engineering-wisdom
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its requirement to ingest untrusted repository content while maintaining file-modification capabilities.
- Ingestion points: The agent is directed to read various repository files including
AGENTS.md,docs/WORKFLOW.md,references/heuristics.md,references/sources.md, and any other code or documentation relevant to a review request (SKILL.md). - Boundary markers: The instructions do not define clear delimiters or provide directives to ignore potential instructions embedded within the ingested repository data.
- Capability inventory: The skill allows the agent to perform 'authorized changes' and 'implementation' tasks, which implies the capability to write to or modify repository files (
SKILL.md). - Sanitization: No explicit sanitization, filtering, or validation of the repository content is specified before the agent processes it.
- [SAFE]: The skill implements safety best practices by requiring an explicit trigger (
$engineering-wisdom) for activation and instructing the agent to keep reviews read-only by default unless changes are specifically authorized (SKILL.md).
Audit Metadata