hai-debug
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, including software logs, error messages, and codebase contents. This creates a surface for indirect prompt injection where malicious instructions hidden in logs or comments could attempt to influence the agent's diagnostic or repair actions.
- Ingestion points: The workflow requires reading available logs, errors, code, recent changes, and existing tests as specified in
SKILL.md(Workflow Step 1). - Boundary markers: There are no specified delimiters or explicit instructions to treat external data as untrusted or to ignore instructions embedded within the logs or code.
- Capability inventory: The skill allows for the execution of reproduction commands, instrumentation of code, and modification of files (file-write) if a fix is requested.
- Sanitization: The instructions do not define any sanitization, filtering, or validation steps for the external content before it is processed or used to generate fixes.
Audit Metadata