The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external content (articles, documents, or URLs) to create or update local wiki pages. This untrusted data is synthesized by the LLM into summaries and cross-references without explicit sanitization or boundary markers, creating a surface where malicious instructions in the source could influence future agent actions when the wiki is queried.
Ingestion points: SKILL.md instructions for 'Ingest a source' read from wiki/raw/ or user-provided paths/URLs.
Boundary markers: None found. The skill does not instruct the agent to ignore or delimit embedded instructions in the source documents.
Capability inventory: SKILL.md allows Bash, Read, Write, and Edit. Scripts list.py, search.py, lint.py, and log.py perform file system operations (read/write/directory listing) based on the wiki content.
Sanitization: None found. Content from wiki/raw/ is directly processed and written into wiki/pages/.

llm-wiki