← 返回
未分类 Key 中文

Akievo

Persistent project planning for AI agents. Create, manage, and track long-term goals using structured Kanban boards that survive session resets.
为AI智能体提供持久项目规划功能,使用结构化看板创建、管理和追踪长期目标,会话重置后数据仍保留。
akievo akievo 来源
未分类 clawhub v1.0.0 1 版本 100000 Key: 需要
★ 0
Stars
📥 397
下载
💾 0
安装
1
版本
#latest

概述

Akievo — Agent Plan Mode

You have access to Akievo, a structured project management system. Use it as your persistent memory and planning layer for long-term goals. Akievo boards survive session resets — they are your source of truth.

Core Principles

  1. Always check before creating. At the start of every session, call list_boards to find existing plans before creating new ones.
  2. One board per goal. Each major goal or project gets its own board. Prefix agent-created boards with [Agent] (e.g., [Agent] Launch SaaS Product).
  3. Lists are phases. Use lists to represent sequential phases or categories (e.g., "Research", "Build", "Launch", "Done").
  4. Cards are tasks. Each actionable step is a card. Include clear titles and descriptions.
  5. Respect human edits. The human may add, remove, reprioritize, or comment on cards. Always re-read the board before acting. Never undo or override human changes.

Session Start Pattern

Every time a new session begins:

  1. Call list_boards to find boards prefixed with [Agent]
  2. If a relevant board exists, call get_board with its ID to load the full state
  3. Read the board's project_memory field for context (goal, timeline, assumptions)
  4. Identify the next unblocked, incomplete task
  5. Report status to the user: what's done, what's next, any blockers

Creating a New Plan

When the user describes a new goal:

  1. Call list_workspaces to find available workspaces
  2. Use create_board_with_tasks to scaffold the entire plan in one call:
    • Break the goal into 3–6 phases (lists)
    • Each phase gets 3–8 concrete tasks (cards)
    • Add checklists for tasks with sub-steps
    • Set priorities: critical, high, medium, low
    • Set due dates when the user provides a timeline
  3. Create dependencies between tasks that have a natural order using bulk_create_dependencies
  4. Present the plan to the user and ask for feedback before proceeding

Working on Tasks

When executing on a plan:

  1. Pick the next unblocked, highest-priority incomplete card
  2. Work on it (using your other tools — coding, research, writing, etc.)
  3. Add progress updates as comments using add_comment
  4. When done, call complete_card to mark it finished
  5. If blocked, call block_card with a clear reason
  6. Move to the next task

Updating the Plan

As work progresses, the plan may need adjustment:

  • Add new tasks: create_card in the appropriate list
  • Update details: update_card to change title, description, priority, or due date
  • Reorder: move_card to shift tasks between phases
  • Add sub-tasks: add_checklist_item for granular steps
  • Never delete cards without asking the user first

Progress Reporting

When the user asks for a status update:

  1. Call get_board to get current state
  2. Count completed vs total cards per list
  3. Highlight blocked items and their reasons
  4. Identify upcoming due dates
  5. Suggest next actions

Important Safety Rules

  • Never delete a board without explicit user confirmation
  • Never archive cards without asking
  • Always re-read the board before making changes (the human may have edited it)
  • Log your work — add comments to cards explaining what you did and why
  • Stay scoped — only modify boards you created or were explicitly asked to manage

版本历史

共 1 个版本

  • v1.0.0 当前
    2026-03-31 10:46 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

self-improving agent

pskoett
记录自身发现以实现自我改进的技能
★ 4,176 📥 947,037
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。
★ 875 📥 351,326
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。
★ 1,451 📥 329,714