The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads project-specific documentation files and is explicitly instructed to follow the rules and instructions found within them. Maliciously crafted instructions in these files could attempt to influence the agent's behavior.
Ingestion points: Reading AGENTS.md, CLAUDE.md, GEMINI.md, and COPILOT.md in the project root.
Boundary markers: Absent; instructions are incorporated directly into the session context.
Capability inventory: Executes shell commands (git status, git log, bd).
Sanitization: Absent; the skill directs the agent to follow found instructions without validation.
[COMMAND_EXECUTION]: The skill executes several shell commands to gather project state and issue tracker information.
Evidence: Execution of git status, git log --oneline --date=short --format="%h %ad %s" -15, bd ready, and bd list --status=in_progress in the Gather State workflow.

onboarding