orka-memory

分类: 工具与效率 | 上传者: joey-barbier | 下载: 0 | 版本: v1.0（最新）

Restore project context from memory files, or initialize them if none exist. Use when user says "where were we", "resume", "catch up", "refresh context", "init memory", or at the start of a new session to reload project state.

更新日志: Source: GitHub https://github.com/joey-barbier/ClaudeCode-Plugin

目录结构

当前层级: plugins/cc-memory/skills/orka-memory/

📄 SKILL.md 3.4 KB

SKILL.md

---
name: orka-memory
description: Restore project context from memory files, or initialize them if none exist. Use when user says "where were we", "resume", "catch up", "refresh context", "init memory", or at the start of a new session to reload project state.
allowed-tools: Read, Write, Glob, Grep, Bash
---

# Orka Memory - Restore Context

Restore project context from existing memory files, or create them from scratch.

## Scope

Reads and creates memory files only. Does NOT:
- Modify project source code
- Make architectural decisions
- Delete existing memory files (only adds/updates)

## Step 1: Find Memory Files

Use Glob to search in priority order:

1. `.claude/memory/**/*.md` - Dedicated memory folder
2. `docs-architecture/**/*.md` - Architecture documentation
3. `CLAUDE.md`, `.claude/CLAUDE.md` - Root config
4. `documentations/**/*.md`, `docs/**/*.md` - General docs
5. `README.md` - Fallback

**If files found** -> go to Step 2 (Restore)
**If no files found** -> go to Step 3 (Initialize)

## Step 2: Restore Context

Read each file found. Priority order:

| File | Content |
|------|---------|
| `PROJECT_STATE.md` | Current state, done/not done, bugs |
| `ARCHITECTURE.md` | Structure, flows, components |
| `DECISIONS.md` | Technical choices, conventions |
| `NEXT_STEPS.md` | Roadmap, next tasks |
| `COMMANDS.md` | Useful commands |
| `CLAUDE.md` | Project instructions |

Produce a structured summary:

```markdown
## Context Restored

**Project**: [Name deduced from files]
**Stack**: [Technologies identified]
**Phase**: [Current state]

### What's DONE
- [Feature 1]
- [Feature 2]

### What REMAINS
- [Priority 1]
- [Priority 2]

### Points of Attention
- [Known bugs, limitations, important decisions]

---
What do we work on next?
```

## Step 3: Initialize Memory

When no memory files exist, offer to create them:

```
No memory files found in this project.

Locations checked:
- .claude/memory/
- docs-architecture/
- CLAUDE.md
- documentations/

I can initialize memory files by analyzing the project. Proceed?
```

If the user agrees:

### 3a. Gather Project Information

Explore the project to determine:
1. **Project name**: from `package.json`, `Cargo.toml`, `setup.py`, `*.xcodeproj`, or directory name
2. **Tech stack**: from config files, dependencies, file extensions
3. **Current state**: from git log, README, existing docs
4. **Project structure**: from directory layout

### 3b. Create `.claude/memory/` with these files:

**PROJECT_STATE.md** - Current phase, what's done/remaining, known bugs, attention points

**ARCHITECTURE.md** - Tech stack, project structure, key components, data flow

**DECISIONS.md** - Conventions (commit format, branch naming, code style) + key decisions table

**NEXT_STEPS.md** - Immediate priority, short term, backlog

**COMMANDS.md** - Dev server, test, build, git workflow commands

### 3c. Confirm

```
Memory initialized at .claude/memory/

Created: PROJECT_STATE.md, ARCHITECTURE.md, DECISIONS.md, NEXT_STEPS.md, COMMANDS.md

Use /cc-memory:orka-memory to reload this context in future sessions.
Use /cc-memory:orka-save to save progression after significant work.
```

## Rules

1. **Never invent** -- only use information from actual files or project analysis
2. **Mark unknowns** as "TBD" rather than guessing
3. **Fill with real data** from project analysis, not empty templates
4. **Always conclude** with "What do we work on next?" or equivalent
5. **Respect existing** -- if some memory files exist, only create missing ones

登录后下载/点赞/收藏 ❤ 10 | ★ 0

orka-memory

目录结构

SKILL.md

举报内容

提示