The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches workspace templates and additional workflow skills from the author's public GitHub repositories (gtrabanco/agentic-workflow) using npx degit and git clone during both bootstrap and upgrade procedures.
[COMMAND_EXECUTION]: Uses the GitHub CLI (gh) to create repository labels (urgent, fix-next) and executes a local shell script (.agentic-workflow/hooks/tests/test-command-guard.sh) to verify the installation of safety hooks. All operations are gated by explicit user consent and discovery findings.
[REMOTE_CODE_EXECUTION]: Utilizes npx to download and run utility packages (degit, skills) from the author's official sources to facilitate the scaffolding process.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes local project manifest files to detect the technology stack and propose relevant configurations.
Ingestion points: Reads various project manifest files including package.json, pyproject.toml, go.mod, Cargo.toml, Gemfile, and others from the target workspace directory.
Boundary markers: No explicit boundary markers or delimiters are defined in the instructions for parsing manifest data.
Capability inventory: The skill possesses capabilities for file system writes, network downloads (npx degit, git clone), and shell command execution (gh label, bash).
Sanitization: Relies on a multi-step interview process where the agent presents findings to the user for explicit confirmation or editing, rather than acting autonomously on ingested data.

init-workspace