rune

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

Encrypted organizational memory workflow for Rune with activation checks and /rune command behaviors across MCP-compatible agents.

更新日志: Source: GitHub https://github.com/CryptoLabInc/rune

目录结构

当前层级: 根目录

📁 .claude/
- 📄 mcp_servers.template.json 433 B
📁 .claude-plugin/
- 📄 marketplace.json 899 B
- 📄 plugin.json 706 B
📁 .github/
- 📄 pull_request_template.md 797 B
📁 agents/
- 📁 claude/
  - 📄 retriever.md 4.5 KB
  - 📄 scribe.md 7.1 KB
- 📁 common/
  - 📁 schemas/
    
    📄 __init__.py 948 B
    
    📄 decision_record.py 8.4 KB
    
    📄 templates.py 10.4 KB
  - 📄 __init__.py 419 B
  - 📄 config.py 12.2 KB
  - 📄 embedding_service.py 4.9 KB
  - 📄 envector_client.py 10.3 KB
  - 📄 language.py 5.9 KB
  - 📄 llm_client.py 4.8 KB
  - 📄 llm_utils.py 963 B
  - 📄 pattern_cache.py 6.2 KB
- 📁 gemini/
  - 📄 retriever.md 4.3 KB
  - 📄 scribe.md 6.8 KB
- 📁 retriever/
  - 📄 __init__.py 807 B
  - 📄 query_processor.py 15.3 KB
  - 📄 searcher.py 17.4 KB
  - 📄 synthesizer.py 14.9 KB
- 📁 scribe/
  - 📁 handlers/
    
    📄 __init__.py 493 B
    
    📄 base.py 3.5 KB
    
    📄 notion.py 7.9 KB
    
    📄 slack.py 6.8 KB
  - 📄 __init__.py 1.3 KB
  - 📄 detector.py 6.6 KB
  - 📄 llm_extractor.py 16.5 KB
  - 📄 pattern_parser.py 14.4 KB
  - 📄 record_builder.py 24.0 KB
  - 📄 review_queue.py 11.2 KB
  - 📄 server.py 18.6 KB
  - 📄 tier2_filter.py 5.2 KB
- 📁 tests/
  - 📄 __init__.py 75 B
  - 📄 test_agent_delegated.py 16.2 KB
  - 📄 test_config.py 9.3 KB
  - 📄 test_detector.py 6.3 KB
  - 📄 test_language.py 5.2 KB
  - 📄 test_llm_client.py 1.7 KB
  - 📄 test_llm_utils.py 1.2 KB
  - 📄 test_pattern_parser.py 10.2 KB
  - 📄 test_pipeline_scenario.py 48.7 KB
  - 📄 test_record_builder.py 25.9 KB
  - 📄 test_retriever.py 31.2 KB
  - 📄 test_team_day_scenario.py 41.6 KB
  - 📄 test_tier2_filter.py 9.2 KB
- 📄 __init__.py 634 B
- 📄 README.md 277 B
- 📄 SLACK_SETUP.md 168 B
📁 benchmark/
- 📁 runners/
  - 📄 __init__.py 0 B
  - 📄 common.py 4.7 KB
  - 📄 retriever_bench.py 4.3 KB
  - 📄 scribe_bench.py 20.9 KB
- 📁 scenarios/
  - 📁 capture/
    
