The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by scanning project manifests such as package.json, Cargo.toml, and pyproject.toml. This represents a potential indirect prompt injection surface where malicious content in these files could attempt to influence the agent's behavior.
Ingestion points: Project manifest files (Cargo.toml, package.json, go.mod, pyproject.toml, Gemfile, Package.swift, mix.exs) and CI configuration files.
Boundary markers: None explicitly defined in the output format instructions for .agents/stack-context.md.
Capability inventory: The skill is authorized to read local files and write to the .agents/ directory. It also mentions usage of library resolution and web search tools.
Sanitization: No explicit sanitization or filtering of the content read from manifests is mentioned before it is summarized into the stack context file.
[DATA_EXPOSURE]: The skill accesses CI configuration files and directory structures to determine project conventions. This data is used solely to generate a local context file and is not transmitted externally.

detect-stack