestOpenAiRemoteCompaction and requestRemoteCompaction accept an optional AbortSignal; the coding-agent compaction caller forwards the existing signal so cancellation now terminates the in-flight fetch instead of stranding the session in the compacting state until the server replies.
`SessionManager.close()` queues `#closePersistWriterInternal()` on the
persist chain. The task awaits `#persistWriter.close()`, which flips
`#closing = true` synchronously before yielding on its inner writer
`close()`. A concurrent `appendMessage()` landing in that yield window
hit the hot path, got the still-cached (but closing) writer back from
`#ensurePersistWriter()`, and threw `Error("Writer closed")` from
`writeSync`. The throw was stashed into `#persistError`; the next async
caller (`flush()` or a later `appendMessage()`) re-threw it as an
unhandled rejection with the original line-1282 stack.
Expose `NdjsonFileWriter.isOpen()` and treat a mid-close cached writer
as a miss in `#ensurePersistWriter()`. `_persist` now falls back to the
async `#rewriteFile()` cold path so the entry — already in
`#fileEntries` — still lands on disk once the close drains.
- Added a dedicated `@sinclair/typebox` specifier remap path and routed bare legacy imports to the in-repo shim during extension rewriting and module resolution.
- Added resolve hooks for both `file` and legacy-file namespaces so legacy extensions load the shim consistently at runtime.
- Replaced fromTypeBox conversion with a JSON-schema validator flow in ai tool handling and execution paths.
- Added recursive schema validation and expanded TypeBox checks for refs, enums, uniqueItems, and constraint keywords.
- Sanitized Azure/CCA tool schemas by dropping unsupported fields and rewriting oneOf tool branches as anyOf.
- Tightened argument and model-config validation, preserving unknown tool fields and adding apiKey plus compatibility flags.
- Added canonical `pi.zod` schema API exports and removed TypeBox package exports/imports.
- Migrated Tool schema typing from TypeBox to shared `TSchema`/Zod flow with legacy TypeBox compatibility.
- Updated AI provider adapters and MCP/agent builders to convert tool params through `toolWireSchema()`.
- Reworked schema validation from AJV to Zod-safe parsing with `fromTypeBox`, `toolWireSchema`, and meta schema checks.
- Added a task.maxRuntimeMs setting with a default disabled state for per-subagent runtime limits.
- Updated subprocess execution to enforce the configured wall-clock timeout, mark timeouts as aborts, and include timeout-specific abort messaging.
- Tracked per-turn context size and context window through task progress/results and updated UI renderers to display current context against window with cumulative tokens rendered separately.
- Replaced custom MockAssistantStream helpers with createMockModel streams across agent tests.
- Removed manual queueMicrotask stream-event scripting in favor of scripted mock responses.
- Consolidated helper fixtures by deleting local aliases and reusing shared user-message/model helpers.
- Updated test assertions to use mock.calls and mock.model metadata for call and context validation.
- Updated plan-mode and hindsight session state lookups to use Array.findLast for selecting the latest assistant or user message.
- Updated postmortem callback iteration to use Array.toReversed before mapping cleanup callbacks.
- Added OpenAI remote-compaction API support with provider-specific endpoint gating.
- Added buildOpenAiNativeHistory, token-budget estimation, and message trimming for remote-compaction.
- Added helpers to preserve and validate remote-compaction metadata in request/response handling.
- Refactored coding-agent compaction to consume pi-ai remote-compaction helpers with converted message history.
- Exported remote-compaction from ai index and documented the new APIs in CHANGELOG.
- Replaced `GenAIAttr` with an `export const enum` in telemetry while preserving all GenAI attribute constants.
- Updated the OTEL stream test fixture to emit an `error` event with an `error` payload instead of a `done` event.
- Added runSubprocess telemetry propagation tests for inheriting parent telemetry and handling missing parent telemetry.
- Task tool sessions now expose and forward parent OpenTelemetry config when creating subagent tasks.
- Subprocess execution now derives child telemetry from the parent config with the subagent identity and child session conversation handling.
- Subagent creation now records a handoff span using the resolved parent telemetry handle before running the child loop.
- Adjusted OutputSink to disable head retention after replace(), resetting counters so later pushes append to the tail and do not trigger stale middle-elision in dump().
- Refined artifact link emission to insert a newline separator only when the minimized output lacked one.
- Added a regression test for replace-plus-push ordering that verifies no elision marker and aligned byte counts.
- Added opt-in telemetry configuration to Agent and session APIs, including Agent#setTelemetry mutator.
- Implemented OpenTelemetry spans for invoke_agent, chat, execute_tool, and handoff paths with metadata and step tracking.
