A skill for operating and diagnosing Temporal environments — namespace administration, capacity management, IAM, certificate rotation, workflow health queries, batch operations, plus bottom-up diagnosis of stuck workflows, non-determinism errors, connectivity problems, certificate expirations, worker health issues, and rate limits.
Applies to both Temporal Cloud (tcld commands) and self-hosted (temporal operator commands) deployments. Data-plane operations (temporal workflow, temporal batch, temporal schedule) work on both.
Warning
This Skill is currently in Public Preview, and will continue to evolve and improve.
It runs temporal and tcld commands against whatever environment your CLIs are authenticated to, and some of those commands have no undo. Before pointing an agent at production, scope the credential you give it — see AGENT-PERMISSIONS.md.
We would love to hear your feedback - positive or negative - over in the Community Slack, in the #topic-ai channel. Bug reports and corrections are also welcome as GitHub issues.
npx skills add temporalio/skill-temporal-ops- Follow prompts
mkdir -p ~/.claude/skills && git clone https://github.com/temporalio/skill-temporal-ops ~/.claude/skills/temporal-ops
Appropriately adjust the installation directory based on your coding agent.
This skill runs temporal and tcld commands against whatever environment your CLIs are authenticated to, and some of those commands have no undo. The skill tells the agent to establish a blast radius and get your approval before running anything in that category — but that is a behavior, not a boundary.
Since you have to set up CLI authentication anyway, it is worth deciding which credential the agent gets: a scoped-down or read-only Service Account limits what any mutating command can reach, no matter what the agent decides to run. AGENT-PERMISSIONS.md covers that, plus command denylists and pre-execution hooks, with examples for Claude Code.
- Cloud namespace admin — create, get, list, delete namespaces; add/remove regions; failover; HA config; retention, tags, codec-server, connectivity rules, search attributes, accepted-client-ca, certificate filters, export
- Cloud capacity — On-Demand vs Provisioned modes, APS/RPS/OPS, TRUs, capacity updates, throttling, APS management
- Cloud IAM — API key lifecycle, user management, user groups, service accounts, roles, namespace permissions
- Cloud certificates — mTLS cert generation, CA upload, certificate filters, rotation, auth method switching
- Cloud export & connectivity — Workflow History Export (S3/GCS), PrivateLink/PSC, connectivity rules, Cloud Ops API
- Self-hosted admin — cluster health, namespace CRUD, search attributes, Nexus endpoints via
temporal operator - Workflow health queries — List Filter queries to find stuck/hung workflows, task-queue poller status, workflow counts
- Batch & lifecycle — cancel, terminate, reset workflows; batch operations; schedules; external activity completion
- Stuck workflows — Event-History-driven triage that identifies the last event, classifies the cause, and proposes a fix
- Non-determinism errors — detection, local replay reproduction, and fix patterns
- Connectivity failures — can't connect, TLS handshake errors, endpoint mismatches, DNS issues
- Certificate problems — x509 errors, expiry, chain verification, openssl diagnostics
- Authentication failures —
UNAUTHENTICATEDvsPERMISSION_DENIED, mTLS identity mapping; Cloud addstcld login, API key expiry, and Cloud role/namespace-permission mismatches - Worker health — no pollers, task-queue backlog, worker registration, build-id mismatches
- Rate limiting —
RESOURCE_EXHAUSTED, Cloud APS/RPS/OPS limits, self-hostedfrontend.rps/frontend.namespaceRPSdynamic config, retry-after patterns - HA failover (Temporal Cloud) — multi-region endpoint, DNS staleness, failover verification
- Ambiguous runtime errors —
context deadline exceeded,workflow is busy - Replay debugging — using the VS Code debugger extension to step through workflow history
- Writing workflows/activities — use skill-temporal-developer
- Worker performance tuning, scaling, capacity planning — use skill-temporal-workertuning
- SDK-specific ergonomics — use skill-temporal-developer