The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes source code and documentation from a user-provided path, creating an injection surface.
Ingestion points: Loads file content using the Read and Glob tools into the agent's context.
Boundary markers: None defined; the agent processes raw content without explicit delimiters.
Capability inventory: Has access to Bash, Grep, and agentdb storage tools.
Sanitization: No specific sanitization or validation logic is applied to the ingested content.
[EXTERNAL_DOWNLOADS]: Mentions the use of npx @claude-flow/cli, which downloads the tool from the NPM registry at execution time.
[COMMAND_EXECUTION]: Utilizes the Bash tool to perform file system scanning and searching operations.

kg-extract