Adds client-capability-gated tool availability: gateway clients declare
capabilities at connect (new inline-widgets cap), chat.send stamps them into
the run context, and every tool assembly path (embedded runner, queued
followups, Codex app-server harness, plugin-only construction plans) drops
tools whose requiredClientCaps the originating client did not declare. The
Canvas plugin ships the first such tool, show_widget: agents pass SVG or an
HTML fragment plus a title; the plugin hosts it as a bounded, retention-scoped
Canvas document and returns the existing canvas preview handle, which web chat
renders as a sandboxed iframe fitted to the widget's reported content height.
Widget frames never get allow-same-origin (per-preview sandbox ceiling,
including the sidebar path) and the Canvas host serves widget documents with a
CSP sandbox header so direct navigation runs in an opaque origin. Verified
live end-to-end on a Testbox with gpt-5.5 (screenshots on the PR). CLI-backed
model backends do not carry client caps yet and stay fail-closed (#102577).
Closes#101790
* fix(mistral): use truncateUtf16Safe for error text truncation
Replace naive .slice(0, maxChars) with truncateUtf16Safe() in
truncateErrorText() to prevent surrogate pair splitting in
Mistral provider error messages shown to users.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(ai): keep Mistral errors UTF-16 safe
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* feat(secrets): resolve SecretRef model credentials at egress via process-local sentinels
SecretRef-managed model-provider credentials now travel as opaque
oc-sent-v1 sentinels through auth storage, stream options, and SDK
config; the guarded model fetch injects real values into headers and
URLs immediately before the SSRF-guarded send and fails closed on
unknown sentinels. packages/ai adapters converge on the host guarded
fetch where the SDK supports custom fetch and unwrap at construction
where it does not. Resolved values (and their percent-encoded forms)
register for exact-value log redaction. Kill switch:
OPENCLAW_SECRET_SENTINELS=off. Also fixes a pre-existing unhandled
rejection race in capNonOkResponseBodyLazily (pipeThrough writer leak).
* test(plugin-sdk): update public surface budget
* feat(cron): add headless code-mode driver and trigger-script evaluator
Part A of cron event triggers: runCodeModeScriptHeadless runs a script
to completion (exec/settle/resume, no snapshots, no session), plus the
cron-facing evaluator with per-job tool catalog cache, semaphore, 16KB
state cap, and closed failure taxonomy.
* fix(cron): correct trigger-script bootstrap flags and cache narrowing
* feat(cron): add event triggers (polled condition-watcher scripts)
Part B: trigger field on cron jobs gated by cron.triggers.enabled;
timer evaluates the script each due tick, quiet ticks leave no run
history, fired runs append the script message to the payload; once
semantics, min-interval floor, SQLite columns, RPC/CLI/agent-tool
surfaces, and docs.
* fix(cron): propagate triggerEval through startup catch-up outcomes
* fix(cron): honor cron staggering on quiet trigger ticks; fix trigger test types
* fix(cron): reject with Error reason in trigger-script abort test
* fix(cron): regenerate protocol/docs/snapshot artifacts and break madge cycle for trigger types
CI fixes for #101195: trigger evaluator result types move to the cron
types leaf (madge cycle), cron tool schema inventory gains trigger,
Swift protocol bindings + docs map + Linux prompt snapshots regenerated.
* fix(cron): drop underscore-dangle names in trigger code sync assert
* fix(cron): adopt registerHeadlessToolSearchCatalog after tool-search symbol localization
Upstream #101831 made registerToolSearchCatalog module-private; fold the
headless ref-only catalog registration into one public seam.
* fix(gateway): support native Windows exec approvals
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
* chore: defer changelog entry to release
* test: use tracked approvals temp directories
---------
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
sessions.delete gains an explicit archivedOnly param (additive protocol
schema change, Swift models regenerated): with archivedOnly=true, the
dispatcher grants operator.write and the handler requires the target to
already be archived (archive-then-delete keeps destructive intent a
two-step action). Without the flag nothing changes: deletes require
operator.admin, so internal fallback/synthetic dispatch, subagent cleanup,
and CLI minting keep their admin contracts, and the session-kill HTTP
endpoint pins admin explicitly since it terminates live runs.
Android sends archivedOnly and offers Delete only on archived rows, where
its bounded (non-admin) operator session is now authorized; active rows
archive first. iOS/web connect with admin and keep unrestricted deletes.
Refs #100712
* fix(anthropic): resolve thinking as disabled when legacy budget is below 1024
When adjustMaxTokensForThinking collapses the thinking budget below the
Anthropic minimum (1024), option resolution now sets thinkingEnabled to
false instead of always forcing it to true. This keeps every downstream
consumer (payload, replay, temperature, tool-choice) consistent — they
all see the same disabled state instead of an enabled flag with a
missing or API-rejected thinking block.
|| → ?? in both builders is defensive: the resolution layer already
prevents invalid budgets from reaching the builder through the normal
path, but ?? preserves an explicit zero when the builder is called
directly.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix(anthropic): guard raw streamAnthropic builder path against sub-minimum budgets
Add budget guards in both Anthropic payload builders so direct
streamAnthropic and bundled-plugin callers (e.g. Mantle) that bypass
option resolution also get the disabled-state rule instead of producing
API-rejected { type: "enabled", budget_tokens: < 1024 } requests.
