skills/canonical/mason/chisel-slicer/Gen Agent Trust Hub

chisel-slicer

Pass

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Several helper scripts (check-slice.py, check-diff.py, deb-list.py, review-diff.py, try-cut) utilize subprocess.run or shell execution to interact with system tools including git, dpkg-deb, and chisel. This behavior is central to the skill's purpose of managing and reviewing repository content. The scripts use list-based arguments for subprocess calls, which mitigates standard shell injection risks.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from external sources, specifically Slice Definition Files (SDFs) in YAML format, git PR diffs, and metadata from Debian packages. Maliciously crafted SDFs could attempt to influence agent behavior during review.
  • Ingestion points: check-slice.py reads SDF files; deb-list.py processes package metadata; review-diff.py processes git diffs.
  • Boundary markers: The skill uses structured YAML parsing for most ingestion.
  • Capability inventory: The agent has the capability to write files to the local repository and execute local scripts.
  • Sanitization: check-slice.py uses a custom _StrictLoader inheriting from yaml.SafeLoader to prevent unsafe object construction during YAML parsing. The instructions also emphasize using agent judgment for review tasks.
  • [EXTERNAL_DOWNLOADS]: The deb-list.py script fetches .deb packages from established Ubuntu mirrors (archive.ubuntu.com and ports.ubuntu.com) to inspect their contents. The orientation script fetches release information from GitHub's infrastructure (github.com and raw.githubusercontent.com). These operations target well-known and trusted infrastructure for the purpose of package analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 21, 2026, 09:01 AM
Security Audit — agent-trust-hub — chisel-slicer