Track the non-message token estimate used for provider-anchored assistant usage, then add only positive current system/tool growth during pre-prompt threshold checks. Restore released collab changelog entries to the immutable 15.13.1 section.
Fixes#2628
Include the current system prompt and active tool schemas when provider-anchored pre-prompt checks estimate threshold pressure, and cover the case with a regression test.
Fixes#2628
Use provider-anchored context usage for pre-prompt context-full threshold checks when available, then add only pending prompt tokens. This keeps OpenAI Responses encrypted reasoning payloads from overcounting local prompt pressure while the visible context usage remains below threshold.
Fixes#2628
- Added `features.unexpectedStopDetection` and `unexpectedStopModel` settings for opt-in behavior.
- Added assistant-stop handling to classify stop reasons and resume generation with retry prompts.
- Added unexpected-stop classifier logic with candidate checks, model fallback, and YES/NO parsing.
- Added retry tracking that caps auto-continues at three attempts and logs a warning when exceeded.
- Added `jsonSchemaToTypeScript` and `renderToolInventory` to generate tool blocks with TypeScript signatures.
- Added `examples` and `TSchema` fields to dump-tool metadata and passed them through prompt rendering.
- Changed Harmony invocation rendering to omit `<|constrain|>json` markers in tool call payloads.
- Added compact native tool list-mode inventory rendering with full `# Tool:` output elsewhere.
The pre-prompt context check ran compaction directly, so snapcompact (or
any strategy) fired before auto-promote ever got a chance — defeating
Auto-Promote Context. It now tries promotion to a larger-context model
first (mirroring the post-turn threshold path) and only compacts when no
target is available.
Auto and manual compaction now project a snapcompact result's
post-compaction size (kept history + frames at the image budget + summary
+ non-message overhead); when it still exceeds the model's usable window,
they downgrade to a context-full LLM summary instead of leaving the
session overflowing.
- Removed the streaming guard that previously rejected /tan while the parent response was still generating.
- Passed "deliverAs: \"nextTurn\"" when sending the background dispatch breadcrumb and kept "triggerTurn: false" so an in-flight turn is not steered.
- Skipped rebuilding chat messages during streaming sessions and updated tests to cover the non-blocking dispatch path.
`#checkTodoCompletion` used to append a `<system-reminder>` and then call
`#scheduleAgentContinue`, so a text-only acknowledgement ("paused at your
instruction") triggered another `agent_end` that re-ran the same check and
fired the next reminder — counter ticked 1/3 → 2/3 → 3/3 inside a single user
pause without any user input. The user perceived three back-to-back reminders
appear from nowhere; the agent felt implicit pressure to invent busy-work or
take destructive ops to silence the loop.
Added `#todoReminderAwaitingProgress`: a reminder sets it, any `toolResult`
(real tool-level progress) or a new user prompt clears it, and
`#checkTodoCompletion` stays silent while it is set. Reset alongside
`#todoReminderCount` on user prompts, session reset, handoff, and the no-op
short-circuits in `#checkTodoCompletion` so the field never gets stuck.
Escalation through `todo.reminders.max` still works when the agent makes
tool-level progress between stops — that is the case the cap was designed for.
Regression coverage in `agent-session-todo-reminder-loop.test.ts` drives a
mocked `agent.continue` to mirror the bug-reported model behaviour and pins
the contract: exactly one reminder per user pause when the agent only
acknowledges; re-escalation when the agent actually calls a tool between stops.
Fixes#2590
- agent-loop: raise repetition-detection floor to 180 chars and clear thinking
replay anchors when collapsing a detected loop.
- providers/google: ignore empty text parts, retain terminal thoughtSignatures,
and stop function-call signatures clobbering the prior block.
- autolearn: capture goal-mode at the turn boundary; harden managed-skill writes
against hard-links/symlinks (O_NOFOLLOW + nlink); refuse minting managed skills
whose name an authored skill already claims.
- eager tasks: thread agentKind through the session so a custom top-level agentId
still gets always-mode delegation; split Eager Tasks prompt into hard vs soft.
- title-generator: race the online title model against a local tiny-model fallback.
- eager-todo: keep the soft reminder aligned with the todo init schema.
- mcp/stdio: keep close() detaching the read loop instead of awaiting it.
