# Send Type-Based Autonomy to your agent
Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.
## Fast path
- Download the package from Yavira.
- Extract it into a folder your agent can access.
- Paste one of the prompts below and point your agent at the extracted folder.
## Suggested prompts
### New install

```text
I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Then review README.md for any prerequisites, environment setup, or post-install checks. Tell me what you changed and call out any manual steps you could not complete.
```
### Upgrade existing

```text
I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run.
```
## Machine-readable fields
```json
{
  "schemaVersion": "1.0",
  "item": {
    "slug": "autonomy-type-based",
    "name": "Type-Based Autonomy",
    "source": "tencent",
    "type": "skill",
    "category": "开发工具",
    "sourceUrl": "https://clawhub.ai/luciusrockwing/autonomy-type-based",
    "canonicalUrl": "https://clawhub.ai/luciusrockwing/autonomy-type-based",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadUrl": "/downloads/autonomy-type-based",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=autonomy-type-based",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "packageFormat": "ZIP package",
    "primaryDoc": "SKILL.md",
    "includedAssets": [
      "README.md",
      "SKILL.md",
      "_meta.json",
      "references/checkpoints.md",
      "templates/HEARTBEAT.md",
      "templates/QUEUE.md"
    ],
    "downloadMode": "redirect",
    "sourceHealth": {
      "source": "tencent",
      "slug": "autonomy-type-based",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-05-10T07:52:37.152Z",
      "expiresAt": "2026-05-17T07:52:37.152Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=autonomy-type-based",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=autonomy-type-based",
        "contentDisposition": "attachment; filename=\"autonomy-type-based-1.0.0.zip\"",
        "redirectLocation": null,
        "bodySnippet": null,
        "slug": "autonomy-type-based"
      },
      "scope": "item",
      "summary": "Item download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this item.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/autonomy-type-based"
    },
    "validation": {
      "installChecklist": [
        "Use the Yavira download entry.",
        "Review SKILL.md after the package is downloaded.",
        "Confirm the extracted package contains the expected setup assets."
      ],
      "postInstallChecks": [
        "Confirm the extracted package includes the expected docs or setup files.",
        "Validate the skill or prompts are available in your target agent workspace.",
        "Capture any manual follow-up steps the agent could not complete."
      ]
    }
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/autonomy-type-based",
    "downloadUrl": "https://openagent3.xyz/downloads/autonomy-type-based",
    "agentUrl": "https://openagent3.xyz/skills/autonomy-type-based/agent",
    "manifestUrl": "https://openagent3.xyz/skills/autonomy-type-based/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/autonomy-type-based/agent.md"
  }
}
```
## Documentation

### Type-Based Autonomy

Transform your agent from reactive to autonomous worker on specific task types.

### Concept

The agent pulls from tasks/QUEUE.md but only works on tasks tagged with specific types:

📚 Research  → ✅ Autonomy works on these
✍️ Writing   → ✅ Autonomy works on these
🔍 Analysis  → ✅ Autonomy works on these

🧹 Maintenance → ❌ Autonomy SKIPS these (cron handles)
💾 Backup      → ❌ Autonomy SKIPS these (cron handles)

Cron jobs handle backups, cleanup, security audits. Autonomy handles research, writing, analysis.

### 1. Task Queue Structure

Every task in tasks/QUEUE.md has a @type: label:

## 🔴 Ready

### 📚 Research (@type:research)
- [ ] @priority:high @type:research Competitor pricing for X product
- [ ] @priority:medium @type:research Ollama model alternatives

### ✍️ Writing (@type:writing)
- [ ] @priority:medium @type:writing Blog post on memory systems
- [ ] @priority:low @type:writing Documentation update

### 🔍 Analysis (@type:analysis)
- [ ] @priority:medium @type:analysis Review weekly metrics
- [ ] @priority:low @type:analysis Analyze token patterns

### 🧹 Maintenance (@type:maintenance)
→ Autonomy IGNORES, cron handles
- [ ] @priority:medium @type:maintenance Old log cleanup

