The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to perform a deep, automated audit of external codebases and sustainability data models without implementing sanitization or boundary delimiters for the analyzed content.
Ingestion points: The skill reads various configuration files (package.json, requirements.txt, go.mod, Gemfile, pom.xml), ORM models, database migration files, and stakeholder engagement documents to map ESG data.
Boundary markers: There are no instructions to utilize delimiters or to ignore embedded instructions within the codebase files being analyzed, which could allow malicious data to influence agent behavior.
Capability inventory: The skill has the capability to read files across the entire project directory and write execution logs to the user's home directory.
Sanitization: The instructions do not specify any validation, escaping, or filtering of the content ingested from the external codebase.
[PERSISTENCE]: The skill attempts to maintain a record of its execution history and outcomes by writing to a hidden platform-specific directory.
Evidence: The "SELF-EVOLUTION TELEMETRY" section instructs the agent to check for the existence of ~/.claude/projects/ and append metadata (outcome, iterations, bottlenecks, and suggestions) to skill-telemetry.md. This allows the skill to persist state and logs across different sessions in a hidden location.

sustainability-metrics