coding-agent-loops
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the
--dangerously-skip-permissionsflag for the Claude Code CLI to enable autonomous execution of tasks. This configuration allows the agent to execute shell commands and modify the file system without seeking user approval, which is intended for unattended operations but bypasses essential security controls.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by directing the agent to ingest and act upon data from the local repository environment.\n - Ingestion points: The agent is instructed to read
git log,git status, and project files likeSESSION_BRIEF.md(found inSKILL.md) to determine its next actions.\n - Boundary markers: There are no markers or explicit instructions to ignore potentially malicious content within the ingested files or git history.\n
- Capability inventory: The agent has the ability to run shell commands and modify files, which are executed without confirmation due to the provided CLI flags.\n
- Sanitization: The skill does not implement any validation or sanitization of the data retrieved from the repository before it is processed as context for the agent.
Audit Metadata