### 2. Heartbeat Flow

Heartbeat → Check urgent → No → Read QUEUE.md → Filter by @type → Pick task → Work → Update QUEUE → Log

Filter logic:

Read all tasks in 🔴 Ready section
ONLY pick tasks with @type:research | @type:writing | @type:analysis
SKIP tasks with @type:maintenance | @type:backup | @type:security

### 3. Task Completion

1. Mark task as In Progress: @agent: @type:research [task description]
2. Work on it
3. Move to Done Today with completion notes
4. Log to memory/[today].md
5. Check GOALS.md and .learnings/ for follow-up tasks

### Research (@type:research)

Definition: Information gathering, investigation, discovery

Examples:

Competitor analysis
API documentation research
Technology exploration
Market research
Best practices investigation

Output format:

## Research: [Topic]

### Findings
- Key point 1
- Key point 2

### Sources
- [Source 1](url)
- [Source 2](url)

### Recommendations
- Recommendation 1
- Recommendation 2

Add follow-up task to Ideas:

- [Idea: @type:analysis @priority:medium Analyze research findings for X]

### Writing (@type:writing)

Definition: Content creation, documentation, communication

Examples:

Blog posts
Documentation updates
Email drafts
Announcements
Guides/tutorials

Output format:

# [Title]

[Content]

Post-completion:

If email: Add to Ideas for review by RA
If blog/doc: Add to Ideas for publishing
If announcement: Add to Ideas for distribution

### Analysis (@type:analysis)

Definition: Data review, metrics analysis, pattern finding

Examples:

Weekly performance review
Token usage analysis
Log analysis
Trend identification
Metrics dashboard creation

Output format:

## Analysis: [Topic]

### Data Reviewed
- [List of data sources]

### Key Findings
- Finding 1 with metric
- Finding 2 with metric

### Patterns
- Pattern 1
- Pattern 2

### Recommendations
- Action 1
- Action 2

Add follow-up tasks:

- [Idea: @type:writing @priority:medium Write analysis report]
- [Idea: @type:research @priority:low Investigate pattern X further]

### Maintenance (@type:maintenance)

Definition: System cleanup, organization, routine tasks

Handler: Cron (NOT autonomy)

Examples:

Old log cleanup
Temp file deletion
File organization
Archive old records

Behavior:

Autonomy SKIPS these tasks
Cron jobs handle them overnight
Manual trigger if urgent, but usually not needed

### Backup (@type:backup)

Definition: Data backup, version control, sync

Handler: Cron (NOT autonomy)

Examples:

GitHub backup
Database backup
Cloud sync

Behavior:

Autonomy SKIPS these tasks
Scheduled twice daily (00:00, 12:00 UTC)

### Security (@type:security)

Definition: Security checks, audits, vulnerability scans

Handler: Cron (NOT autonomy)

Examples:

Security audit
Permission check
Credential review

Behavior:

Autonomy SKIPS these tasks
Monthly security audit cron (1st of month)

### Priority System

Priority affects task selection order:

PriorityWhen to useSelection@priority:urgentTime-sensitive, deadline < 24hPick FIRST@priority:highImportant, deadline 2-3 daysPick SECONDpriority:mediumNormal importancePick THIRDpriority:lowNice to have, no deadlinePick LAST

### GOALS.md Integration

Every task should support RA's long-term goal: MONEY

When creating tasks:

Check GOALS.md for current objectives
Link tasks to money-making activities
Ask: "How does this help RA make money?"

