The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to act on human feedback provided through external channels like chat or Linear threads, which could be exploited for indirect prompt injection.
Ingestion points: Chat messages and comments in Linear threads (references/answer-arrives.md).
Boundary markers: No specific boundary markers or instruction-ignoring delimiters are required for the agent when processing these inputs.
Capability inventory: The agent can perform administrative actions via API calls to /admin/* endpoints and update ticket statuses in Linear (references/levers.md, references/replan.md).
Sanitization: The instructions do not mandate sanitizing the input from human responses before using them to trigger administrative state changes.
[COMMAND_EXECUTION]: The skill uses project-specific and standard CLI tools to execute workflow tasks.
Evidence: The SOP involves running linearis for issue management and bun for type-checking routines (references/replan.md, references/validate-failing.md).
[SAFE]: Interaction with the vendor's administrative API endpoints on the catalystcloud.dev domain is part of the documented workflow for managing the system's internal state (references/levers.md).

catalyst-sop