sprint-team
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Multiple roles (Researcher, Coder, Validator, Integrator) are designed to execute shell commands provided via template variables like {{ci_commands}} and {{coverage_command}}, or through standard development tools such as git, gh (GitHub CLI), and prek. This grants the agents significant control over the local execution environment as part of their intended roles.- [DATA_EXFILTRATION]: The skill uses the GitHub CLI (gh) to view, create, and merge pull requests. This represents a capability to transfer information from the local project environment to the configured remote repository.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as agents ingest untrusted data from the codebase.
- Ingestion points: references/researcher.md reads source files via {{src_files}} and captures CI tool output; references/validator.md reads PR diffs via gh pr diff.
- Boundary markers: No explicit boundary markers or instructions to ignore embedded prompts are present in the role definitions when processing external data.
- Capability inventory: Significant capabilities are available to the agents, including arbitrary shell execution via ci_commands and interaction with the git repository.
- Sanitization: No sanitization or validation of the ingested source code or tool output is specified before the data is processed by the agents.
Audit Metadata