conflux-rust-integration-test
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- Privilege Escalation (HIGH): In the 'Pitfalls' section, the documentation explicitly instructs the agent or user to 'rerun with escalated permissions' to resolve an 'Operation not permitted' error in the
~/.cache/uvdirectory. This encourages the use ofsudoor administrative privileges for routine cache management, posing a risk of system-wide compromise or unauthorized file modification. - Command Execution (HIGH): The skill provides multiple workflows for executing shell commands via
uv run --no-sync pytest. This establishes a direct pathway for executing arbitrary Python code on the local system. - Indirect Prompt Injection Surface (HIGH): (Category 8) The skill's primary function involves creating and modifying test modules (
integration_tests/tests/...) and then executing them. This creates a high-capability attack surface where malicious instructions embedded in external PR descriptions or test data could be transformed into executable code by the agent and subsequently run. - Ingestion points: Reads and writes to
integration_tests/tests/. - Boundary markers: None identified in the provided instructions.
- Capability inventory: Subprocess execution via
uv runand file write operations. - Sanitization: None specified for test content generation.
Recommendations
- AI detected serious security threats
Audit Metadata