The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill triages markdown files based on their content, headers, and status markers (e.g., "status:", "Superseded by") to determine whether to harvest knowledge, merge files, or delete them. This ingestion of untrusted data from the repository creates a surface for instructions embedded in documents to influence the agent's decision-making process.
Ingestion points: Headers of markdown files and summary documents like README.md and INDEX.md as described in references/workflow.md.
Boundary markers: None identified.
Capability inventory: The skill has the capability to write to files, delete files, and execute git commit operations as detailed in references/operations.md and references/workflow.md.
Sanitization: None identified for processing document content.
[COMMAND_EXECUTION]: The scripts/scope.mjs utility executes the git log command via node:child_process to retrieve the last commit date for files in the repository to identify orphaned documents.
[DYNAMIC_EXECUTION]: The scripts/setup-modules.mjs file utilizes dynamic import() to load functionality from the sibling lodestar-setup skill. This mechanism is used to share runtime logic and configuration across the vendor's tool suite.

lodestar-docs