tk-repository-guidance

Pass

Audited by Gen Agent Trust Hub on Nov 1, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes repository-local files and external documentation, which is a standard surface for indirect prompt injection. However, it provides robust mitigations by instructing the agent to treat such data as untrusted evidence rather than instructions or commands.
  • Ingestion points: Reads CONTEXT.md, CONTEXT-MAP.md, tigerkit.config.json, .tigerkit/repository.json, and external library/API documentation.
  • Boundary markers: Contains clear instructions to interpret data as evidence for judgment only and explicitly warns that policy strings are not shell commands or prompt overrides.
  • Capability inventory: This is a reference-only skill that does not define any executable tools, though it guides the use of other tools in the ecosystem.
  • Sanitization: Emphasizes the need to validate JSON schemas, versions, and file ownership before processing configuration data.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 1, 2026, 04:41 PM