The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of an external package dos-kernel from the PyPI registry.
Evidence: The setup instructions specify python -m pip install 'dos-kernel==<reviewed-version>' in the How It Works section of SKILL.md.
[COMMAND_EXECUTION]: The skill relies on executing shell commands to perform its core functions.
Evidence: Examples in SKILL.md show the use of dos commit-audit --workspace . HEAD --json and dos verify --workspace . PLAN PHASE --json --no-ci.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from git repositories.
Ingestion points: The dos tool reads commit subjects, bodies, and file diffs from the local git history and working tree as described in SKILL.md.
Boundary markers: There are no documented boundary markers or instructions provided to the agent to treat the git-sourced data as untrusted or to ignore embedded instructions within commit messages.
Capability inventory: The skill possesses the capability to install software via pip and execute shell commands (dos CLI) as seen in SKILL.md.
Sanitization: The skill does not describe any sanitization or escaping of the git content before it is processed by the dos tool or interpreted by the agent.

dos-verify-done-claims