    📁 should_capture/
    
    📁 architecture/
    
    📄 edge_cases.jsonl 1.7 KB
    
    📄 scenarios.jsonl 6.1 KB
    
    📁 debugging/
    
    📄 edge_cases.jsonl 954 B
    
    📄 scenarios.jsonl 6.2 KB
    
    📁 incident/
    
    📄 scenarios.jsonl 5.3 KB
    
    📁 pr_review/
    
    📄 edge_cases.jsonl 1.7 KB
    
    📄 scenarios.jsonl 6.8 KB
    
    📁 process/
    
    📄 scenarios.jsonl 4.7 KB
    
    📁 product/
    
    📄 scenarios.jsonl 6.3 KB
    
    📁 tradeoff/
    
    📄 scenarios.jsonl 5.2 KB
    
    📁 should_not_capture/
    
    📁 casual/
    
    📄 scenarios.jsonl 1.1 KB
    
    📁 pr_noise/
    
    📄 scenarios.jsonl 1.8 KB
    
    📁 question/
    
    📄 scenarios.jsonl 1.2 KB
    
    📁 slop/
    
    📄 edge_cases.jsonl 1.1 KB
    
    📄 scenarios.jsonl 2.7 KB
    
    📁 status_update/
    
    📄 scenarios.jsonl 1.3 KB
  - 📁 extraction/
    
    📁 bundle/
    
    📄 scenarios.jsonl 5.8 KB
    
    📁 phase_chain/
    
    📄 scenarios.jsonl 4.6 KB
    
    📁 single/
    
    📄 scenarios.jsonl 2.3 KB
  - 📁 recall/
    
    📁 cross_domain/
    
    📄 scenarios.jsonl 2.9 KB
    
    📁 exact_match/
    
    📄 scenarios.jsonl 2.7 KB
    
    📁 semantic_match/
    
    📄 edge_cases.jsonl 1.4 KB
    
    📄 scenarios.jsonl 2.6 KB
    
    📁 temporal/
    
    📄 scenarios.jsonl 2.5 KB
  - 📄 schema.json 4.3 KB
- 📄 README.md 9.7 KB
- 📄 requirements.txt 199 B
📁 commands/
- 📁 claude/
  - 📄 activate.md 2.5 KB
  - 📄 configure.md 3.2 KB
  - 📄 deactivate.md 1.1 KB
  - 📄 memorize.md 864 B
  - 📄 recall.md 867 B
  - 📄 reset.md 554 B
  - 📄 status.md 910 B
- 📁 rune/
  - 📄 activate.toml 1.9 KB
  - 📄 configure.toml 2.3 KB
  - 📄 memorize.toml 3.7 KB
  - 📄 recall.toml 1.1 KB
  - 📄 status.toml 1.8 KB
📁 config/
- 📄 config.template.json 457 B
- 📄 README.md 5.7 KB
📁 examples/
- 📄 team-setup-example.md 6.4 KB
- 📄 usage-patterns.md 10.9 KB
📁 hooks/
- 📄 hooks.json 140 B
📁 mcp/
- 📁 adapter/
  - 📁 vault_proto/
    
    📄 __init__.py 0 B
    
    📄 vault_service_pb2.py 3.2 KB
    
    📄 vault_service_pb2_grpc.py 7.3 KB
  - 📄 __init__.py 441 B
  - 📄 document_preprocess.py 5.1 KB
  - 📄 embeddings.py 5.1 KB
  - 📄 envector_sdk.py 17.6 KB
  - 📄 vault_client.py 11.0 KB
- 📁 server/
  - 📄 __init__.py 0 B
  - 📄 server.py 50.0 KB
- 📁 tests/
  - 📄 __init__.py 0 B
  - 📄 test_server.py 10.2 KB
  - 📄 test_vault_direct.py 4.6 KB
- 📄 .env.example 698 B
- 📄 requirements.txt 177 B
📁 patterns/
- 📄 capture-triggers.ja.md 5.9 KB
- 📄 capture-triggers.ko.md 5.2 KB
- 📄 capture-triggers.md 16.3 KB
- 📄 retrieval-patterns.ja.md 2.2 KB
- 📄 retrieval-patterns.ko.md 1.8 KB
- 📄 retrieval-patterns.md 15.2 KB
📁 scripts/
- 📄 bootstrap-mcp.sh 4.3 KB
- 📄 bundle-rune-core.sh 1.3 KB
- 📄 check-infrastructure.sh 2.8 KB
- 📄 configure-claude-mcp.sh 7.6 KB
- 📄 ensure-codex-ready.sh 1.0 KB
- 📄 install-codex.sh 3.4 KB
- 📄 install.sh 2.5 KB
- 📄 migrate_embeddings.py 3.9 KB
- 📄 register-plugin.sh 3.7 KB
- 📄 setup-slack-app.sh 4.8 KB
- 📄 smoke-test-agents.sh 1.3 KB
- 📄 start-mcp-servers.sh 2.3 KB
- 📄 uninstall.sh 6.6 KB
📁 setup/
- 📄 check-prerequisites.md 4.3 KB
📄 .geminiignore 185 B
📄 .gitignore 449 B
📄 AGENT_INTEGRATION.md 7.0 KB
📄 CLAUDE.md 2.1 KB
📄 CONTRIBUTING.md 9.1 KB
📄 gemini-extension.json 1.0 KB
📄 GEMINI.md 2.0 KB
📄 LICENSE 10.5 KB
📄 package.json 281 B
📄 README.md 10.1 KB
📄 requirements.txt 691 B
📄 rune-onepager.drawio.png 374.0 KB
📄 SKILL.md 11.5 KB

