deliver-slices
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from potentially untrusted external sources, such as GitHub issues and pull request comments, which could contain malicious instructions.
- Ingestion points: Workflow step 1 in
SKILL.mdinvolves reading GitHub issues and PR comments to plan implementation slices. - Boundary markers: The supervisor agent creates 'slice packets' (defined in
references/slice-packet.md) to scope child agent activities. - Capability inventory: The skill utilizes git operations (commit, push, checkout) and executes local validation scripts as part of the delivery process (
SKILL.md). - Sanitization: No technical sanitization of external content is specified; the skill relies on the supervisor agent's review and briefing process.
- [COMMAND_EXECUTION]: The skill performs repository management and validation via shell commands.
- Evidence:
SKILL.mddirects the agent to rungit status,git commit,git push, and repository-specific validation scripts to confirm implementation correctness.
Audit Metadata