← 返回
未分类 中文

Projitive

Projitive is an MCP-first governance skill for agent-driven delivery. Use this before changing task states or writing governance artifacts. Core flow: taskNe...
Projitive 是一种以 MCP 为先的治理技能,用于代理驱动的交付。在更改任务状态或编写治理文档之前使用。核心流程:任务 Ne...
yinxulai yinxulai 来源
未分类 clawhub v2.1.3 1 版本 100000 Key: 无需
★ 0
Stars
📥 285
下载
💾 0
安装
1
版本
#latest

概述

MCP Setup (Default Required)

Install (always use latest):

npm install -g @projitive/mcp@latest

> @projitive/mcp must be kept up to date — outdated versions may be missing tools or contain bugs.

Key methods:

  • Discovery: projectLocate / projectScan / projectNext
  • Context: projectContext / taskList / taskContext / roadmapContext
  • Execution: taskNext
  • Planning: taskCreate / roadmapCreate

Init: Repo Governance Setup

When .projitive/ does not exist or is incomplete, call projectInit(projectPath="") immediately. Do NOT ask the user to do this manually.

Mandatory Governance Files

  • README.md — scope, glossary, required reading
  • roadmap.md — stage goals and milestones
  • tasks.md — task pool with status + evidence
  • designs/ — design decisions and rationale
  • reports/ — execution reports and evidence
  • templates/ — per-tool response templates

Execution Loop

  1. taskNext — pick actionable task
  2. taskContext — get evidence and hints
  3. Execute; update designs/ and reports/
  4. taskContext — verify alignment
  5. Repeat

If taskNext returns empty: projectContextroadmapContexttaskCreate 1–3 focused TODOs (+ roadmapCreate if needed) → taskNext.

Task Status Rules

  • TODOIN_PROGRESS: execution started, scope clear
  • IN_PROGRESSDONE: evidence exists, criteria met
  • IN_PROGRESSBLOCKED: external dependency blocks progress
  • BLOCKEDTODO/IN_PROGRESS: after blocker resolution is documented

Never mark DONE without evidence references.

Evidence Rules

  • Design rationale → .projitive/designs/
  • Execution outcome → .projitive/reports/
  • Task updates must reference exact evidence files

Always-on Rules

  • Resolve governance root before any task decisions
  • Governance state writes MUST go through MCP tools; never directly edit tasks.md/roadmap.md
  • Keep status transitions explicit and evidence-linked
  • Preserve ID stability in roadmap/tasks/designs/reports
  • Prefer smallest valid step that moves one task forward

Fallback (no MCP)

  1. Find nearest .projitive/
  2. Read README.md + roadmap.md + tasks.md
  3. Execute one narrow task
  4. Write evidence to designs/ or reports/
  5. Update task status in tasks.md

版本历史

共 1 个版本

  • v2.1.3 当前
    2026-05-07 12:35 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

ai-agent

Find Skills

root
帮助用户发现和安装智能体技能,当用户询问如「如何做X」、「找X的技能」、「有能做...的吗」等问题时
★ 1,517 📥 573,360
ai-agent

Self-Improving + Proactive Agent

ivangdavila
自我反思+自我批评+自我学习+自组织记忆。智能体评估自身工作、发现错误并持续改进。
★ 1,441 📥 328,337
ai-agent

Agent Browser

rez0
用于 AI 代理的浏览器自动化 CLI。当用户需要与网站交互(包括浏览页面、填写表单、点击按钮、截图等)时使用。
★ 865 📥 343,891