The Agent Skills Directory

[SAFE]: The skill performs read-only analysis of local project files and metadata using standard command-line tools like ripgrep, tokei, and wc. It does not attempt to access sensitive system files, hardcode credentials, or establish network connections.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of specifications, code modules, and ADRs (Architecture Decision Records) to generate its coverage map. This creates a surface for indirect prompt injection where malicious content within those files could potentially influence the resulting assessment report.
Ingestion points: File content harvested via ripgrep (rg), output from the doctrine CLI suite, and directory listings.
Boundary markers: The instructions do not specify the use of delimiters or specific "ignore" instructions for the content being read.
Capability inventory: The skill's capabilities are limited to reading local files and executing a specific set of analysis utilities to produce a markdown artifact; it has no network or file-writing capabilities beyond creating the report.
Sanitization: There is no mention of sanitizing or escaping the content read from external files before it is processed by the agent.

spec-coverage-assessment