- Added a telemetry helper module, OpenTelemetry request/usage types, and dependency wiring with no-op behavior when tracer SDK is absent.
- Added OTEL end-to-end tests and fixed coding-agent OutputSink realignment and artifact-link newline output issues.
When initialServiceTier is resolved from settings (e.g. the user has
serviceTier: priority configured as a default), sdk.ts wrote model_change
and thinking_level_change entries for the new session but omitted
service_tier_change. The stats parser derives priority-tier premium counts
from service_tier_change entries, so sessions started in fast mode without
an explicit /fast toggle had no tier entry and were undercounted in omp-stats.
Mirror the thinking_level_change pattern: write service_tier_change
alongside the other initial entries when initialServiceTier is set.
isUrlLikeSpecifier matched Windows absolute paths (e.g. `C:\foo`) because the URL-scheme regex `^[A-Za-z][A-Za-z\d+.-]*:` happily eats a single drive letter. When a legacy plugin extension imported a bare-specifier dep from its own `node_modules`, rewriteBareImportsForLegacyExtension resolved it to an absolute path, then toRewrittenImportSpecifier short-circuited pathToFileURL and embedded the raw Windows path into the mirrored TS source.
The TS string-literal parser then ate \n, \U, \y and friends, producing nonsense package specifiers like `C:Usersjames.ompagentextensionssupipowers\node_modulesyamldistindex.js` that Bun rejected with `Cannot find package …`. Net effect: every legacy extension that pulls in any node_modules dep failed to load on Windows.
Fix: reject `^[A-Za-z]:[\\/]` in isUrlLikeSpecifier before the URL-scheme test so drive-letter paths flow through pathToFileURL and reach the mirror as proper `file:///C:/...` URLs.
- Added a shared session command helper that aggregates extension, prompt, and skill slash commands.
- Updated ACP, extension UI, runtime-init, and task executor extension contexts to return session command data instead of empty arrays.
- Updated root marker detection to read the directory once and match glob markers against top-level entries, preventing recursive scans when checking project roots.
- Adjusted root-marker glob error handling to warn on directory listing failures and treat the marker set as absent when unreadable.
- Reduced marksman's configured warmup timeout from 15,000 ms to 2,000 ms in LSP defaults.
- Updated collectPayload to accept intervening blank lines as empty payload when requirePayload is true and no payload has been collected yet.
- Added a fallback path alongside the HL_EDIT_SEP check so blank insertions without a separator are tolerated as required-payload op payload lines.
Commit d43655541 made splitHashlineInputs silently drop sections whose
body has no operations, so '@<path>\n' alone now yields zero sections
and trips the 'exactly one hashline section' guard before reaching the
no-op or local-URL checks the tests intended to exercise.
- 'returns no-op error for unchanged content' now uses '= 1<hash>..1<hash>'
with the existing line as payload (true no-op via formatLineHash).
- 'returns a handled error when the source path is a local URL' appends
'+ EOF\n' so the section is emitted and resolveToCwd surfaces the
internal-scheme rejection.
- Updated hashline splitting to skip emitting a section when a path header has no following operations.
- Added a trimmed non-empty check so only sections with diff content are returned.
- Added tests for duplicate and trailing headers to ensure empty header-only chunks are silently dropped.
Brush-core applied POSIX parameter expansion to $env before dispatching a command, mangling PowerShell references like Write-Host $env:SystemRoot to :SystemRoot. Move the fix down to env-var application: every brush session now defines env=$env as an internal (non-exported) shell variable, so the bash expansion of $env yields the literal $env and PowerShell tokens reach the child intact. User assignments (env=prod; echo "$env:8080") still shadow the fallback in their command scope, so the POSIX bash contract is preserved.
Fixes#1079
- Added a `resolveSearchRepoScope` helper that uses an explicit `repo` when provided, skips defaulting when a query already contains a repo/org/user/owner scope qualifier, and otherwise resolves the current checkout via `resolveDefaultRepoMemoized`.
- Updated `search_issues`, `search_prs`, `search_code`, and `search_commits` to use the resolver before composing API queries, defaulting `repo` when omitted but silently falling back to an unscoped search on resolution failure.
- Documented the new search-repo defaulting rules in tool prompts, user docs, and the package changelog.
The agentic commit pipeline (`omp commit`) wrote its commit and then sat
spinning on Ctrl+C because nothing in `Commit.run()` drained the lingering
event-loop handles: `installH2Fetch()` keeps idle HTTP/2 sockets warm to
the provider, the Settings autosave timer can still be armed, and the
AgentSession's extension/runner machinery holds onto async-job and OAuth
refresh state even after `session.dispose()` releases what it knows how
to. Mirror the `runPrintMode` exit path from `main.ts` by calling
`postmortem.quit(0)` once `runCommitCommand` resolves so the CLI returns
to the shell, runs registered cleanup callbacks, then exits — same model
the non-interactive launch flow already uses.
