skills/tacoda/tack/tack-characterize/Gen Agent Trust Hub

tack-characterize

Pass

Audited by Gen Agent Trust Hub on Nov 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill guides the agent to perform extensive local command execution to establish a project baseline. This includes running installation commands (npm ci, pip install), build processes (go build, cargo build), and test suites (pytest, rspec, cargo test) as detailed in the recipes/ directory and SKILL.md.
  • [EXTERNAL_DOWNLOADS]: The skill's recipes recommend fetching standard testing and mocking utilities from well-known package registries and repositories. Examples include the go-cmp module from Google's GitHub repository, as well as various packages for JavaScript, Python, Ruby, and Rust (e.g., vitest, syrupy, VCR, insta).
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze potentially untrusted third-party codebases, which presents an attack surface for indirect prompt injection.
  • Ingestion points: The agent is instructed to read various files within a repository, such as dependency manifests (package.json, go.mod), documentation, and existing agent configuration files (CLAUDE.md, .cursorrules) as specified in SKILL.md and references/characterize.md.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the analyzed files are provided in the instructions.
  • Capability inventory: The skill utilizes significant capabilities, including the ability to execute arbitrary shell commands and create/commit files (the characterization report) as described across all files.
  • Sanitization: The instructions do not prescribe any sanitization or validation of the data read from the codebase before it is processed or used to generate reports.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 22, 2026, 01:30 AM