The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a local Python script (scripts/repo_inventory.py) used to generate a repository inventory. The script performs standard filesystem operations (directory walking and file listing) using standard libraries without network access or privileged operations.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it is designed to ingest and analyze arbitrary content from a target repository (e.g., README files, manifests, and CI configurations).
Ingestion points: Files identified during repository discovery, such as README.md, package.json, and .github/workflows/ (referenced in SKILL.md and references/repo-analysis-playbook.md).
Boundary markers: The skill uses structured templates (REPO_CONTEXT.md and REVIEW_BRIEF.md) to delimit the agent's output.
Capability inventory: The skill utilizes local filesystem access via a Python script and standard agent file-reading tools; no network exfiltration or code execution capabilities are defined within the skill.
Sanitization: Instructions explicitly require the agent to ground all claims in repository evidence and to use templates to maintain structure.
[DATA_EXFILTRATION]: The skill templates prompt the user to document configuration and secrets (e.g., .env.example). While this is intended for architectural documentation and setup instructions, it identifies a surface where sensitive configuration details could be inadvertently included in the generated documents if not properly managed by the operator.

repo-context-builder