- Added `hasSteeringMessages` to `executeToolCalls` polling so queued steering is detected without dequeuing.
- Retained fallback to `getSteeringMessages` by consuming messages when no peek callback exists.
- Added optional `hasSteeringMessages` config hook and limited steering checks to boundaries.
- Fixed interrupted tool-batch steering by keeping queued messages until boundary handling.
- Added idle text and image submissions to steer queueing when no input waiter exists.
- Auto-continued resumable sessions after queued steering and preserved submit metadata.
- Renamed all functions, types, and constants in @oh-my-pi/snapcompact to namespace-relative names (`snapcompactCompact` → `compact`, `renderSnapcompactFrames` → `renderMany`, `snapcompactFrameCount` → `frames`, `SnapcompactShape` → `Shape`, `SNAPCOMPACT_SHAPES` → `SHAPES`, …).
- Converted every consumer to `import * as snapcompact` member access: `agent/compaction.ts`, `coding-agent` `agent-session.ts`/`session-manager.ts`/`snapcompact-inline.ts`, and all affected tests.
- Renamed internal `geometry` locals to `geo` in `snapcompact.ts` to avoid TDZ collisions with the new `geometry` export.
- Updated `docs/compaction.md` prose and added a Breaking Changes entry to the snapcompact changelog documenting the full rename map.
- Added `renderSnapcompactFrames()` and `snapcompactFrameCount()` to @oh-my-pi/snapcompact for paging arbitrary text into PNG image blocks without dim-marker bookkeeping.
- Widened the agent loop's `transformProviderContext` hook to `(context, model) => Context` so per-request transforms can gate on the dispatch model's capabilities.
- Added `SnapcompactInlineTransformer` rendering the system prompt and large historical tool results as snapcompact frames on vision models: vision gate, per-provider image budgets, 3k-token floor, savings-margin gate, skip-last rule, and hash-keyed render caches swept to live tool calls.
- Added default-off `snapcompact.systemPrompt` and `snapcompact.toolResults` settings under a new Context → Experimental group, composed after secret obfuscation in `sdk.ts` so frames are built per-request and never persisted to session.jsonl.
- Added prompt stubs (`snapcompact-system-stub.md`, `snapcompact-system-frames-note.md`, `snapcompact-toolresult-note.md`) and unit tests covering frame paging, no-mutate guarantees, budget caps, gates, and render caching.
- Mapped Codex `end_turn:false` terminal events to `pause_turn` stop details in response stream parsing.
- Updated `agent-loop` to re-sample `pause_turn` turns, reset on tool calls, and cap continuations at 8.
- Added coverage for pause-turn mapping and continuation-capping in agent and AI stream tests.
- Extended `AgentTool.concurrency` to accept per-call resolver functions and resolved concurrency mode from each tool call, falling back to exclusive on resolver errors.
- Updated BashTool to schedule non-PTY calls as shared and PTY calls as exclusive so non-interactive bash calls can run in parallel within one message.
- Tracked in-use persistent shell sessions in the bash executor and routed overlapping calls on the same session key to isolated one-shot shells while preserving owner session availability.
The shake-strategy post-shake threshold check was reading
#estimatePendingPromptTokens([]) while #checkCompaction triggered on
calculateContextTokens(assistantMessage.usage). The local estimator
ignored block.thinkingSignature payloads (OpenAI Responses encrypted
reasoning items, Anthropic signed thinking blocks, etc.), so on a
thinking-heavy session the estimate sat ~0.9–2× below provider-reported
usage. Once the two straddled the threshold, the #2119 dead-loop guard
never fired, shake reported 'handled', and #scheduleAutoContinuePrompt
re-injected the auto-continue developer prompt every turn — 53 injections
in a real 25-minute repro session before an external timeout.
Thread the trigger's provider-anchored contextTokens through
#runAutoCompaction → #runAutoShake for the threshold and incomplete
paths, then evaluate residual pressure as triggerContextTokens −
result.tokensFreed with an 80% recovery-band hysteresis. Re-checking
against the raw threshold (even on the corrected metric) would still let
shake reclaim a trickle of the previous turn's elidable blocks and land
just under the line every turn; the band closes that oscillation.
As defense in depth, estimateTokens() now charges thinkingSignature and
redactedThinking.data alongside the visible thinking text so every
other site that uses the estimator (idle compaction, pre-prompt check,
status line) tracks provider usage on replay.
