The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several command-line tools through Bash to manage configuration and inspect the environment. It executes a bundled shell script templates/resolve-config.sh to handle JSON configuration merging and profile resolution. It also uses git, gh (GitHub CLI), and jq to read repository state, auth status, labels, and issue templates (SKILL.md, REFERENCE.md, resolve-config.sh). Additionally, it uses curl to fetch a JSON schema for validation.
[EXTERNAL_DOWNLOADS]: The skill fetches a necessary validation schema from the author's public GitHub repository. The source is https://raw.githubusercontent.com/TitusKirch/skills/main/tituskirch-skills.schema.json (SKILL.md). This file provides the single source of truth for configuration keys, types, and defaults.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from repository files which could contain malicious instructions.
Ingestion points: The skill reads .tituskirch-skills.json and issue templates located in .github/ISSUE_TEMPLATE/ (SKILL.md, REFERENCE.md).
Boundary markers: It implements a strict 'Plan -> Confirm -> Write' workflow, ensuring the user reviews every proposed change before any file is written or committed.
Capability inventory: The skill has the ability to write files, edit content, and execute network/CLI commands via Bash tools like git, gh, and curl (SKILL.md allowed-tools).
Sanitization: It relies on human-in-the-loop confirmation as a primary defense against unauthorized actions triggered by processed data.

tituskirch-skills-config