The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection, as it ingests and processes text directly from code comments in the repository. Maliciously crafted comments could attempt to influence the agent's reporting behavior or output format.
Ingestion points: Local repository files scanned via grep as defined in SKILL.md.
Boundary markers: Absent; the skill does not use specific delimiters for extracted comment content to separate data from instructions.
Capability inventory: Local file reading, shell command execution (grep, git blame), and optional file writing for the ledger.
Sanitization: Absent; comment content is used directly for the final report output.
[COMMAND_EXECUTION]: The skill utilizes standard local utilities (grep and git) to perform codebase analysis. These operations are restricted to the local project environment and the current user's execution context.

ponytail-debt