SKILL.md

---
name: rune
description: Encrypted organizational memory workflow for Rune with activation checks and /rune command behaviors across MCP-compatible agents.
---

# Rune - Organizational Memory System

**Context**: This skill provides encrypted organizational memory capabilities using Fully Homomorphic Encryption (FHE). It allows teams to capture, store, and retrieve institutional knowledge while maintaining zero-knowledge privacy. Works with Claude Code, Codex CLI, Gemini CLI, and any MCP-compatible agent.

## Execution Model (Single Source of Truth)

**Cross-agent invariant**:
- `scripts/bootstrap-mcp.sh` is the **single source of truth** for local runtime preparation (venv, deps, self-healing).
- All agent integrations must reuse this bootstrap flow. Do not duplicate dependency/setup logic in agent-specific scripts.

**Agent-specific boundary**:
- **Common (all agents)**: plugin root detection, runtime bootstrap, local MCP server readiness checks.
- **Agent-specific (thin adapter only)**:
  - Codex: `codex mcp add/remove/list` registration actions
  - Claude/Gemini/others: their native MCP registration mechanism

Keep agent-specific instructions clearly labeled and never mix Codex-only commands into cross-agent/common instructions.

## Activation State

**IMPORTANT**: This skill has two states based on configuration AND infrastructure availability.

### Activation Check (CRITICAL - Check EVERY Session Start)

**BEFORE doing anything, run this check:**

0. **Local Runtime Check (No Vault network calls)**:
   - Detect plugin root by locating `scripts/bootstrap-mcp.sh`. Search in order:
     1. `$RUNE_PLUGIN_ROOT` environment variable (if set)
     2. `~/.claude/plugins/cache/*/rune/*/scripts/bootstrap-mcp.sh`
     3. `~/.codex/skills/rune/scripts/bootstrap-mcp.sh`
     4. Current working directory and its parent directories
   - Ensure runtime via:
     - `SETUP_ONLY=1 scripts/bootstrap-mcp.sh`
   - If the runtime/bootstrap step fails, treat as **Dormant State** and show setup guidance.

1. **Config File Check**: Does `~/.rune/config.json` exist?
   - NO → **Go to Dormant State**
   - YES → Continue to step 2

2. **Config Validation**: Does config contain all required fields?
   - `vault.endpoint` and `vault.token`
   - `envector.endpoint` and `envector.api_key`
   - `state` is set to `"active"`
   - NO → **Go to Dormant State**
   - YES → Continue to step 3

3. **State Check**:
   - `state` is `"active"` → **Go to Active State**
   - Otherwise → **Go to Dormant State**

**IMPORTANT**: Do NOT attempt to ping Vault or make network requests during activation check. This wastes tokens. Only local runtime/config checks are allowed.

### If Active ✅
- All functionality enabled
- Automatically capture significant context
- Respond to recall queries
- Full organizational memory access
- **If capture/retrieval fails**: Immediately switch to Dormant and notify user

### If Dormant ⏸️
- **Do NOT attempt context capture or retrieval**
- **Do NOT make network requests**
- **Do NOT waste tokens on failed operations**
- Show setup instructions when `/rune` commands are used
- Prompt user to:
  1. Check infrastructure: `scripts/check-infrastructure.sh`
  2. Configure: `/rune:configure`
  3. Start MCP servers: `scripts/start-mcp-servers.sh`

