The Agent Skills Directory

[SAFE]: The skill performs intended architectural analysis using established software engineering methodologies. It converts source code into a documentation-based knowledge base to aid AI understanding of complex systems.
[COMMAND_EXECUTION]: The skill executes local Python scripts and platform-provided CLI commands.
Evidence: It runs scripts/scan_repo.py to index files and scripts/validate_kb.py to check for documentation dead links and quality anchors.
Evidence: It utilizes the platform command teamai codebase --extract to perform AST-based extraction of code structures.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes untrusted source code and documentation.
Ingestion points: Reads files from multiple local repository paths (Go, Python, Java, TS, etc.) and product documentation (Markdown, YAML) provided by the user in Phase 0.
Boundary markers: While the skill employs Markdown templates and structured 'AI Quick Understanding' tables, it does not implement strict security delimiters or specific warnings to ignore embedded instructions within the code being analyzed.
Capability inventory: The agent uses grep, glob, read_file, and write_to_file tools, and executes the provided local Python scripts within the project environment.
Sanitization: Content is extracted and summarized by the LLM based on templates; no automated sanitization or filtering of potential injection strings in the source code is performed.

team-wiki-codebase