{
  "markdown": "# consensus-rnd\n\nEnglish canonical public identity document. 中文 companion: [README.zh-CN.md](./README.zh-CN.md).\n\n`consensus-rnd` is a cross-platform Agent Skills publication repository. Its published skill is `sshx`, a worker-delegated inline consensus contract for high-risk decisions and implementation plans.\n\nThis repository is not an application runtime. Its deliverables are skills under `skills/<name>/` plus platform manifests that expose the same skill tree to Claude Code, Codex, Cursor, and Gemini.\n\n## What It Provides\n\n| skill | What it is for | Runtime shape |\n|---|---|---|\n| `sshx` | Worker-delegated inline consensus (轻量 worker-delegated inline 共识方法论) with isolated, independent perspectives, a fixed thinking truth table, and a same-shape review gate. | Prompt-level contract; no daemon, GitHub, git, label, release, or lifecycle authority. |\n\n## Core\n\nThe engine is not \"run the same prompt several times and vote.\" It is biased, independent, multi-angle convergence:\n\n- **Biased independent perspectives**: thinking and review workers start from different priors and must not read same-round peer output before sealing their own verdicts.\n- **Meta-judge convergence**: disagreement is reduced into fixed exits: consensus, a compatible concrete plan, or a bounded stall that is surfaced honestly.\n- **Concrete plans pass the gate**: an obvious direction still needs independent validation before implementation.\n- **Symmetric verification**: finished work passes independent reviewers and the fixed review truth table before it is declared done.\n- **Pure orchestration**: the caller dispatches workers and aggregates bounded conclusions; implementation, verification, repair, and design solving remain delegated.\n\n## Quick Start\n\n### Claude Code\n\n```bash\n/plugin marketplace add ChronoAIProject/consensus-rnd\n/plugin install consensus-rnd@consensus-rnd\n```\n\nFor local repository development, you may manually create `.claude/skills -> ../skills`; it is ignored and is not a published Claude entrypoint.\n\n### Codex / Cursor\n\nPoint the platform plugin mechanism at this repository. `.codex-plugin/plugin.json` and `.cursor-plugin/plugin.json` expose skills through `\"skills\": \"./skills/\"`.\n\n### Gemini CLI\n\nInstall as an extension. `gemini-extension.json` uses `GEMINI.md` as the context entrypoint and lists the available skill.\n\n### Direct Copy\n\nCopy `skills/<name>/` into the agent's personal skills directory, such as Claude Code's `~/.claude/skills/`.\n\n## Architecture\n\n`skills/` is the shared product surface. Each skill owns its contract in `SKILL.md`; heavy references may live beside it, and mechanical behavior lives under that skill's `scripts/` tree.\n\nPlatform manifests point different agents at the same skills:\n\n```text\n.\n├── .claude-plugin/        # Claude Code: plugin.json + marketplace.json\n├── .codex-plugin/         # Codex: plugin.json\n├── .cursor-plugin/        # Cursor: plugin.json\n├── gemini-extension.json + GEMINI.md   # Gemini: extension manifest + context entry\n├── package.json           # npm-style metadata / version anchor\n├── AGENTS.md -> CLAUDE.md # Cross-agent rules, symlinked\n├── CLAUDE.md              # Agent guide for work inside this repository\n├── README.md              # English canonical public identity document\n├── README.zh-CN.md        # Chinese companion public identity document\n├── LICENSE                # MIT\n├── skills/<name>/         # Each skill; SKILL.md is required\n└── .version-bump.json     # Manifest version synchronization map\n```\n\n## License\n\n[MIT](./LICENSE) © ChronoAIProject\n",
  "bytes": 3563,
  "sha": "f29db90bc12c483c6c990131ebdf4fcc929280971d67e4a279909c6a972f6b18",
  "repo_slug": "chronoaiproject/consensus-rnd",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_chronoaiproject_consensus_rnd_2d0fe9e7/readme"
}