- stream loop: fix collapsing and tool-call thought-signature handling.
- Kept queued resolve handlers pending when a downgraded forced tool choice completes without invoking the requested tool.
- Covered the skipped-tool and invoked-tool queue paths in tool-choice queue tests.
Fixes#2546
The sqlite_master probe runs before any other statement and can take a
read lock or trigger WAL recovery, so under the same concurrent-startup
race fixed in #2421 it could throw SQLITE_BUSY before PRAGMA
busy_timeout ran. Install the busy handler immediately after opening
the Database, matching every other init path.
Fixes#2421
- Added a generation-aware early-exit check in `#waitForPostPromptRecovery` to return when the prompt turn is superseded.
- Passed the current prompt generation into the post-prompt recovery wait after prompting.
- Ensured aborted prompts no longer block while later queued retry/follow-up turns drain.
Two controller bugs from review:
- The post-stop nudge still queued a passive `nextTurn` message during goal
mode (goal mode only disabled `autoContinue`). That message rides the goal
continuation and can divert the goal loop into capture. Return early when
goal mode is active.
- `#suppressNext` was latched before the fire-and-forget `sendCustomMessage`.
It must arm synchronously (the synthetic turn's `agent_end` fires inside
`sendCustomMessage` before it resolves), but a rejected or *deferred* dispatch
(ACP clients downgrade `triggerTurn` to a queue) then produces no `agent_end`,
so the latch swallowed the next real stop. `sendCustomMessage` now returns
whether it actually started a turn; the controller disarms the latch when no
turn ran (rejection or deferral).
`task/executor.ts` widens its pending-message array to `Promise<unknown>[]` to
absorb the new return type (the resolved values are discarded).
Addresses review threads on PR #2542 (threads 5, 9, 12).
- Routed `customType: "handoff"` messages to the compact divider path in Agent Hub and UI helpers.
- Added handoff summary expansion that extracts context text and strips `<handoff-context>` wrappers.
- Refactored shared divider rendering into `SummaryDividerComponent` used by compaction and handoff messages.
- Added a `snapcompact-savings.jsonl` append-only journal for per-tool savings records.
- Extended `SnapcompactInlineTransformer` to compute `savedTokens` and emit savings to an optional sink.
- Integrated `createSnapcompactSavingsRecorder` into `createAgentSession` for session-aware tool-result metrics.
- Skipped journaling entries without a session and deduped duplicate `toolCallId`s per recorder lifecycle.
- Added tests for savings-sink emissions and savings-journal read/write behavior.
- Added session-domain modules and exports for session-entries, context, listing, loader, and migrations.
- Changed persistence to async append writes plus writeTextAtomic, removing sync line APIs.
- Added compaction-aware session context rebuild with dangling tool-call cleanup.
- Added resumable session resolution with status inference, id/stem/suffix matching, and backup recovery.
- Validated queued toolChoice against active tools in agent and coding-agent sessions.
- Rejected queued forced choices with reason "unavailable" when selected tools were inactive.
- Dropped provider toolChoice payloads when requested function tools were not offered.
- Probed Tokio worker-thread support and fell back to current-thread runtime creation.
Adds a fastModeScope setting (both|openai|claude, default both). setFastMode(true)
derives the service tier from it (both->priority, openai->openai-only,
claude->claude-only) instead of hardcoding unscoped priority; the off path and
the already-on no-op are unchanged. /fast status now reports the active scope.
Default both preserves existing behavior.
- Fixed initial assistant persistence by synchronously materializing in-memory entries and keeping a writer open.
- Fixed mid-close append handling to write entries through a one-shot sync writer instead of queuing a rewrite.
- Fixed persist-task gating by tracking pending writes before starting immediate persistence.
- Queued steering now drains after session settlement, so aborted auto-continued turns no longer leave queued messages stranded.
- Resumable-state detection now treats tool-result messages as resumable so continue can process queued steering after an interrupted tool execution.
- Regression tests were added for queued steer draining after abort and after an interrupted tool result.
- Replaced queued-message interrupt flow with session abort calls on empty submit and escape.
- Removed interrupting state and notifyInterrupting teardown paths from abort handling.
- Updated AgentSession queue operations to use shared steering and follow-up queue views.
