doc-and-modernize
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to identify the repository (e.g.,
git remote -v,git log -1) and to perform a feasibility spike by running a project's own build, install, and test commands. These actions are fundamental to the skill's purpose of analyzing and modernizing codebases. - [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from the repository being analyzed, including READMEs, manifests (
package.json,go.mod), and source code. - Ingestion points: Project manifests, documentation files, CI configurations, and source code files from the local repository.
- Boundary markers: The instructions require the agent to provide citations (file and line numbers) for every claim, use a confidence assessment table (High/Inferred/Unverified), and follow a strict local-first data processing rule.
- Capability inventory: The agent can execute subprocess calls for Git operations and codebase-specific task runners (npm, make, cargo, etc.) to verify project state.
- Sanitization: The skill includes a specific directive to redact credentials and tokens from Git remote URLs before recording them in generated documents.
- [DYNAMIC_EXECUTION]: The skill dynamically identifies and executes build, test, and dependency installation commands based on the technologies detected in the repository's manifests. This risk is mitigated by the instruction to verify commands against task runners or CI configurations rather than assuming them.
Audit Metadata