The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from git diffs which could contain malicious instructions designed to influence the agent's behavior during the simplification or validation phases.
Ingestion points: The skill gathers code changes using various git diff commands in Phase 0.
Capability inventory: The skill has the authority to modify files directly in Phase 2 and execute arbitrary validation commands (tests, lint, build) in the Validate phase.
Boundary markers: There are no explicit instructions or delimiters defined to separate the untrusted diff content from the agent's core instructions.
Sanitization: No sanitization or filtering of the ingested diff data is mentioned.
[COMMAND_EXECUTION]: The skill utilizes shell commands to interact with the repository and environment.
Evidence: Phase 0 uses git diff variations. The Validate phase executes tests, lint, typecheck, and build tools. These are standard operations for a developer tool but represent an execution surface triggered by the skill.

simplify