用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/paxlabs-inc/matrix-core --skill project-planning命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
| name | project-planning |
| description | Use to know about Planning Artifacts used in project management. |
Throughout the project-planning instructions you will see often refernces to types in the form of [Epic], [Spec], [Story], [Task], [Research], and [Decision].
This is to reinforce to you the reader or agent that these are distinct artifact types with specific roles.
Projects flow through five distinct phases. Understanding where artifacts belong is critical for agents executing this skill.
Agents make tactical decisions autonomously (which artifact to create, when to move statuses, validating content) but escalate strategic decisions to Q (scope changes, timeline shifts, technical refactoring, resource constraints).
Planning — Create artifacts defining the work (PRD → Epic → Spec → Stories → Tasks)
references/workflow/planning-phase.md) for detailed artifact creation sequence and validation gatesExecution — Move work to completion through status transitions
references/workflow/execution-phase.md) for execution guidanceClosing — Archive completed work and formally close Epic/Project
references/workflow/closing-phase.md) for closing proceduresRetrospective — Capture learnings (can follow Epic or Project completion)
references/workflow/retrospective-phase.md) for retrospective guidancePause and escalate to Q when encountering:
→ See: skill_reference(references/workflow/overview.md) for complete escalation matrix, artifact relationship diagram, and agent decision model.
When estimation is required, read the resources:
references/estimation.md) for detailed estimation techniques and best practices.Artifacts maintain clear relationships to ensure traceability and context throughout the project lifecycle.
references/relationships.md) for detailed relationship types and usage guidelines for all artifact types (PRD, Epic, Research, Story, Task, Retrospective, Decision).Each artifact type has valid status progressions that guide state management throughout the project lifecycle.
references/status-flow.md for detailed status flow diagrams and descriptions for all artifact types (PRD, Epic, Research, Story, Task, Retrospective, Decision).All artifacts follow consistent schema patterns for organization and linking. Each artifact type has specific frontmatter requirements and content sections.
→ Use the linked skill resources for detailed schema definitions, artifact content structure, and link type definitions for all artifact types (PRD, Epic, Research, Story, Task, Retrospective, Decision).
See below for specific artifact schemas:
schema/01-artifact-prd.md)schema/02-artifact-epic.md)schema/03-artifact-spec.md)schema/04-artifact-research.md)schema/05-artifact-decision.md)schema/06-artifact-story.md)schema/07-artifact-task.md)schema/08-artifact-retrospective.md)Multi-source deep research through Matrix web search, fetch, and the router-owned Exa MCP. Searches, extracts, synthesizes, and delivers cited reports with explicit grounding.
Grounded semantic search and asynchronous research through Matrix's router-owned Exa MCP. Use for controlled web evidence, company research, known-URL extraction, outbound briefs, or source-grounded drafts.
The all-encompassing default for the Paxeer/Matrix runtime. One skill that builds software (files + shell + supervised services + git), runs research (web search + fetch + full Playwright browser), and operates the Paxeer network end-to-end (reads across RPC/explorer/portfolio/price/precompiles, and wallet-gated writes: transfers, DEX swaps, perps, token launches, payment streams, staking, scheduled txs). Verbs build/modify/deliver are lifecycle stages, not domains. Every outcome is grounded in a real tool result — file bytes, exit code, service log, commit hash, fetched content, a quote, or a tx hash — never asserted from memory. Destructive, irreversible, or value-moving actions stop to confirm when unbounded, unauthorized, or unclear.
基于 SOC 职业分类