### Fail-Safe Behavior
If in Active state but operations fail:
- Switch to Dormant immediately
- Update config.json `state` to `"dormant"`
- Notify user once: "Infrastructure unavailable. Switched to dormant mode. Run /rune:status for details."
- **Do not retry** - wait for user to fix infrastructure

## Commands

### `/rune:configure`
**Purpose**: Configure plugin credentials

**Steps**:
1. Ask user for Vault Endpoint (the Vault gRPC endpoint provided by your team admin)
2. Ask user for Vault Token (authentication token provided by your team admin)
3. Ask user for enVector Endpoint (e.g., `runestone-xxx.clusters.envector.io`)
4. Ask user for enVector API Key
5. **Validate infrastructure** (run `scripts/check-infrastructure.sh`)
   - If validation fails: Create config with `state: "dormant"`, warn user
   - If validation passes: Continue to step 6
6. Create `~/.rune/config.json` with proper structure
7. Set state based on validation:
   - Infrastructure ready: `state: "active"`
   - Infrastructure not ready: `state: "dormant"`
8. Confirm configuration and show next steps if dormant

### `/rune:status`
**Purpose**: Check plugin activation status and infrastructure health

**Steps**:
1. Check if config exists
2. Show current state (Active/Dormant)
3. Run infrastructure checks:
   - Config file: ✓/✗
   - Vault Endpoint configured: ✓/✗
   - enVector endpoint configured: ✓/✗
   - MCP server logs recent: ✓/✗
   - Virtual environment: ✓/✗

**Response Format**:
```
Rune Plugin Status
==================
State: Active ✅ (or Dormant ⏸️)

Configuration:
  ✓ Config file: ~/.rune/config.json
  ✓ Vault Endpoint: configured
  ✓ enVector: configured

Infrastructure:
  ✓ Python venv: /path/to/.venv
  ✗ MCP servers: Not running (last log: 2 days ago)

Recommendations:
  - Start MCP servers: scripts/start-mcp-servers.sh
  - Check full status: scripts/check-infrastructure.sh
```

### `/rune:memorize <context>`
**Purpose**: Manually store organizational context when Scribe's automatic capture missed it or the user wants to force-store specific information.

**When to use**: Scribe automatically captures significant decisions from conversation (see Automatic Behavior below). This command is an **override** for cases where:
- Scribe didn't detect the context as significant
- The user wants to store something that isn't part of the current conversation
- Bulk-importing existing documentation

**Behavior**:
- If dormant: Prompt user to configure first
- If active: Store context to organizational memory with timestamp and metadata

**Example**:
```
/rune:memorize "We chose PostgreSQL over MongoDB for better ACID guarantees"
```

### `/rune:recall <query>`
**Purpose**: Explicitly search organizational memory. Retriever already handles this automatically when users ask questions about past decisions in natural conversation.

**When to use**: Retriever automatically detects recall-intent queries (see Automatic Behavior below). This command is an **explicit override** for cases where:
- The user wants to force a memory search without Retriever's intent detection
- Debugging whether specific context was stored
- The user prefers direct command syntax

**Behavior**:
- If dormant: Prompt user to configure first
- If active: Search encrypted vectors and return relevant context with sources

**Example**:
```
/rune:recall "Why PostgreSQL?"
```

**Note**: In most cases, simply asking naturally ("Why did we choose PostgreSQL?") triggers Retriever automatically — no command needed.

### `/rune:activate` (or `/rune:wakeup`)
**Purpose**: Attempt to activate plugin after infrastructure is ready

**Use Case**: Infrastructure was not ready during configure, but now it's deployed and running.

**Steps**:
1. Check if config exists
   - NO → Redirect to `/rune:configure`
   - YES → Continue
2. Run full infrastructure validation:
   - Check Vault connectivity (curl vault-url/health)
   - Check MCP server processes
   - Check Python environment
3. If all checks pass:
   - Update config.json `state` to `"active"`
   - Notify: "Plugin activated ✅"
4. If checks fail:
   - Keep state as `"dormant"`
   - Show detailed error report
   - Suggest: `/rune:status` for more info