New regression test pins the contract; existing dispatch test bumped
its mocked tokensFreed so its happy-path scenario lands inside the new
recovery band.
Fixes#2275
- The `snapcompact-summary.md` prompt was removed from `packages/agent` to further centralize `snapcompact`-related concerns within the dedicated `@oh-my-pi/snapcompact` package.
- Introduced `isSnapcompactShape` for runtime validation of `SnapcompactShape` overrides, enhancing data integrity when loading configuration from dynamic sources.
- Updated README and native module comments to reflect the standalone `snapcompact` package structure.
- Added a new @oh-my-pi/snapcompact package and redirected compaction call sites to it.
- Added provider-aware snapcompact shape resolution for model-specific mixed-frame behavior.
- Added optional image detail support by extending ImageContent and passing hints through OpenAI providers.
- Added native snapcompact render options, including 5x8/8x8 font loading and palette/geometry controls.
Interrupting mid-tool execution (e.g. Enter with a pending steer) drained the steering queue into the dying run — it landed in history without a response and the post-abort resume saw an empty queue, so the agent stopped instead of continuing. Steering/follow-up/aside queue polls in runLoopBody and the post-tool-call check in executeToolCalls are now skipped once the run's abort signal fires, leaving the queue intact for Agent.continue().
Adds snapcompactCompact() in compaction/snapcompact.ts: instead of an LLM-generated summary, discarded history is printed onto dense 2576px PNG frames with the public-domain X.org 5x8 pixel font and re-attached to the compaction summary message as image blocks. Fully local — no model call; ~7x cheaper than raw text at near-parity recall. CompactionSummaryMessage now charges per attached frame in estimateTokens(), frames persist under preserveData.snapcompact with an 8-frame budget that evicts middle-out (session-head frame pinned so head and tail both survive). Rasterization and PNG encoding run in native code via renderSnapcompactPng().
Adds pruneSupersededToolResults() and the opt-in PruneConfig.supersedeKey hook: when a tool call shares a key with a newer one (e.g. a re-read of the same file), the older result is pruned even inside the protectTokens window and replaced with a [Superseded by a newer read of this file] placeholder. Adds readToolSupersedeKey() and the shared splitReadSelector() implementing the read-tool path/selector grammar (including the .. range alias and L-prefix forms) so selector-free reads supersede range reads of the same file and URL-scheme paths are exempt. Strips selectors before tracking in <read-files> compaction lists, so reads dedupe to the base path and match write/edit paths when splitting read-only vs modified lists (selector-polluted lists from earlier compactions self-heal on the next pass). Gated by the new compaction.supersedeReads setting (default on).
Splits the array-form defaultConvertToLlm into a single-message convertMessageToLlm that embedders can delegate every core role to instead of duplicating the conversion. Adds the optional images field on CompactionSummaryMessage so the converter attaches snapcompact frames after the summary text (snapcompact strategy lands next). Renames every convertToLlm call site in compaction.ts and branch-summarization.ts to the canonical defaultConvertToLlm.\n\nNote: the bundled CHANGELOG entries also cover the supersede-reads, snapcompact, and steering-queue fixes that follow in this batch (the entries land in directly adjacent lines and cannot be split by diff).
Anthropic rejects tool_result blocks when is_error is true and content is
empty after trimming whitespace. Fill a placeholder at encode time and in
coerceToolResult so wedged sessions recover on the next request.
Added AgentLoopConfig.getDisableReasoning so the agent loop refreshes disableReasoning on every model call, matching getReasoning. Mid-run thinking-level transitions in and out of off now propagate to the next request instead of sticking on the value captured at prompt start.\n\nFixes #2239
Propagated explicit thinking-off state through the agent loop so provider requests receive disableReasoning instead of an undefined effort. Added Ollama and agent-session regressions for the :off path.\n\nFixes #2239
Converted provider-facing tool parameters to wire JSON Schema before redaction so live Zod instances are not deep-cloned into plain objects.
Fixes#2146
- Replaced minLevel/maxLevel range with explicit efforts array plus baked effortMap/supportsDisplay wire facts.
- Removed runtime enrichment layer and modelOmitsReasoningEffort; providers now read baked fields.
- Fixed dotted Opus 4.7/4.8 ids missing adaptive display via classifier-based predicates (#1373).
- Bumped model cache schema to v4 to invalidate pre-efforts rows.
