The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local repository (source code and existing documentation) which could theoretically contain malicious instructions designed to manipulate the agent's behavior during wiki generation.
Ingestion points: Reads all source code files and existing documentation such as README.md and ARCHITECTURE.md during the survey and authoring phases (references/generate-and-refresh.md).
Boundary markers: Absent; the instructions do not specify the use of delimiters or specific instruction-ignore wrappers when reading or interpolating the content of these files.
Capability inventory: The skill uses shell execution (exec) for environment detection and git operations, and file-write capabilities (write, edit) via the OpenKnowledge MCP server.
Sanitization: The skill includes instructional guidance for the agent to omit secrets and internal infrastructure details for public-facing wikis, but lacks automated sanitization of ingested content.
[COMMAND_EXECUTION]: The skill utilizes shell commands to perform repository introspection and file management.
Evidence: Uses exec to run ls -A, cat, git rev-parse HEAD, and git diff --stat to detect repository state and changes (references/generate-and-refresh.md).

codebase-wiki