wiki-conventions

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides structural conventions for a wiki-like organization of markdown files. It defines data formats, directory layouts, and logging procedures without executing dangerous commands or making network requests. Accessing a local cache at ~/.claude/wiki.local.md is for session persistence and follows standard tool behavior.
  • [SAFE]: The skill defines a workflow that ingests untrusted data from a raw/ directory, creating a potential surface for indirect prompt injection.
  • Ingestion points: Content from the raw/ directory and user-defined guidance in CLAUDE.md (if present) are ingested into the agent context in SKILL.md.
  • Boundary markers: The conventions do not specify delimiters or warnings to isolate ingested source content from the agent's instructions.
  • Capability inventory: The skill describes the creation and modification of markdown files in the wiki/ directory and root index/log files across multiple operations.
  • Sanitization: No sanitization or validation of the content of raw source files is defined within the provided conventions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 12:10 PM
Security Audit — agent-trust-hub — wiki-conventions