convez.ai
Agente
YeLuo45

superpower-clockless

YeLuo45/superpower-clockless

Ver no GitHub
InicialPluginScore · de 0 a 100

跨 Agent 安装器:为 Hermes、OpenClaw、Cursor、Claude Code、Codex 集成 ai-superpower 与提案生命周期技能

Instalação

$npx convez-ai YeLuo45/superpower-clockless

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 5 meses
Versão
—
Issues abertas
1
Contribuidores
2

README

A tradução para o português deste README ainda está na fila.

superpower-clockless

Cross-agent installer for the Superpower proposal system.

It packages two capabilities into one portable project:

  • ai-superpower: API-first project/proposal storage with audit logs, CSV locking, validation, and lifecycle transitions.
  • prj-proposals-manager: platform-agnostic proposal lifecycle skill for intake, PRD, TDD, development handoff, acceptance, deployment, and delivery.

The design follows the agentmemory pattern: one shared local service, plus thin per-agent adapters for MCP/config/skills.

Supported Agents

AgentIntegration
Hermes~/.hermes/config.yaml MCP block + skill copy
OpenClaw~/.openclaw/openclaw.json MCP block + extension skill copy
Cursor~/.cursor/mcp.json MCP block + always-on rule
Claude Code~/.claude.json MCP block + CLAUDE.md workflow note
Codex CLI~/.codex/config.toml MCP block + AGENTS.md workflow note

Installation Without Python

superpower-clockless is a Python application. If you don't have Python installed, use one of the methods below.

Option 1: Bootstrap Script (Linux/macOS)

Run the bootstrap script — it detects your OS, installs Python 3 if missing, then sets up superpower-clockless:

curl -fsSL https://raw.githubusercontent.com/YeLuo45/superpower-clockless/main/bootstrap.sh | bash

The bootstrap script:

  1. Detects OS (Linux/macOS)
  2. Installs Python 3.10+ via system package manager (apt/brew/yum)
  3. Creates a Python virtual environment at ~/.superpower-clockless/venv
  4. Installs superpower-clockless into the venv
  5. Writes ~/.superpower-clockless/env with API key export prompt

After bootstrap, add to shell profile:

# Unix
echo 'source ~/.superpower-clockless/env' >> ~/.bashrc
source ~/.bashrc

Option 2: Docker

# Pull latest image
docker pull yeluo45/superpower-clockless:latest

# Run with API key and port
docker run -d \
  --name superpower-clockless \
  -p 8000:8000 \
  -e AI_SUPERPOWER_API_KEY="<your-key>" \
  yeluo45/superpower-clockless:latest

Access ai-superpower API at http://localhost:8000. The container runs the ai-superpower server + superpower-clockless MCP bridge.

Option 3: Standalone Binary (Windows)

Download the standalone Windows executable from GitHub Releases:

# Download and run installer (no Python required)
irm https://raw.githubusercontent.com/YeLuo45/superpower-clockless/main/bootstrap.ps1 | iex

The bootstrap.ps1 script downloads and runs directly — no GitHub Releases attachment needed.


Quick Start

# Install superpower-clockless in editable mode (from local repo clone)
pip install -e .

# Set your ai-superpower API key (get it from https://github.com/YeLuo45/ai-superpower)
export AI_SUPERPOWER_API_KEY="<your-key>"

# List all supported agents and their integration status
superpower-clockless agents

# Show available MCP bridge tools without starting the stdio loop
superpower-clockless mcp-info

# Preview what an hermes install would do (no files written)
superpower-clockless explain hermes

# Dry run: show planned filesystem changes for hermes integration
superpower-clockless install hermes --dry-run

# Full install: configure hermes + start ai-superpower server on http://127.0.0.1:8000
superpower-clockless install hermes --api-url http://127.0.0.1:8000 --start-server

During install, superpower-clockless reads AI_SUPERPOWER_API_KEY or --api-key and writes an env file with the export. Unix/macOS uses ~/.superpower-clockless/env and Windows uses ~/.superpower-clockless/env.bat.