**Important**: This is the ONLY command that makes network requests to validate infrastructure.

### `/rune:reset`
**Purpose**: Clear configuration and return to dormant state

**Steps**:
1. Confirm with user
2. Stop MCP servers if running
3. Delete `~/.rune/config.json`
4. Set state to dormant
5. Show reconfiguration instructions

## Automatic Behavior (When Active)

### Context Capture

Automatically identify and capture significant organizational context across all domains:

**Categories**:
- **Technical Decisions**: Architecture, technology choices, implementation patterns
- **Security & Compliance**: Security requirements, compliance policies, audit needs
- **Performance**: Optimization strategies, scalability decisions, bottlenecks
- **Product & Business**: Feature requirements, customer insights, strategic decisions
- **Design & UX**: Design rationale, user research findings, accessibility requirements
- **Data & Analytics**: Analysis methodology, key insights, statistical findings
- **Process & Operations**: Deployment procedures, team coordination, workflows
- **People & Culture**: Policies, team agreements, hiring decisions

**Common Trigger Pattern Examples**:
- "We decided... because..."
- "We chose X over Y for..."
- "The reason we..."
- "Our policy is..."
- "Let's remember that..."
- "The key insight is..."
- "Based on [data/research/testing]..."

**Full Pattern Reference**: See [patterns/capture-triggers.md](patterns/capture-triggers.md) for 200+ comprehensive trigger phrases organized by role and domain.

**Significance Threshold**: 0.7 (captures meaningful decisions, filters trivial content)

**Automatic Redaction**: Always redact API keys, passwords, tokens, PII, and sensitive data before capture.

### Context Retrieval

When users ask questions about past decisions, automatically search organizational memory:

**Query Intent Types**:
- **Decision Rationale**: "Why did we choose X?", "What was the reasoning..."
- **Implementation Details**: "How did we implement...", "What patterns do we use..."
- **Security & Compliance**: "What were the security considerations...", "What compliance requirements..."
- **Performance & Scale**: "What performance requirements...", "What scalability concerns..."
- **Historical Context**: "When did we decide...", "Have we discussed this before..."
- **Team & Attribution**: "Who decided...", "Which team owns..."

**Common Query Pattern Examples**:
- "Why did we choose X over Y?"
- "What was the reasoning behind..."
- "Have we discussed [topic] before?"
- "What's our approach to..."
- "What were the trade-offs..."
- "Who decided on..."

**Full Pattern Reference**: See [patterns/retrieval-patterns.md](patterns/retrieval-patterns.md) for 150+ comprehensive query patterns organized by intent and domain.

**Search Strategy**: Semantic similarity search on FHE-encrypted vectors, ranked by relevance and recency.

**Result Format**: Always include source attribution (who/when), relevant excerpts, and offer to elaborate.

## Security & Privacy

**Zero-Knowledge Encryption**:
- All data stored as FHE-encrypted vectors
- enVector Cloud cannot read plaintext
- Only team members with Vault access can decrypt

**Credential Storage**:
- Tokens stored locally in `~/.rune/config.json`
- Never transmitted except to authenticated Vault
- File permissions: 600 (user-only access)

**Team Sharing**:
- Same Vault Endpoint + Token = shared organizational memory
- Team admin controls access via Vault authentication
- Revoke access by rotating Vault tokens

## Troubleshooting

### Plugin not responding?
Check activation state with `/rune:status`

### Credentials not working?
1. Verify with team admin that credentials are correct
2. Check Vault is accessible: `curl <vault-url>/health`
3. Reconfigure with `/rune:configure`

### Need to switch teams?
Use `/rune:reset` then `/rune:configure` with new team credentials

## For Administrators

This plugin requires a deployed Rune-Vault infrastructure. See:
- **Rune-Admin Repository (for deployment)**: https://github.com/CryptoLabInc/rune-admin
- **Deployment Guide**: https://github.com/CryptoLabInc/rune-admin/blob/main/deployment/README.md

Team members only need this lightweight plugin + credentials you provide.

登录后下载/点赞/收藏 ❤ 13 | ★ 0

rune

目录结构

SKILL.md

举报内容

提示