@zen-agentchecks…zen/docs-thermos-refresh-e403e1 → master7 files · +81 −10review: @rohanprabhureview: @sarthakreview: @acsrujanreview: @lingalarahul7review: @anshugarg15updated 3w agoRoll the apps/thermos .md set forward for the eight changes that landed since the last docs PR (#9919):
docs/timer-shards/ARCHITECTURE.md Metrics section now documents the four new pipeline metrics (timer_shards.client.timers_started, shard.timers_dispatched, shard.fire_lag_seconds, shard.dispatch.batch_size), including their emit-site (StartTimers for client, DispatchBatch for shard), tagging (timer_type:<int>), and the WriteUtilization (bool, error) → error refactor. The workers/WORKERS.md timer_shards/ row points at the same set with links into ARCHITECTURE.md.AGENTS.md gets a new Temporal SDK Metrics subsection covering the Tally → DogStatsD handler, the doubled-segment temporal.temporal_<sdk_name> naming rationale, the migration history from the OTLP-based path, and the reuse of utils.ProvideStatsdClient. Root CLAUDE.md DataDog bullet cross-links to it so the SDK-metrics specifics live in one place.workers/WORKERS.md subsection under Polling Triggers Temporal Workflow documents the unified PollingFailureDisableThreshold = 300 semantic across both FetchLatestAuthFromApollo and the new HandleToolErrorTriggerRuns activity, the shared disableAtThresholdAndPersist flow, and the Apollo /disable reason (polling_failure_in_composio_infra). workers/activity_timeouts.md gains a row for the new 120s handletoolerrortriggerruns activity. ent/CLAUDE.md updates the consecutive_failure_count entry from "planned" to "wired" with the threshold semantics. Root CLAUDE.md worker bullets call out the new path + the shared disableTriggersInApollo chunking helper.?version=latest / toolkit_versions[<tk>]=latest resolution (PRs #10006, #10011) — handlers/CLAUDE.md gets a new Version resolution on list + retrieve endpoints subsection documenting the shared resolution rules: NormalizeRequestedVersion, PickListEntityVersion, ResolveToolkitDefaultVersion, ResolveEffectiveVersion, and HasLatestToolkitVersionOverride. The pre-#10006 behaviour and the latest:base out-of-scope caveat are spelled out so the next reader can match the helpers back to the observed precedence chain.Docs-only change. Sanity checks on the surrounding tree:
cd apps/thermos && go build ./... — clean.cd apps/thermos && go vet ./... — clean.apps/thermos/workers/timer_shards/client.go → emitClientTimersStarted (per-chunk, post-ack, nil-safe per backend).apps/thermos/workers/timer_shards/shardworkflow/metrics.go → emitDispatchMetrics (per-batch, success-only, both StatsD + OTel).apps/thermos/workers/timer_shards/shardworkflow/workflow.go → WriteUtilization now returns error not (bool, error).apps/thermos/lib/temporal/metrics_handler.go → ProvideTemporalMetricsHandler, Tally Prefix: "temporal", Separator: ".", DogstatsdReporter reuses utils.ProvideStatsdClient.apps/thermos/workers/batched_polling_triggers.go → PollingFailureDisableThreshold = 300, disableAtThresholdAndPersist, HandleToolErrorTriggerRuns, BuildNewPollingRunAfterToolError, 120s handleToolErrorOptions.StartToCloseTimeout.apps/thermos/workers/batched_polling_trigger_cleaner.go → disableTriggersInApollo with apolloDisableMaxBatchSize = 500.apps/thermos/lib/versioning/utils.go → NormalizeRequestedVersion, PickListEntityVersion.apps/thermos/ent/registrydb/utils/toolkit.go → HasLatestToolkitVersionOverride, ResolveEffectiveVersion, ResolveToolkitDefaultVersion.apps/thermos/handlers/tools/list.go + apps/thermos/handlers/triggers/list.go use PickListEntityVersion; lib/tools/retrieve.go + handlers/triggers/retrieve.go use NormalizeRequestedVersion + ResolveEffectiveVersion..md files in apps/thermos/ enumerated and reviewed; only the 7 files touched above had real drift since #9919.Origin: cron-a3c59f02d29a / zen-cron-68b880dc9f00 Triggered by: rahul.lingala@composio.dev | Source: unknown Session: https://zen-api-production-4c98.up.railway.app/dashboard/#/chat/zen-cron-68b880dc9f00
Based on git blame analysis of 7 file(s):
| Contributor | Contribution | Files |
|---|---|---|
| Sarthak Agrawal |
| 40% |
| 5 |
| Zen | 28% | 7 |
| abir-taheer | 24% | 2 |
| Rohan Prabhu | 5% | 2 |
| lingalarahul7 | 2% | 4 |
Suggested reviewers: Zen — edited several of the changed files today (CLAUDE, workers, activity_timeouts) and can quickly validate recent doc/handler updates. Sarthak Agrawal — the largest recent contributor across ARCHITECTURE, AGENTS and other thermos docs, good for a deeper architectural/consistency review.
🤖 Based on git blame with recency weighting (recent edits count more).
:white_check_mark: All modified and coverable lines are covered by tests.
| Flag | Coverage Δ | |
|---|---|---|
| e2e-tests | ? | |
| self-hosted-tests | 5.60% <ø> (ø) | |
| thermos-service-isolation-tests | 65.62% <ø> (?) | |
| thermos-unit-tests | 7.30% <ø> (ø) | |
| unit-tests | ? |
Flags with carried forward coverage won't be shown. Click here to find out more. see 831 files with indirect coverage changes