# Unix / macOS
export AI_SUPERPOWER_API_KEY="<your-key>"
:: Windows
@echo off
set "AI_SUPERPOWER_API_KEY=<your-key>"

Source this file from shell startup scripts when you want the key available in new terminal sessions.

By default, install first bootstraps a local ai-superpower scaffold at ~/.superpower-clockless/ai-superpower, then wires the selected agent to it. Use --skip-core only when ai-superpower is already installed elsewhere.

Install other hosts by changing the agent name:

superpower-clockless install cursor
superpower-clockless install claude-code
superpower-clockless install codex
superpower-clockless install openclaw

Runtime Model

superpower-clockless now bootstraps a starter ai-superpower core when needed, then wires agents to that local service.

Hermes / OpenClaw / Cursor / Claude Code / Codex
        | config + MCP + skill/rules
        v
superpower-clockless MCP bridge + adapter
        |
        v
ai-superpower REST API (default http://127.0.0.1:8000)
        |
        v
projects.csv / proposals.csv / audit.log

Repository Layout

src/superpower_clockless/
  api_client.py                # ai-superpower REST client
  core.py                      # bundled ai-superpower core bootstrap
  doctor.py                    # post-install validation checks
  explain.py                   # non-mutating install preview plans
  mcp_server.py                # minimal MCP stdio bridge
  installer.py                 # CLI installer and config merge logic
  catalog/agents.json          # supported agent matrix
  templates/skills/            # bundled prj-proposals-manager skill
  templates/ai-superpower/     # ai-superpower package metadata snapshot
  templates/agents/            # host instruction blocks

tests/
  test_api_client.py              # REST client behavior tests
  test_mcp_server.py              # MCP bridge behavior tests
  test_installer.py               # installer behavior tests

MCP Tools

superpower-clockless mcp starts a stdio JSON-RPC bridge. The bridge exposes these tools:

  • health
  • project_list, project_get
  • proposal_list, proposal_get, proposal_create
  • proposal_update_fields, proposal_update_status

Use superpower-clockless mcp-info to inspect tool names without starting the stdio loop.

Doctor

Run the post-install doctor to verify local host wiring and ai-superpower connectivity without mutating files or data:

superpower-clockless doctor --agent hermes
superpower-clockless doctor --agent all
superpower-clockless doctor --json

The doctor checks catalog metadata, host config file presence, MCP server entries, skill/rule files, and GET /health on the configured ai-superpower API URL.

Explain

Preview installer changes before writing any files:

superpower-clockless explain hermes
superpower-clockless explain all --json
superpower-clockless explain codex --start-server

The explain command reuses the install planner in dry-run mode and reports expanded config paths, skill paths, MCP server keys, API URL, and planned actions.

Safety Rules

  • All project/proposal data writes must go through ai-superpower API/CLI.
  • CSV files are data storage, not a user-editing interface.
  • Existing agent config files are merged, not replaced.
  • Default install bootstraps ai-superpower core before agent wiring; use --skip-core for adapter-only mode.
  • --dry-run shows planned filesystem changes without writing.

Windows Support

On Windows systems:

  • ~/.superpower-clockless/env.bat is written for API key exports
  • set "AI_SUPERPOWER_API_KEY=<your-key>" is the equivalent of export
  • Paths like ~/.hermes/config.yaml resolve to %USERPROFILE%\.hermes\config.yaml
  • PowerShell users: run .\.superpower-clockless\env.bat or add it to $PROFILE

Multi-Language Documentation

FileLanguage
README.mdEnglish
README-zh.mdChinese (中文)
README-de.mdGerman (Deutsch)
README-fr.mdFrench (Français)
README-ja.mdJapanese (日本語)

Development

python -m pip install -e .[dev]
python -m pytest -q
python -m superpower_clockless.cli agents
python -m superpower_clockless.cli mcp-info
python -m superpower_clockless.cli install hermes --dry-run
Ver o original no GitHub

Bom para

Relacionados