cc-codex-spec-bootstrap
Pass
Audited by Gen Agent Trust Hub on Mar 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the 'abcoder' utility via npm and executes 'gitnexus' through npx. While these originate from third-party sources (CloudWeGo and a community contributor), they are standard tools required for the skill's primary function of code analysis.
- [COMMAND_EXECUTION]: Orchestrates several CLI operations including npx commands, Python scripts located within the project's .trellis directory, and Codex CLI execution. It also provides instructions for configuring MCP servers for both Claude Code and Codex agents to enable tool access.
- [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection. It analyzes untrusted local repository content to automatically generate task descriptions (PRDs) which are then consumed and executed by Codex agents.
- Ingestion points: GitNexus and ABCoder analysis of the target codebase (SKILL.md, references/mcp-setup.md).
- Boundary markers: None explicitly implemented in the generated PRD templates to separate untrusted code data from agent instructions.
- Capability inventory: The parallel Codex agents have the capability to read any file in the repository and write output to the .trellis/spec directory.
- Sanitization: No explicit sanitization or escaping mechanisms are defined for code patterns extracted from the repository before they are interpolated into the agent tasks.
Audit Metadata