Add proof-anthropic-thinking-budget.mts driving real production
functions with terminal output and negative control.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix(anthropic): normalize legacy thinking budgets
Co-authored-by: Pick-cat <huang.ting3@xydigit.com>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* fix(memory-host-sdk): handle stdout/stderr stream errors in runCliCommand
Register error handlers on stdout and stderr streams in
runCliCommand() to prevent uncaught exceptions when a pipe
breaks (e.g. EPIPE after child process exit).
Without these listeners, Node.js throws an uncaught exception
that crashes the process.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(memory-host-sdk): extract listenForChildOutputErrors helper
Refactor the inline stream error handling into a reusable
listenForChildOutputErrors helper, matching the pattern
established in the merged #101370 (agent-core).
The helper registers error listeners on stdout/stderr via a
shared loop, keeping the pattern consistent across the codebase.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(memory-host-sdk): harden qmd stream failures
* fix(memory-host-sdk): harden qmd stream failures
* docs(changelog): note qmd stream hardening
* chore: keep release changelog out of contributor PR
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* fix(agent-core): handle stdout/stderr stream errors in harness exec
Register error handlers on stdout and stderr streams in
NodeExecutionEnv.exec() to prevent uncaught exceptions when
a pipe breaks (e.g. EPIPE after child process exit).
Without these listeners, Node.js throws an uncaught exception
that crashes the entire agent-core process.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(agent-core): address review — hoist-safe mock and curly lint
- Use vi.hoisted pattern for spawnMock to fix ReferenceError in CI
- Wrap one-line if guard in braces per repo curly rule
Co-Authored-By: Claude <noreply@anthropic.com>
* chore: keep release notes in PR context
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
Prevent user-authored [cron:] text from clearing reset or deleted session memory while preserving structured cron-run opacity.
Co-authored-by: ly-wang19 <ly-wang19@users.noreply.github.com>
The active user turn was decorated at request-build time with an inbound
metadata block (Conversation info / reply-target / sender / forwarded /
chat-history), but the session store persists the bare text and the LLM boundary
strips that block from historical replay. So a user message's serialized bytes
changed retroactively — decorated as the active turn, bare as history — on every
turn, invalidating any prefix/exact-match provider prompt cache from that point
onward, every turn. (#90811 fixed the timestamp half of this same asymmetry.)
Rework of the earlier "strip the active turn" approach (which lost model-visible
context — ClawSweeper's P1) into one that keeps the context AND recovers the
cache cost:
- Route current-turn inbound metadata out of the user prompt and into the hidden
runtime-context custom message (the mechanism from openclaw#89428), and
relocate that carrier to the ABSOLUTE TAIL of the wire request (after the
active user turn and any tool-call scaffolding). The user turn stays bare and
byte-identical in both the active and historical positions; the volatile
carrier vanishes next turn exactly where the assistant reply begins anyway, so
the request is an append-only prefix-extension through the active user turn.
- A durable marker (UserMessage.runtimeContextCarrier, set by convertToLlm from
the carrier's details) lets the Anthropic SDK transport, the managed Anthropic
transport (anthropic-transport-stream / anthropic-payload-policy), and
OpenAI-completions skip the carrier when selecting the deepest cache_control
breakpoint, keeping the anchor on the last stable user turn.
Storage stays BARE (invariant preserved); runtime-only turns (room events) keep
their existing inline behavior, which is byte-stable because room-event/system
context is not strip-eligible; legacy bare-stored sessions are unchanged.
Reviewed with adversarial gpt-5.5/codex sweeps (correctness, byte-identity,
compatibility, test adequacy): found + fixed the managed Anthropic transport
cache-anchor path and added the missing OpenAI-completions/managed-transport
breakpoint tests; a theoretical runtime-only mutation was verified unreachable
and pinned with a test. Final sweep clean.
Co-authored-by: openclaw#89428 (runtime-context handoff approach)
The ChatGPT backend routes requests by session_id (codex-cli sends it); without it each request lands on an arbitrary machine and the prompt cache misses. Measured on live traffic: 56.0% -> 81.8% TRUE cache hit rate (12-turn A/B, identical prompts).
Tracks per-skill usage from the skill.used diagnostic event (trusted-only
delivery, file-scoped identity), sweeps workshop-created skills daily from
gateway maintenance (active -> stale 30d -> archived 90d, pinned bypass,
restore-only unarchive, files never touched), filters archived skills from
snapshots fail-open, reports workspace-scoped overlap candidates, and adds
openclaw skills curator CLI, additive gateway methods, and a warn-only
doctor finding. Zero new config keys; SQLite/Kysely state only.
Compaction failure and timeout recovery can no longer leave a session transcript assistant-last and wedged after restoring pre-compaction state.
Transcript-continuation failures now carry a typed agent-core error code and no longer demote the active model through model fallback scoring.
Surface: embedded agent runner compaction recovery, model fallback classification, packages/agent-core.
Refs #100312. Includes regression coverage for #99943.
