validate-bug
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute compiler commands and to "rebuild and rerun the unchanged runtime PoC" provided by external reproduction steps. It also directs the agent to execute shell commands using
uv run --no-project <script> manifest <validate-step-dir>and to run a local script../../scripts/records.py. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: Untrusted data enters the context through
finding.jsonand the "PoC package" provided by the reproduction phase (referenced in SKILL.md). - Boundary markers: Absent. There are no instructions to use delimiters or warnings to ignore instructions embedded within the findings or PoC data.
- Capability inventory: The skill has the ability to execute subprocess calls (compiler/runtime commands), run Python scripts via
uv, and perform file system operations to save receipts and manifest directories. - Sanitization: Absent. The instructions explicitly state to rerun the "unchanged" PoC, meaning no filtering or sanitization of potentially malicious code or instructions is performed.
Audit Metadata