A collection of AI-related tools and Cursor skills.
Worked examples of the three layers that keep an AI agent on course: a context file and work brief (Data), constraints that prevent drift (Guardrails), and targeted prompts and tools (Skills). Data examples cover generic, software, business, everyday, and resume writing.
See management/README.md.
Capture product context — mission, must-haves, boundaries, principles, stack, constraints — so AI agents stay aligned. Asks 8 direct questions, one at a time, and writes a single file: docs/context.md. Say "you tell me" on any question and the agent researches it for you. The first run in a repo also wires AGENTS.md so every future agent — Cursor, Claude Code, Codex — reads docs/context.md before non-trivial work.
Install skill:
git clone --depth 1 https://github.com/hferello/ai.git /tmp/cursor-skill && bash /tmp/cursor-skill/context/install.sh && rm -rf /tmp/cursor-skillSee context/README.md for full documentation.
Feature design workflow. Asks clarifying questions first, then creates planning files in features/<task>/. Trigger with /feature.
Install:
git clone --depth 1 https://github.com/hferello/ai.git /tmp/cursor-feature-skill && bash /tmp/cursor-feature-skill/feature/install.sh && rm -rf /tmp/cursor-feature-skillSee feature/README.md for full documentation.
This repo also ships Claude Code plugins for the same two workflows:
contextplugin ->/contextfeatureplugin ->/feature
- Add this repo as a marketplace:
/plugin marketplace add https://github.com/hferello/ai
- Install either or both plugins:
/plugin install context@hferello-ai
/plugin install feature@hferello-ai
The marketplace definition lives at .claude-plugin/marketplace.json.
For plugin packaging details (generated files vs curated files), see plugins/README.md.