Github

<pre style='white-space:pre-wrap;width:81ex'>fix: prevent cold image analysis from timing out during setup (#133045) * fix: prevent cold image analysis from timing out during setup * test: keep agent tool type checks within the shard budget</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(talk): initialize browser voice for OpenAI login (#133048)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix: Responses sessions compact before reaching context limit (#130993) * fix(agents): prevent false Responses compaction * fix(ai): reject incomplete Responses usage snapshots * fix(agents): keep staged compaction progress alive * fix(agents): align recovery compaction timeout ownership * fix(agents): preserve custom compaction watchdogs * fix(compaction): preserve timeout and task state ownership * fix(compaction): restore omitted latest requests *

<pre style='white-space:pre-wrap;width:81ex'>test: reuse built runtime in agent exec E2E (#133044)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(models): verify native CLI auth in detailed lists (#130749) Detailed model listings now use the provider-owned native CLI authentication result for configured CLI routes. Anthropic API credentials no longer satisfy a separate Claude CLI route. The default model list remains lazy and starts no native authentication probe. Co-authored-by: Marvinthebored <peter@lindsey.jp> Co-authored-by: Ayaan Zaidi <hi@obviy.us></pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(ui): label the configured default in the session permissions picker (#132989) * feat(ui): label configured default session permissions Expose the configured exec posture as optional agent display metadata and show it in the chat and New Session permission pickers. Preserve plain Default for allowlist and non-equivalent security/ask pairs. Keep session modes unset when inheriting config; the label does not affect authorization, execution, filesyst

<pre style='white-space:pre-wrap;width:81ex'>test(agents): isolate archive browser cleanup dependency (#133042)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): restore system-agent delegation in channel turns (#128084) * fix(agents): isolate probe runtime generation * test(qa): prove channel system-agent delegation --------- Co-authored-by: VACInc <3279061+VACInc@users.noreply.github.com> Co-authored-by: RoboClaw <309084314+roboclaw-bot@users.noreply.github.com></pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(mantis): add long-held active turn recipe (#128117) Capture the proven mechanics from proof run 32616859410 (PR #127950) as a reusable recipe: dual 600s timeout config patch, Code Mode exec events file, streamed chunkDelayMs hold, and the observe-loop pattern around the 60s cap. Require agents to record recipe-suggestion.md after scenario-mechanics iteration so future discoveries persist.</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(doctor): migrate legacy Claude CLI allowlists to canonical refs (#126866) `agents.defaults.modelPolicy.allow` seeded from a legacy model map kept `claude-cli/*` refs that doctor never rewrote, while the model map migrated to canonical `anthropic/*` keys — leaving no selectable Claude model after upgrade. `openclaw doctor --fix` now canonicalizes explicit allowlists in the same pass: legacy refs migrate to provider refs with their claude-cli runtim

<pre style='white-space:pre-wrap;width:81ex'>fix(cli-runner): preserve partial output on aborted Claude live turns (#126883) Claude CLI live turns that were aborted (user /stop) or hit a caller deadline rejected in failClaudeTurn, so assistant text the streaming parser had already accepted was dropped from the transcript. The turn now resolves that partial output with a terminalInterruption disposition (aborted | timeout, classified by the canonical signal-reason rule) and the CLI runner settles

<pre style='white-space:pre-wrap;width:81ex'>feat(mantis): give the proof agent a developer shell and gateway restart (#128094) * feat(mantis): add exec and restart lane primitives Give the proof agent a developer shell inside each SUT container and an in-container gateway restart so it can design scenarios like a local developer: patch openclaw.json and restart, stage plugins and fixtures, run node/tsx against the read-only repo root, inspect SQLite state. - container script: exec (docker exec

<pre style='white-space:pre-wrap;width:81ex'>fix(heartbeat): deliver exec completions with disabled cadence (#126895) With agents.defaults.heartbeat.every set to "0m", a background exec completion queued its system event and requested a targeted exec-event wake, but the shared unscheduled-wake policy only admitted immediate wakes, so the follow-up agent turn silently never ran. Admit the exact exec-event/event producer shape (session or agent target required) in one per-source predicate, dispatc

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): resume yielded requesters after visible spawns (#128096)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): bind failed wait recovery to tool call (#128098)</pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(mantis): publish trusted lane fact digests and a differential line (#128089) Pass verdicts rested on agent prose only. The evidence builder now digests each lane's trusted mantis-lane-facts.json (sends, bot messages, edits, deletes, provider requests, injected Bot API faults, observed seconds, attempt, sanitized user inputs) into an additive per-lane digest field and a comparison.differential line listing the counts that changed between baseline

<pre style='white-space:pre-wrap;width:81ex'>fix(web): keep short multiline user messages expanded (#128085) * fix(web): keep short user prompts expanded * test(ui): drop non-regressing disclosure cases</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(ui): keep one agent run in one transcript response (#126278) * fix(ui): compose agent run transcript responses * fix(ui): preserve explicit run transcript boundaries * fix(ui): preserve run transcript status * fix(ui): preserve live agent run content * fix(ui): preserve semantic agent run frames * fix(ui): complete agent run transcript ownership * fix(ui): stabilize agent run transcript lifecycle * fix(ui): complete agent run frame ownership * fix

<pre style='white-space:pre-wrap;width:81ex'>fix(gateway): restore HTTP subagent completion announces [AI-assisted] (#128068) * fix(gateway): restore HTTP subagent completion announces * test(gateway): prove HTTP resolver retires with instance</pre>