Examples:

### 📚 Research
- [ ] @priority:high @type:research Competitor pricing (GOAL: pricing strategy for new product)
- [ ] @priority:medium @type:research Market fit analysis (GOAL: validate product idea)

### ✍️ Writing
- [ ] @priority:high @type:writing Sales email template (GOAL: improve conversion)
- [ ] @priority:medium @type:blog Marketing post (GOAL: drive traffic)

Post-completion:

Update GOALS.md with progress notes if relevant
Add to Ideas: follow-up tasks that advance goals

### .learnings/ Integration

When completing tasks, add findings to .learnings/:

After research task:

## [LRN-20260216-001] research-findings
Research: Competitor pricing analysis

Key findings: [summarize]

After analysis task:

## [LRN-20260216-002] analysis-insights
Analysis: Token usage patterns

Key insights: [summarize]

After problem:

## [ERR-20260216-001] research-issue
Error: API rate limit during research

Fix: [document the fix]

### Adding Tasks

RA adds tasks directly to queue:

## 🔴 Ready
- [ ] @type:research @priority:high Analyze competitor X pricing

The agent discovers tasks during work and adds to Ideas:

## 💡 Ideas
- [Idea: @type:research @priority:medium Investigate Ollama alternative models]

### Updating Tasks

When starting:

## 🟡 In Progress
- [ ] @agent: @type:research @priority:high Competitor pricing analysis
  - Started: 2026-02-16 14:00 UTC
  - Progress: Gathering data

When complete:

## ✅ Done Today
- [x] @agent: @type:research @priority:high Competitor pricing analysis
  - Completed: 2026-02-16 14:25 UTC
  - Output: tasks/outputs/competitor-pricing-analysis.md

When blocked:

## 🔵 Blocked
- [ ] @type:writing @priority:medium Email draft (needs: RA review)

### Clearing Done Today

Daily routine (during heartbeat or cron):

Move completed tasks from Done Today to tasks/archive/ if you want to keep history
Or simply delete if not needed
Clear section with: ## ✅ Done Today (empty)

### Token Budget

Recommendation: 4 sessions/day, ~3-8K tokens each = 12-32K/day

Session strategy:

TimeTask TypeTokensFocus09:00 AMResearch8-10KDeep investigation13:00 PMWriting5-7KContent creation17:00 PMAnalysis3-5KData review21:00 PMLearning2-3KReview learnings

When to stop:

If tokens remaining < 5K for the day
If queue has no tasks of allowed types
If RA is actively messaging (human priority)

### Coordination With Cron

Autonomy and cron work in parallel but on different task types:

SystemTask TypesScheduleAutonomyResearch, Writing, AnalysisAny time (heartbeat-driven)CronBackup, Maintenance, SecurityScheduled (midnight, noon, daily)

No conflicts - they work on completely different things.

### When to Use This Skill

Use this skill when:

You want autonomy focused on value-add tasks (research, writing, analysis)
You want maximum token efficiency
Tasks can be clearly categorized by type
You want to expand task types later (add coding, testing as they come up)
You want clear separation between autonomy and maintenance

### When NOT to Use This Skill

Do not use this skill when:

You want continuous work on any task type → Use autonomy-windowed skill
Task types are unclear or ambiguous → Use simpler queue
You want the agent to handle maintenance tasks → Those belong to cron
You want predictable work hours → Use autonomy-windowed for time-based control

### Quick Reference

Task types for autonomy: @type:research | @type:writing | @type:analysis

Task types for cron: @type:maintenance | @type:backup | @type:security

Priority order: urgent → high → medium → low

Queue location: tasks/QUEUE.md

See templates/QUEUE.md for full template structure
## Trust
- Source: tencent
- Verification: Indexed source record
- Publisher: luciusrockwing
- Version: 1.0.0
## Source health
- Status: healthy
- Item download looks usable.
- Yavira can redirect you to the upstream package for this item.
- Health scope: item
- Reason: direct_download_ok
- Checked at: 2026-05-10T07:52:37.152Z
- Expires at: 2026-05-17T07:52:37.152Z
- Recommended action: Download for OpenClaw
## Links
- [Detail page](https://openagent3.xyz/skills/autonomy-type-based)
- [Send to Agent page](https://openagent3.xyz/skills/autonomy-type-based/agent)
- [JSON manifest](https://openagent3.xyz/skills/autonomy-type-based/agent.json)
- [Markdown brief](https://openagent3.xyz/skills/autonomy-type-based/agent.md)
- [Download page](https://openagent3.xyz/downloads/autonomy-type-based)