The Agent Skills Directory

[SAFE]: The skill instructions are focused exclusively on read-only analysis of the project workspace. It identifies framework versions, directory structures, and architectural patterns from standard configuration files like package.json and tsconfig.json.
[DATA_EXPOSURE]: The skill analyzes database schemas and integration points to document the data model and authentication methods. This behavior is restricted to the local workspace and intended for generating a context briefing for the user, with no evidence of data exfiltration or credential harvesting.
[INDIRECT_PROMPT_INJECTION]: While the skill processes external project files which could serve as a vector for indirect injection, the instructions are strictly defined for extraction and reporting. The lack of file-write or code-execution commands in the procedure significantly mitigates the risk of an automated attack triggering harmful actions.

codebase-context