Also widens the private bash-tool helpers' `notices` parameter to
`readonly string[]` so `bun check:types` keeps passing — the public outer
arm already accepted `readonly string[]` and an upstream commit had only
partially propagated the change.
Fixes#1041
The previous examples used " • " in the source file and inserted "·"
as the new content. Some agents were copying the middle-dot literally
into real edits as if it were format scaffolding, since the demo
inserts were near-twins of the existing string.
The new example uses TITLE = "Mr" → "Mrs" with "Dr" inserts:
ASCII-only, clearly distinct from any payload separator, and obviously
domain content rather than punctuation the model could confuse for
syntax. Every original op shape is preserved (single-line replace,
multiline replace, insert AFTER/BEFORE, append, delete, blank, both
anti-patterns). Pure prompt change; parser/schema/runtime untouched.
The ClaudePluginManifest interface was missing the `commands` field
(the standard Claude plugin format), and resolvePluginDir only checked
the legacy `slash-commands` key. Plugins declaring their command path
via `"commands": "..."` silently fell back to the hardcoded
`<plugin-root>/commands/` directory, which doesn't exist for
Claude-format plugins, so no slash commands were ever loaded.
Changes:
- Added `commands?: string` to ClaudePluginManifest.
- Changed resolvePluginDir to accept ReadonlyArray<keyof
ClaudePluginManifest> and iterate in priority order; the first
non-empty match wins.
- loadSlashCommands now passes ["commands", "slash-commands"] so
the canonical Claude plugin key takes precedence over the legacy one.
- Added two regression tests: one covering the `commands` key in
isolation, one verifying its precedence over `slash-commands` when
both fields are present.
Fixes#1076
- Updated `collectPayload` to treat blank lines inside a payload run as empty entries when subsequent payload text is present.
- Added lookahead-based end-of-run detection so blank lines before a non-payload operation remain section separators.
- Added tests verifying blank-line recovery inside payloads and non-consumption of trailing blanks between hashline sections.
Loaded cached standard provider discovery models into ModelRegistry at startup so retry fallback validation can resolve Ollama Cloud models that are already visible through --list-models.
Added regression coverage for cached ollama-cloud fallback selectors and fixed a readonly notices type error exposed by the focused type check.
Fixes#1052
- anthropic: add signal to AnthropicSearchParams, callSearch(), and
AnthropicProvider.search()
- exa: add signal to ExaSearchParams, callExaSearch(), and ExaProvider.search()
- jina: add signal to JinaSearchParams, callJinaSearch(), and JinaProvider.search()
- zai: add signal to ZaiSearchParams, callZaiTool(), and ZaiProvider.search()
- gemini: add signal to GeminiSearchParams, callGeminiSearch() and
buildInit() so both fetchWithRetry calls (initial + auth-refresh retry)
carry the signal
The five providers listed above never forwarded SearchParams.signal to the
underlying HTTP layer, so pressing Esc during a web_search call had no effect
and the session froze until the request resolved or Ctrl+C was pressed.
brave, kimi, perplexity, searxng, tavily, synthetic, codex, kagi, and parallel
already thread the signal correctly and are unchanged.
Fixes#1044
- Added stripOutputNotice to output-meta to remove appended truncation notices when output metadata is available.
- Updated bash, eval, browser, read, and ssh renderers to strip the notice before display so the styled warning line is not duplicated.
- Left fallback behavior unchanged so outputs without a notice continue through unchanged.
TypeScript 6 tightens mutability checks; the notices parameter was typed
as string[] but the caller held a readonly string[] (from the options
object). Widening both private method overloads to readonly string[]
satisfies the type checker without any runtime change — filter(Boolean)
works on readonly arrays.
discoverAgents() called listClaudePluginRoots() unconditionally, so agents
from Claude Code marketplace plugins appeared in /agents and the Agent
Control Center even when claude-plugins was listed in disabledProviders.
Guard the listClaudePluginRoots() call with isProviderEnabled("claude-plugins"),
returning an empty roots array when the provider is disabled — matching how
filterProviders() handles every other capability's provider set.
Added regression test that verifies both the enabled path (agents visible)
and the disabled path (agents absent) using a real temp-directory plugin
registry fixture.
Fixes#1075
- Updated `formatBashFixupNotice` to wrap the stripped-pattern warning in a `<system-warning>` wrapper.
- Reworded the notice to clarify output is already truncated and stderr is merged into stdout.
- Extended the Bash interceptor test to verify the warning tag appears for head/tail stripping.
can1357