The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs repository lanes to read and follow 'repository's own instructions' before performing edits, creating a surface for instructions embedded in external data to influence agent behavior.
Ingestion points: Target repository instructions and planner backlog logs described in the 'Tier 2 — repo lanes' and 'Tier 1 — management' sections.
Boundary markers: No specific delimiters or safety instructions are provided to distinguish repository-supplied instructions from the skill's own operational rules.
Capability inventory: Agents possess the ability to perform file edits, push code, open pull requests, and execute verification commands (e.g., re-running stated commands).
Sanitization: The skill does not mention any validation or filtering of the instructions read from external repositories.
[COMMAND_EXECUTION]: The documentation references a suite of management commands (e.g., fleet blockers, fleet verdicts, fleet register, fleet hb) implemented in scripts/fleet.mjs. These scripts are used for operational tasks such as monitoring heartbeat, checking registry status, and verifying merge conditions.

fleet-roles