wiki-ingest

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted user data from an Obsidian vault, which presents a risk of indirect prompt injection where malicious instructions in a note could influence the agent's behavior during distillation.
  • Ingestion points: Source notes are read from the vault in SKILL.md (Step 4a) to be synthesized into wiki pages.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the source content as untrusted data.
  • Capability inventory: The skill has the capability to read and write files in the local vault and execute bundled shell and Python scripts.
  • Sanitization: The skill relies on the LLM's synthesis capabilities and lacks explicit sanitization of input content.
  • [COMMAND_EXECUTION]: The skill invokes local scripts for core functionality, specifically scan-vault.sh for file discovery and select-batch.py for sorting and capping ingest batches. These are bundled POSIX-compliant shell and Python scripts using standard libraries.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 05:05 PM
Security Audit — agent-trust-hub — wiki-ingest