skills/coco-research/coco/coco-ship/Gen Agent Trust Hub

coco-ship

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: Uses standard git and gh commands to manage local branch state and interact with remote repositories. The commands are structured with safety in mind, such as using heredocs with quoted delimiters to prevent unintended shell expansion of content.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads output from git diff and git log to assist in drafting commit messages and PR summaries. While this content originates from the project files, any potential risk from embedded instructions in code is mitigated by the skill's requirement for a separate, explicit user confirmation step before any code is pushed or a PR is created.
  • [DATA_EXFILTRATION]: Transmits code and metadata to the organization's repositories. This behavior is consistent with the skill's stated purpose of shipping work to vendor-managed infrastructure.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 06:11 PM