ci-fix-tests
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
npx codeceptjs runandgit checkoutto execute tests and manage files. There is a risk of command injection if scenario names, which are interpolated into the--grepargument, contain malicious shell characters derived from test files or traces. - [REMOTE_CODE_EXECUTION]: The agent is tasked with editing JavaScript/TypeScript test files and executing the modified code. This runtime modification and execution of scripts constitutes a dynamic code execution pattern.
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted data from
output/trace_*/artifacts and ARIA snapshots generated during test runs. If the application under test produces malicious output, it could influence the agent's repair decisions. - Ingestion points: Failed test trace artifacts and ARIA snapshots located in the
output/directory. - Boundary markers: Absent; the skill does not specify delimiters or instructions to ignore embedded commands in the traces.
- Capability inventory: Ability to write to the file system (modifying tests) and execute shell commands (
npx,git). - Sanitization: None; the skill lacks explicit validation for the contents of the traces or the scenario names before they are used in high-privilege operations.
Audit Metadata