concurrent-branches
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No evidence of malicious behavior, data exfiltration, or credential theft was found. The instructions align with legitimate software development workflows.
- [COMMAND_EXECUTION]: The skill suggests using standard command-line utilities including git, jq, grep, and make to manage repository state and validate merges. These tools are used appropriately for the skill's stated purpose of assisting in conflict resolution.
- [INDIRECT_PROMPT_INJECTION]: The skill describes processes that ingest and act upon data from external repositories.
- Ingestion points: Git metadata, commit logs, and project configuration files (manifest.json, package.json, pyproject.toml) referenced in SKILL.md.
- Boundary markers: Absent.
- Capability inventory: Local command execution via git, jq, grep, and make as documented in SKILL.md.
- Sanitization: Absent.
- Note: These surfaces are standard for developer-oriented tools and do not represent an unusual security risk in this context.
Audit Metadata