safe-skill

name	safe-skill
description	A minimal example of a Claude Code skill whose runtime grants stay inside the scope its skillsig manifest declared.
allowed-tools	["Read","Edit","Bash(git status*)","Bash(git diff*)"]

safe-skill

A toy fixture used by skillsig verify to demonstrate the happy path: every entry in allowed-tools above is covered by the embedded skillsig: block below, so the verdict comes out TRUSTED.

Embedded skillsig manifest

skillsig: v1
skill_id: skillsig-examples/safe-skill
version: 0.1.0
declares:
  model: claude-opus-4-7
  tools:
    - Read
    - Edit
    - Bash(git status*)
    - Bash(git diff*)
  fs_write:
    - "${WORKSPACE}/**"
  network_egress: []
attestation:
  sigstore_bundle: ./skillsig.bundle