- Propagated isAborting through session state and collab payloads to suppress late updates.
- Filtered dot-only or blank thinking blocks so they no longer render as assistant thought.
- Adjusted assistant-message and streaming-reveal logic to use visible-thinking helpers for consistency.
- Coalesced concurrent interruptAndFlushQueuedMessages() calls through one in-flight promise.
- Replaced continue()-retry logic with agent.prompt() to flush queued messages from empty contexts.
- Skipped queued-message flush replay while compacting or streaming to avoid turn overlap.
- Updated flush path to consume queued steering first, then follow-ups, via dequeuing helper.
- Added a regression test for empty-state interrupt-and-flush delivering queued steers safely.
- Coalesced repeated interrupt-and-flush calls into one queued-steer resume flow.
- Retried queued continues on AgentBusyError after waitForIdle up to a 30s timeout.
- Handled queue-flush failures in input controllers with warning logs and TUI error display.
- Updated AgentSession.setModel and setModelTemporary to validate credentials with hasConfiguredAuth instead of eagerly resolving API keys.
- Changed ModelRegistry.hasConfiguredAuth to probe configured credentials without executing command-backed key programs or refreshing OAuth tokens.
- Added model switch auth tests that verify resolver calls were removed and unconfigured models are rejected synchronously.
- Replaced unknown model contextWindow/maxTokens sentinels with nullable values across types and catalog data.
- Mapped request token calculations to treat null maxTokens as unlimited output caps.
- Updated remote compaction and context checks to ignore unknown limits by using Infinity/0 fallbacks.
- Adjusted CLI/model registry flows to skip cap enforcement for null limits and render unknown values as '-'.
- InputController now acknowledged empty-submit interrupts before flushing queued steering or queued-abort paths to preserve interrupting feedback.
- EventController now recreated the loading animation on interrupt acknowledgement and deferred end-of-turn teardown while queued resume recovery was still arming.
- AgentSession exposed an isResumingQueuedMessages phase around interruptAndFlushQueuedMessages so stale agent_end events no longer tear down the loader prematurely.
Concurrent omp --session restores after an unclean shutdown crashed
in SqliteAuthCredentialStore.#initializeSchema() with
SQLITE_BUSY_RECOVERY because the multi-statement schema run installed
PRAGMA busy_timeout=5000 AFTER PRAGMA journal_mode=WAL, the first
lock-taking statement during WAL recovery. Bun's default busy_timeout
is 0, so the lock conflict surfaces immediately.
- packages/ai/src/auth-storage.ts: hoisted PRAGMA busy_timeout to a
standalone first statement, dropped it from the multi-statement
schema run, wrapped SqliteAuthCredentialStore.open() in a 4-attempt
exponential-backoff retry loop on the SQLITE_BUSY family, and the
exhausted-retry error now includes the DB path. Exported
isSqliteBusyError(err) (matches code prefix 'SQLITE_BUSY').
- packages/coding-agent/src/session/agent-storage.ts: same hoist and
the existing retry loop now uses isSqliteBusyError so
SQLITE_BUSY_RECOVERY / _SNAPSHOT / _TIMEOUT also trigger backoff.
- Hoisted busy_timeout before journal_mode=WAL in every other shared
SQLite open path: history-storage, autoresearch/storage,
memories/storage, github-cache, report-tool-issue (auto-QA),
catalog/model-cache; stats/db.ts now sets busy_timeout at all.
- packages/ai/test/auth-storage-sqlite-busy.test.ts pins the contract:
isSqliteBusyError matches every BUSY extended code (rejects
SQLITE_LOCKED, non-errors, strings); open() leaves the connection in
WAL mode (proves busy_timeout ran before journal_mode); open() retries
through synthetic SQLITE_BUSY_RECOVERY; non-BUSY errors (SQLITE_CORRUPT)
short-circuit; exhausted retries throw an error mentioning the DB path
with exactly 3 sleeps for a 4-attempt budget.
Fixes#2421
- Re-polled steering at the loop yield boundary and included it in the pre-stop pending batch so late messages are processed immediately.
- Added session-side draining for stranded queued messages, scheduling an auto-continue when a prompt settles and follow-ups or steers remain.
- Added a regression test for late steering injection at yield and updated mid-turn collab prompt handling to keep steering messages in the pending display queue until consumed.
