The Agent Skills Directory

[SAFE]: The skill is designed for architectural documentation management. It defines a clear process for writing records to docs/adr/ and updating an index in docs/adr/README.md. No malicious code, unauthorized network access, or credential exposure was detected.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it reads and processes existing documentation to inform its decisions. However, this is inherent to its primary purpose and assessed as safe.
Ingestion points: The skill reads existing files from the docs/adr/ directory during the 'Consult' and 'Supersede' modes to recover context.
Boundary markers: None present; the skill treats the existing document content as trusted architectural context.
Capability inventory: The skill is limited to reading from and writing to the docs/adr/ directory and updating the docs/adr/README.md file.
Sanitization: No explicit sanitization or validation of the content read from documentation files is performed.

decision-record