<pre style='white-space:pre-wrap;width:81ex'>improve(ui): align chat branch and PR rows (#125232) * improve(ui): align chat branch and PR rows * improve(ui): collapse absent PR number slot * improve(ui): give Create PR a neutral chip surface * improve(ui): rebalance merged PR row contrast * improve(ui): use standard squircle for PR rows * fix(ui): consolidate static PR link styling</pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(voice-call): let realtime agents end calls (#125525)</pre>

<pre style='white-space:pre-wrap;width:81ex'>refactor(agents): unify exec approval wait outcomes (#125524) * refactor(agents): unify exec approval wait outcomes * docs(plan): record approval wait cleanup</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(openai): prevent realtime barge-in truncation errors (#125502) * fix(openai): clamp realtime barge-in truncation Clamp OpenAI Realtime barge-in truncation to the audio duration produced for each assistant item. This prevents interruption failures observed live when 3,700 ms of decoded audio was paired with a 3,760 ms transport-clock timestamp. * fix(openai): propagate PCM sideband audio format The browser sideband negotiated PCM16, but its bridge

<pre style='white-space:pre-wrap;width:81ex'>test(doctor): isolate Claude CLI health checks (#122627) Amp-Thread-ID: https://ampcode.com/threads/T-019fee8d-665d-707b-a380-23f2a6a1ce03 Co-authored-by: Amp <amp@ampcode.com></pre>

<pre style='white-space:pre-wrap;width:81ex'>improve(gateway): compose live session stress probes (#122519) * test(gateway): compose live session stress probes Amp-Thread-ID: https://ampcode.com/threads/T-019feaaa-c7ed-769e-9f29-a3612bec72e7 * fix(ai): resume after Responses compaction checkpoints Amp-Thread-ID: https://ampcode.com/threads/T-019feaaa-c7ed-769e-9f29-a3612bec72e7 * test(gateway): compose multi-session subagent probes Amp-Thread-ID: https://ampcode.com/threads/T-019feaaa-c7ed-769e-

<pre style='white-space:pre-wrap;width:81ex'>chore(skills): unify issue/PR repair sweeps into openclaw-repair-sweep (#120465) Merge openclaw-small-bugfix-sweep, openclaw-landable-bug-sweep, and openclaw-autonomous-issue-sweep into one openclaw-repair-sweep skill with scope/batch_size/workers inputs and sensible defaults. Fold in the transferable doctrine from the retired personal super-orchestrator runbook: broken-main repair as standing sweep work, mandatory follow-up refactor pass, defect-clas

<pre style='white-space:pre-wrap;width:81ex'>perf(agents): reuse provider policy manifest registry (#119305)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(openai): support the current speech model snapshot (#118475) * fix(openai): support the current speech model snapshot * docs(openai): clarify speech instructions apply to the model family</pre>

<pre style='white-space:pre-wrap;width:81ex'>perf(system-agent): reuse setup metadata per activation (#118352) * perf(system-agent): reuse setup metadata per activation * chore: remove release-owned changelog entry --------- Co-authored-by: Peter Steinberger <steipete@mac-studio-sf2.local></pre>

<pre style='white-space:pre-wrap;width:81ex'>refactor: extract selected agent retirement</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): settle user turn persistence before reconciliation (#117393)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(codex): guide recovery from unimported agent auth (#116807) * fix(codex): diagnose unimported agent auth Add agent-scoped Codex migration guidance and expose a validated --agent target so non-default credentials cannot be imported into the default agent store. Fixes #116763 * fix(codex): surface auth migration before route planning * fix(codex): scope auth migration recovery * fix(codex): preserve API key auth fallback * test: repair memory import

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): avoid provider discovery when cache pruning is off (#117396)</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(tui): cancel buffered submissions on shutdown [AI-assisted] (#117331) * fix(tui): cancel buffered submissions on shutdown * refactor(tui): preserve callable submit coalescer * test(tui): run local PTY smoke against built CLI</pre>

<pre style='white-space:pre-wrap;width:81ex'>refactor(agents): drop legacy subagent state recovery (#115634) * refactor(agents): remove legacy subagent state recovery * test(agents): use canonical subagent persistence fixtures * fix(ci): repair code mode matrix checks * fix(agents): make canonical subagent state validation exact * docs(agents): clarify canonical delivery storage * docs(agents): trim redundant state helper comment * style: format compatibility checks</pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(skills): add Skill Workshop lifecycle hooks (#115606) * feat(skills): add lifecycle hook contracts * feat(plugins): expose skill hook contracts * feat(plugins): identify skill evaluators * feat(skills): persist proposal evaluation lifecycle * feat(skills): add agent evaluation action * feat(skills): emit committed skill lifecycle changes * feat(gateway): expose skill proposal evaluation lifecycle * feat(ui): add Skill Workshop evaluations * fix(s

<pre style='white-space:pre-wrap;width:81ex'>test(ui): cover imported Claude history redaction (#115683) Co-authored-by: Peter Steinberger <steipete@gmail.com></pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(claws): materialize CLAW.md prompts (#113454) * fix(claws): align prompt lifecycle validation * test(claws): cover Unicode frontmatter * feat(claws): materialize CLAW.md prompts * fix(claws): preserve portable provenance import</pre>

<pre style='white-space:pre-wrap;width:81ex'>revert(openai): remove loopback Codex proxy support (#114616) Reverting PR from runaway clanker, sorry! Won't happen again!</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(plugins): prevent bundled discovery loss and out-of-scope activation (#114547) * fix(plugins): prevent missing and out-of-scope plugin activation * fix(plugins): clear complete metadata on lifecycle invalidation * fix(plugins): clear metadata without lifecycle import cycles * test(agents): preserve plugin metadata lifecycle in model mock * test(agents): align Ollama auth with the installed plugin index * test(auto-reply): preserve plugin metadata