Language: English | 繁體中文

TeaPrompt

TeaPrompt is a reflective prompt and workflow library for AI-assisted engineering work.

Core principle:

Doing the right thing > doing things right.

The repository contains:

• reflective-prompt-library/: reusable Markdown prompts for thinking, planning, implementation, review, research, risk handling, and context handoff.
• reflective-prompt-library/skills/: concise SKILL.md workflow wrappers that turn the prompt library into practical agent workflows.
• reflective-prompt-library/plans/: design notes and follow-up plans for code tooling or workflow automation.
• reflective-prompt-library/SKILL_INSTALLATION.md: install instructions for Claude Code, Codex, Cursor, Antigravity CLI / IDE, and OpenCode.
• reflective-prompt-library/METHODOLOGY_MAP.md: classification map for choosing the right strictness level.
• reflective-prompt-library/LANGUAGE_POLICY.md: language policy for English operational docs and localized prompt sources.

Quick Start

Use the prompt library directly:

reflective-prompt-library/README.md

Or copy a workflow skill into a compatible skills directory:

.claude/skills/<skill-name>/SKILL.md
~/.codex/skills/<skill-name>/SKILL.md
.agents/skills/<skill-name>/SKILL.md

For host-specific paths, see Skill Installation Guide.

For repository method boundaries, see Methodology Map. For language conventions, see Language Policy.

Recommended starting points:

• reflective-dispatch: choose the smallest useful workflow.
• reflective-brief: clarify goals, assumptions, scope, acceptance criteria, and falsifiability.
• reflective-spec-plan: write specs, usage-first docs, and task slices.
• reflective-implement: implement with tests, traceability, and verification.
• reflective-review: review code, plans, specs, and AI outputs.
• reflective-risk: gate high-risk work before execution.

Design Philosophy

TeaPrompt keeps prompts and workflows separate:

• Prompts provide nuance, judgment frames, and reusable wording.
• Skills provide repeatable execution shape.
• Plans capture future code or workflow automation without overengineering the current library.

The workflow layer intentionally uses a small number of broad, composable skills instead of one skill per prompt.

License

MIT. See LICENSE.