- Added optional `useless` flags to tool result types and payload builders.
- Added `pruneUseless` and `dropUeless` options to control uneventful result pruning.
- Changed compaction and shake passes to prune or ignore non-error useless tool results.
- Changed conversation serialization to omit useless toolCall/toolResult pairs from output.
- Added coverage for useless tagging, pruning, and serialization behavior.
- Added a SessionFocusController to switch transcript and input context between main and subagent sessions.
- Added agent-hub Enter activation and double-left return behavior for focused local agents.
- Added view-session-based event and render logic to avoid stale focus-session state.
- Added status-line focused agent display with ghost icon and focused-mode border dimming.
- Updated `codexResets.autoRedeem` from a boolean to `unset`/`yes`/`no` in schema and settings types, with updated description text.
- Adjusted Codex auto-reset evaluation to skip checks in `no` mode and to prompt once in `unset` mode before spending, while preserving immediate spend for `yes`.
- Migrated legacy boolean config values to tri-state during settings load and added tests for mode defaults, helper behavior, and migration results.
- Added `src/export/share.ts`: `/share` now snapshots the session JSON, gzips and seals it with a fresh AES-256-GCM key, and pushes the blob to a secret gist or the share server (1 MB cap with image/string/entry truncation via `sealToFit`); links are `<serverUrl>/<id>#<key>` with the key only in the fragment.
- Added `share-loader.js` and `scripts/generate-share-viewer.ts` building the static viewer the relay serves at `GET /s/<id>`: it fetches the sealed blob, decrypts in-browser, and hands the JSON to the export template via `window.__OMP_SESSION_DATA__`.
- Reworked the `/share` command in `command-controller.ts`/`builtin-registry.ts` off the plaintext-gist HTML upload, exported `LoadedCustomShare`, and exposed the session `SecretObfuscator` getter on `AgentSession` for redaction.
- Added `share.serverUrl` and `share.redactSecrets` settings backed by `DEFAULT_SHARE_URL` from pi-wire.
- HTML exports now embed subagent transcripts: `collectSubSessions` walks `<session>/<AgentId>.jsonl` recursively into `SessionData.subSessions`, with `includeSubSessions` opt-out and the exported `buildSessionData` reused by share snapshots.
- Added `share.test.ts` (snapshot/seal/server-url contracts) and `export-subsessions.test.ts`.
- Recomputed `tools.discoveryMode: "auto"` in the deferred MCP closure in `sdk.ts` once the real tool count is known: a toolset crossing the threshold now flips discovery on, registers and activates `search_tool_bm25`, and skips `activateAll` instead of force-activating every MCP tool.
- Guarded the deferred MCP task against disposed sessions: added `AgentSession.isDisposed` and `enableMCPDiscovery()`, and the late connect now calls `disconnectAll()` instead of refreshing tools onto a dead session.
- Cleared `#fastPathKey`/`#fastPathItems` in `AssistantMessageComponent.invalidate()` so theme/symbol changes rebuild reused Markdown children instead of keeping stale captured themes.
- Memoized unusable read summaries as a `false` sentinel in `read.ts` so the per-session LRU no longer retains full sources of unsummarizable files.
- Broadened `HAS_REF_DEF` in `markdown.ts` to match backslash-escaped reference labels (`[a\]b]: x`) and cleared frozen stream-lex state on blank `setText()`.
- Added regression tests: deferred auto-discovery flip and mid-connect dispose (`sdk-mcp-auto-discovery.test.ts` + `many-tools-mcp.ts` fixture), fast-path child rebuild on invalidate, and escaped-ref-def incremental-lex equivalence.
- Added optional `isError` markers on inline tool-result and message models.
- Skipped error-marked tool results during swap planning, savings estimation, and live transformations.
- Added tests confirming large error tool results stay text-only and are not considered for inline swaps.
- Updated snapcompact selectors and previews to pass `ShapeTarget` into `resolveShape` so `auto` is model-tuned.
- Applied `providerFrameBudget` in session compaction so generated archives stay within provider image caps.
- Replaced inline hard-coded image limits with `providerImageBudget` and skipped rasterization at cap.
- Added OpenRouter inline-transformer tests for cap exhaustion and existing-image budget exhaustion behavior.
