The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use various system and Temporal-specific CLI tools including temporal, tcld, openssl, nc, dig, and nslookup for diagnostics and operations. The instructions in SKILL.md define a 'Philosophy' of operator discipline that mandates human-in-the-loop approval for all destructive operations, such as namespace deletion or batch terminations.
[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests and processes data from external systems via CLI outputs.
Ingestion points: The agent is guided to read potentially untrusted data from Workflow histories and Task Queue descriptions (e.g., in references/triage/workflow-stuck.md and references/ops/workflow-health.md).
Boundary markers: No specific delimiters or boundary markers are instructed for use when interpolating this external data into the agent's context.
Capability inventory: The skill utilizes command execution for administrative tasks, including irreversible actions like deleting namespaces (tcld namespace delete) or revoking access keys (tcld apikey delete).
Sanitization: The instructions do not provide explicit filtering or sanitization logic for the data parsed from CLI outputs, relying instead on the agent's interpretation and the user's manual review of proposed actions.

temporal-ops