Move bundled models, model cache/manager, thinking metadata, effort helpers,
provider descriptors/discovery, wire constants, and model identity utilities
into the new @oh-my-pi/pi-catalog package.
Update pi-ai to keep provider runtime/auth concerns, move catalog provider
metadata into CATALOG_PROVIDERS, and migrate coding-agent, agent, stats, docs,
and tests to import catalog values from pi-catalog.
Split coding-agent model registry helpers into discovery, roles, and models
config modules while preserving registry orchestration.
BREAKING CHANGE: @oh-my-pi/pi-ai no longer exports catalog subpaths such as
/models, /model-cache, /model-manager, /model-thinking, /effort,
/provider-models*, discovery helpers, and provider wire constants; use the
matching @oh-my-pi/pi-catalog subpaths instead.
- Rewrote prescriptive prose to MUST/NEVER/SHOULD/MAY phrasing.
- Pruned internal mechanism the agent can't act on from tool prompts.
- Fixed garbled grammar and a stale plan-title placeholder.
- Made ssh tool description synchronous via cached host info.
Completes the injectable-fetch transport wiring (15.10.8) that the feature
left half-done, fixing the deterministic CI test failures:
- compaction.compact() rebuilt summaryOptions field-by-field but dropped
`fetch`, so the injected transport never reached
requestOpenAiRemoteCompaction / generateSummary's remote path. Thread it.
- Read-tool URL pipeline had no fetch seam: renderHtmlToText gained a
fetchOverride param but renderUrl/ToolSession never carried one, so the
jina/parallel reader backends always used global fetch. Add
ToolSession.fetch -> renderUrl -> renderHtmlToText (defaults to global).
- searchWithParallel mirrored extractWithParallel but missed the fetch
option; add it.
- Repair tests whose deleted hookFetch interceptors were never replaced
with a FetchImpl seam (fetch-kagi-toggle, web-search-parallel,
issue-970 discovery).
- Update issue-1746 POSIX case to the #2154 preserved-scrollback contract:
unknown-viewport streaming deferral is now platform-independent.
- Added optional FetchImpl fields to compaction, proxy, AI, coding-agent, and mnemopi options.
- Threaded injected fetch implementations through OAuth, discovery, and search/LLM request flows.
- Removed exported hookFetch utility and its package entrypoint from utils.
- Replaced global-fetch test monkeypatching with per-test FetchImpl mocks across test suites.
- Added `AssistantMessage.upstreamProvider` and emitted it as `pi.gen_ai.response.upstream_provider` telemetry.
- Clamped OpenAI-family output caps to `OPENAI_MAX_OUTPUT_TOKENS` and model limits to avoid oversized token requests.
- Kept committed tool-call blocks on TTSR/user-interrupt aborts, pairing them with a labeled placeholder result.
- Dropped incomplete tool calls only on anonymous `abort()` where partial args are unsafe to replay.
- Stopped aborted runs from waiting on provider iterator cleanup.
- Ran afterToolCall for completed executions after a run aborts.
- Ignored unknown Anthropic content blocks while preserving known ones.
- Kept sampling params and tool-result caching for disabled thinking.
- Clarified the `onRunEnd` documentation to note warning emission through `onTelemetryWarning` with `console.warn` fallback.
- Updated the non-fatal `onRunEnd` test to collect warnings via `onTelemetryWarning` instead of overriding `console.warn`.
- Added a short async pause in the test to let run-loop warning callbacks flush before asserting their presence.
- Updated `docs/ttsr-injection-lifecycle.md` to document `astCondition` behavior, including registration rules and tool-stream matching.
- Extended `packages/natives/test/native.test.ts` with new `astMatch` coverage for Smart matching, metavariable consistency, parse errors, and empty-language rejection.
- Deferred non-interrupting asides to the outer drain when a turn was finishing, preventing an extra model turn from firing before queued follow-ups.
- Coerced post-hook tool results before emission and only marked interrupted tool calls as skipped when execution failed, preserving completed call outcomes.
- Reworked Anthropic CCH patching with native `Buffer.indexOf`, and logged a warning while sending the body unchanged when the billing placeholder could not be anchored.
- Removed `maxToolCallsPerTurn` from `AgentOptions`, `AgentLoopConfig`, and config serialization.
- Removed stream-loop cap enforcement, including the `toolcall_end` abort path and capped assistant messages.
- Removed Anthropic Opus 4.8 batch-cap resolver and agent-session sync logic from coding-agent.
