The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it reads and processes data from external, potentially untrusted project files.
Ingestion points: The ProjectFingerprinter scans local project directories recursively and reads content from files including package.json, pyproject.toml, go.mod, Cargo.toml, Dockerfile, Chart.yaml, and .github/workflows.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potential commands embedded within the data extracted from these files.
Capability inventory: The skill is designed to perform filesystem read operations (recursive walks and file parsing) to extract feature vectors.
Sanitization: The instructions do not specify any sanitization, validation, or filtering of the strings extracted from project configuration files before they are processed by the recommendation engine.

contextual-intelligence