graincrawl

Pass

Audited by Gen Agent Trust Hub on Jul 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs shell operations to build, test, and release the graincrawl CLI. This includes using go build, go test, make smoke, and goreleaser. It also installs a binary to ~/.local/bin/graincrawl and pushes tags via git.
  • [EXTERNAL_DOWNLOADS]: It manages external software dependencies and installations. This includes downloading Go modules via go mod tidy and installing the tool from the author's Homebrew tap (vincentkoc/tap/graincrawl), which is a vendor-owned resource.
  • [PROMPT_INJECTION]: An indirect prompt injection surface is present as the skill processes external repository documents and user notes. 1. Ingestion points: Reads repository files (AGENTS.md, SPEC.md) and Granola notes/transcripts. 2. Boundary markers: Instructions explicitly call for preserving read-only boundaries and avoiding unauthorized Keychain access. 3. Capability inventory: Ability to execute shell commands and modify local files. 4. Sanitization: No specific content sanitization for processed files is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 29, 2026, 02:50 AM
Security Audit — agent-trust-hub — graincrawl