- Updated tests and changelogs to align with uncapped tool-call behavior and dropped cap-specific cases.
- Introduced `AsideMessage` as a message-or-thunk union so aside providers can defer injection decisions.
- Updated agent loop handling to resolve aside thunks at injection time and skip entries that returned `null`, then switched the session yield queue to `drainLazy` for deferred message building.
- Added tests validating lazy aside evaluation and staleness-aware dropping when everything becomes stale after dequeueing.
- Added a new aside-message source on `Agent` and exposed it in `AgentLoopConfig` as `getAsideMessages`.
- Updated the agent loop to poll aside messages after tool batches and before yielding, merging them with follow-up messages before continuing.
- Changed coding-agent session and yield queue handling to pull queued background messages via `drainMessages` at step boundaries instead of streaming-only injection.
- Updated `buildOpenAiNativeHistory` to maintain known and custom tool-call ID sets incrementally while appending provider payload history.
- Rebuilt call-ID state when a full-snapshot payload replaced history so stale outputs were no longer emitted after reset.
- Added compaction regression tests for codex provider payload call-ID registration and stale-result dropping.
- Scope.
- Summary.
- Type.
- Updated `executeToolCalls` to always pass the active `toolSignal` into `tool.execute` rather than bypassing it for non-abortable tools.
- Removed the `nonAbortable` option from `AgentTool` and from the read, write, and edit tools so they can no longer opt out of abort handling.
- Documented the cancellation behavior change in the affected tool docs and package changelogs.
- Updated `waitForDiagnostics` to accept exact document-version matches immediately and otherwise wait for a quiescence window before using the latest publish.
- Removed the old unversioned-acceptance option and applied the settle-based wait logic through inline and deferred diagnostics fetch paths.
- Added an LSP writethrough regression test ensuring stale unversioned diagnostics are ignored in favor of later fresh publishes.
- Updated `convertToLlm` logic to emit `developer` role for custom, hook, and file-mention inputs.
- Simplified OpenAI compact output filtering to retain only `user` and `assistant` messages, removing legacy `system-reminder` pattern checks.
- Adjusted compaction and session tests to match the new developer-role mapping and expected compacted content.
- Removed `<turn-aborted>` guidance injection from `transformMessages`, deleted `turn-aborted-guidance.md`, and dropped the synthetic abort note path for aborted/error turns.
- Updated `c`/`.` continue shortcuts to submit `manual-continue.md` as a hidden synthetic `developer` message via `session.prompt(..., { synthetic: true })` instead of sending an empty user turn.
- Updated tests and changelog notes in AI and coding-agent to reflect the revised abort-context and continue behavior.
The !sawTerminalEvent guard must check signal.aborted first and throw
the caller's reason (not a generic disconnect message) so that
Agent.abort('user-interrupt') surfaces the custom reason in
errorMessage instead of overwriting it.
- Export ProxyMessageEventStream class (avoid ReturnType<> per AGENTS.md)
- Use hookFetch from @oh-my-pi/pi-utils instead of manual globalThis.fetch mutation
- Use Promise.withResolvers() instead of new Promise for timeout races
- Use typed Context object instead of 'as never' cast
- Add client-initiated abort test (verifies stopReason='aborted' path)
- Remove 'as IteratorResult' cast on timeout sentinel
When the proxy SSE stream ends without sending a done or error
terminal event, streamProxy now throws an error instead of silently
calling stream.end() with no arguments. The thrown error is caught by
the existing catch block, which pushes an error event with
stopReason='error' and resolves finalResultPromise.
Previously, server disconnects produced a silent success: the stream
iterator completed normally with stopReason='stop' (the default) and
empty content, and finalResultPromise never resolved, causing
callers that awaited stream.result() to hang indefinitely.
This matches the pattern used by the codex provider
(openai-codex-responses.ts) which already guards against missing
terminal events.
- Added `EventLoopKeepalive` in `instrumentedCompleteSimple` to keep the event loop alive during oneshot completion calls.
- This prevented long-lived completion promises from freezing host spinners in Bun by keeping timer servicing active.
- Added `/tan` slash command registration and interactive handling.
- Added TanCommandController validation and async task scheduling for `/tan` dispatch.
- Added session cloning that suppresses breadcrumbs, copies artifacts, and handles abort cleanup.
- Added `promptCacheKey` support in Agent and inherited `providerPromptCacheKey` in session creation.
can1357