krt-harness-wise

Pass

Audited by Gen Agent Trust Hub on May 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill operates entirely within the scope of documentation and planning. It does not perform network operations, execute remote code, or modify system configurations. All recommended workflows are intended for agent-to-agent handoffs.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository content (code and documentation), which is an inherent surface for indirect prompt injection. It mitigates this risk through a comprehensive classification and summarization workflow (defined in references/document-classification.md and references/context-curation.md). This process requires the agent to verify evidence against the current code and tests, preventing the blind adoption of instructions found in external files.
  • [COMMAND_EXECUTION]: The skill has the capability to write files to the local system. This capability is strictly constrained by the rules in references/harness-artifacts.md, which limit file creation to markdown handoff documents within specified directories (e.g., docs/harnesses/). It contains explicit prohibitions against writing application source code, configuration files, or tests during the harness creation process.
Audit Metadata
Risk Level
SAFE
Analyzed
May 26, 2026, 08:32 PM
Security Audit — agent-trust-hub — krt-harness-wise