The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the user's local repository, including source code, configuration files, and documentation, to perform its analysis and generate reports.
Ingestion points: Local source files in src/, package/, and documentation files are cataloged and analyzed across all modules.
Capability inventory: The skill executes shell commands (rg, grep, find) and documentation generators (cargo doc, go doc, npx typedoc) on these local files.
Boundary markers: The skill does not explicitly define delimiters for untrusted content when presenting findings in the final report, which could lead to confusion between instructions and data.
Sanitization: There is no evidence of sanitization or escaping of file content before it is processed by shell commands or included in outputs.
[COMMAND_EXECUTION]: The workflow relies on executing several command-line utilities to inventory and audit the API surface.
Evidence: The skill invokes ripgrep (rg), grep, find, git status, and yq to extract information from the filesystem.
[EXTERNAL_DOWNLOADS]: The skill references external documentation and utilizes package execution tools.
Evidence: The skill uses npx typedoc to generate documentation for TypeScript projects, which may fetch the package from the NPM registry.
Evidence: The exemplar-research.md module provides links to official documentation for well-known services and libraries, including GitHub, Stripe, and standard language libraries (tokio, pandas, requests).

api-review