* fix(agents): persist embedded runner session transcripts (#77823)
Run persistCliTurnTranscript and post-turn compaction for executionTrace.runner embedded,
matching CLI turns so assistant text reaches session JSONL for webchat/Feishu-style runs.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(agents): narrow embedded transcript mirror with assistant dedupe (#77823)
Embedded runs pass embeddedAssistantGapFill so persistCliTurnTranscript skips
re-appending the user prompt Pi owns and only appends assistant text when the
transcript tail lacks equivalent visible assistant content.
Adds CLI transcript regression coverage for gap-fill dedupe.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(agents): dedupe embedded transcript gap fill by tail
* fix: persist embedded session transcripts (#77839) (thanks @neeravmakwana)
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Ayaan Zaidi <hi@obviy.us>
* feat: generalize pending-final-delivery for subagents and main session
(cherry picked from commit 677fcbfaf87c8cd6de8b5bd02099b29b7d49e916)
* feat(agents): implement Phase 2 durable final delivery for main sessions
(cherry picked from commit b4e39f0ddf6dbd3f0d3b9226df8e714ad722f751)
* fix(agents): narrow heartbeat deferral to pending final delivery
* fix(agents): clear final delivery after dispatch
* fix(agents): gate durable delivery retry capture
---------
Co-authored-by: Mert Basar <MertBasar0@users.noreply.github.com>
Reduce WebUI/Gateway latency churn by avoiding redundant session reloads, carrying session keys through transcript update events, and deferring explicit media provider discovery. Includes changelog attribution and closes the referenced runtime latency issues.
Fixes#75452.
Heartbeat runs can use a per-turn model override via agents.defaults.heartbeat.model. Before this change, the run metadata was written back to the shared session store, so the next normal turn could inherit the heartbeat provider/model and a smaller context window.
This lands the contributor fix plus maintainer polish:
- preserve existing session runtime model/provider/context metadata when persisting heartbeat turns
- avoid creating invalid provider/model pairs for legacy model-only session entries
- leave empty prior runtime state unset for heartbeat-only turns
- keep normal non-heartbeat runtime persistence unchanged
- add focused regression coverage for the session-store edge cases
- refresh heartbeat docs and changelog attribution
Validation:
- pnpm test src/agents/command/session-store.test.ts src/agents/openclaw-tools.session-status.test.ts
- pnpm exec oxfmt --check --threads=1 src/agents/agent-command.ts src/agents/command/session-store.ts src/agents/command/session-store.test.ts CHANGELOG.md docs/gateway/heartbeat.md
- git diff --check
- GitHub checks on 42a00dcf38: clean; no active checks and no relevant failures
Duplicate PR #75567 was already closed; #75557 is the canonical fix.
The required-typed param introduced in 9987e7797f broke
attempt-execution.cli.test.ts and auth-profile-runtime-contract.test.ts
which construct runAgentAttempt params without an originalProvider field.
Make it optional and explicitly require the typeof check before passing
to isClaudeCliProvider so a missing field correctly skips the seed
(defensive default for fallback paths that didn't plumb the original
provider through, no-op for non-fallback paths).
Addresses review on #72069:
- Codex P1 ("Gate Claude prelude seeding by source provider"): the
guard checked the *current* fallback candidate but not the failed
attempt. A session that still carried a stale
cliSessionBindings["claude-cli"] from an unrelated past run would
inject Claude transcript context into a fallback chain that started
on a different provider (e.g. openai -> openai-codex), leaking
irrelevant prior conversation. Plumb `originalProvider` (the
user-requested provider for the chain) through to runAgentAttempt
and require `isClaudeCliProvider(originalProvider)` before reading
Claude history.
- Codex P2 ("Prefer latest compact boundary when summary is missing"):
the resolver always preferred the most recent explicit summary, so
a later compaction without its own summary entry (rare crash case)
paired stale summary text with post-latest-boundary turns. Restructure
readClaudeCliFallbackSeed to queue summaries into pendingSummary and
flush each boundary's pair atomically. A boundary with no preceding
summary now correctly falls back to the boundary's own content
rather than serving an older summary alongside fresh turns.
- Greptile P2 (newest-first break vs sparse coverage): the
formatFallbackTurns walk intentionally stops on the first oversized
turn so the prelude stays a contiguous "what was happening just
before the failure" window. Document the design choice inline so a
future maintainer doesn't reflexively change it to skip-and-continue.
Tests:
- New gateway cases for the boundary-without-summary edge case and
for trailing summaries written without a paired boundary.
- existing 33 attempt-execution + 14 cli-session-history tests still
pass; broader src/agents/command suite stays green (63/63).
Local default oxlint did not run --type-aware so the warning was missed
on the initial commit; CI surfaced it via check-lint. Hoist the heading
into a named const so its length is read directly without the assertion.
When a claude-cli attempt failed with a fallbackable error (e.g. a 402
billing limit), the next candidate -- typically a non-CLI provider --
ran with no prior conversation context. Claude Code keeps its own
JSONL session under ~/.claude/projects/, but the fallback runner only
sees what OpenClaw assembles from its own transcript, which is empty
for claude-cli sessions. The fallback model therefore behaved as if
the conversation just started, even though Claude later resumed fine.
Resolution mirrors what Claude Code itself does on resume after
compaction: prefer the explicit `/compact` summary, then append the
most recent post-boundary turns up to a char budget. Concretely:
- `readClaudeCliFallbackSeed` (gateway): walks the Claude JSONL with
awareness of `type: "summary"` and `type: "system",
subtype: "compact_boundary"` entries. Pre-boundary turns are dropped
(they are represented by the summary); post-boundary turns become
the recent-window. Multiple compactions are handled by preferring
the latest summary. Path safety reuses the existing
`resolveClaudeCliSessionFilePath` validation.
- `formatClaudeCliFallbackPrelude` / `buildClaudeCliFallbackContext\
Prelude` (agents helpers): format the harvested seed into a labeled
prelude. Tool blocks are coalesced to compact "(tool call: name)" /
"(tool result: …)" hints to keep the prompt budget honest. Newest
turns are kept first when truncating; the summary is clearly
labeled "(truncated)" if it overflows.
- `resolveFallbackRetryPrompt`: gains an optional
`priorContextPrelude` that prepends before the existing retry
marker. Empty/whitespace preludes are ignored; first-attempt prompts
are unchanged.
- `runAgentAttempt`: builds the prelude when `isFallbackRetry === true`
AND the new candidate is non-claude-cli AND a Claude-cli session
binding is present. Same-provider fallbacks (claude-cli to
claude-cli) are unaffected because Claude's own --resume still works.
Verified the new tests (12 in cli-session-history, 12 added to
attempt-execution) catch the regression: removing the prelude prepend
in resolveFallbackRetryPrompt makes both new prelude cases fail,
restoring the original cold-start behavior.
References:
- https://code.claude.com/docs/en/how-claude-code-works
- "Inside Claude Code: The Session File Format"
https://databunny.medium.com/inside-claude-code-the-session-file-format-and-how-to-inspect-it-b9998e66d56b
Clean up local Claude stdio one-shot runs before returning from embedded `openclaw agent --local`, including bundle MCP loopback teardown for local process resources.
Keeps gateway-owned MCP loopback cleanup internal to the Gateway, documents the local-vs-gateway behavior, and aligns the stale OpenAI provider-runtime fixture with the current unsupported Codex mini route.
