The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted Markdown files from a wiki vault to detect structural decay and provenance gaps. This creates an indirect prompt injection surface where malicious instructions hidden in the wiki content could attempt to influence the agent's findings, report generation, or suggested patches.
Ingestion points: The skill reads all Markdown files within the provided vault path, including wiki/**/*.md, wiki/index.md, wiki/log.md, and _meta/taxonomy.md (processed in SKILL.md steps 1-3).
Boundary markers: The skill does not explicitly instruct the agent to use boundary markers or delimiters when processing vault data, nor does it include a warning to ignore instructions found within the vault content.
Capability inventory: The agent has the capability to read files, write structured reports to _agent/reports/, append to wiki/log.md, and propose or apply patches to the vault (per the 'Write permissions' section in SKILL.md).
Sanitization: The scripts/wiki-lint-core.mjs script strips Markdown code blocks before performing its regex-based link and taxonomy analysis. This reduces the risk of the deterministic checks being influenced by code-like injection patterns, though the agent's subsequent qualitative analysis phase processes the raw content.

wiki-lint