darkmatter-repo-setup

Pass

Audited by Gen Agent Trust Hub on Sep 27, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill clones configuration templates and shared resources from the author's GitHub repositories.
  • Fetches the canonical project template from git@github.com:darkmatter/template.git.
  • Fetches shared instruction sets from git@github.com:darkmatter/skills.git.
  • [REMOTE_CODE_EXECUTION]: The skill executes a remote installer directly from a GitHub repository using the Nix package manager.
  • Evidence: nix run github:darkmatter/skills#install in SKILL.md.
  • [COMMAND_EXECUTION]: The skill executes various local shell commands for toolchain installation, repository auditing, and validation.
  • Installs the effect-solutions global CLI using bun add -g.
  • Executes project validation commands including bun run check, bun run test, nix flake check, and docker compose config.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the repository it is auditing, which serves as an ingestion point for potentially adversarial data.
  • Ingestion points: Reads package.json, flake.nix, AGENTS.md, and other top-level repository files in SKILL.md (Step 1: Pre-flight).
  • Boundary markers: None identified; the skill instructions do not include specific delimiters or warnings to ignore instructions embedded within the target repo's files.
  • Capability inventory: The skill has significant capabilities including remote script execution (nix run), file writing/modification (SKILL.md Step 3), and network operations (git clone).
  • Sanitization: None identified; the agent is instructed to adapt content from templates and audit existing files without explicit sanitization of the input data.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 27, 2026, 05:10 AM