analyze-github-action-logs
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill constructs and executes shell commands using the
ghCLI andgrepwith parameters (workflow,repo,run_id,count) that are directly derived from user input or external data. If these inputs are not sanitized, they could be used to execute arbitrary shell commands via injection characters like semicolons or backticks. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and analyze GitHub Actions logs, which are untrusted external data sources. A malicious actor could inject instructions into the logs (e.g., through failing test output or printed messages) that may influence the analysis performed by subagents or the final report synthesis. \n
- Ingestion points: Raw logs retrieved via
gh run view --logand saved to/tmp/actions-run-<run_id>.log. \n - Boundary markers: The instructions do not specify clear delimiters or directives for the subagents to ignore instructions embedded within the processed logs. \n
- Capability inventory: The skill uses the
ghCLI,grep, and the ability to spawn subagents for analysis. \n - Sanitization: There is no evidence of log content sanitization or verification before it is passed to subagents.
Audit Metadata