← Back to Skills Marketplace
paitesanshi

ClawBoss - AI Productivity Coach

by Lei Wang · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
569
Downloads
0
Stars
2
Active Installs
1
Versions
Install in OpenClaw
/install clawboss
Description
Transforms your OpenClaw agent into an AI coach using the GROW model to guide goal setting, track progress, adapt intensity, and facilitate reflection with m...
README (SKILL.md)

ClawBoss Skill

Description: AI productivity coach using professional coaching techniques (GROW model, powerful questions, adaptive accountability)

Version: 1.0.0

Author: ClawBoss Contributors


What It Does

ClawBoss transforms your OpenClaw agent into a productivity coach that:

  • Guides goal setting using the GROW coaching model
  • Tracks progress with non-judgmental accountability
  • Adapts intensity based on your momentum
  • Facilitates reflection for continuous improvement
  • Maintains memory of your patterns and insights

Unlike traditional task managers, ClawBoss uses coaching questions to help you discover your own solutions and stay motivated.


Tools

1. clawboss-breakdown

Description: Interactive GROW model session for goal setting and task breakdown

Usage:

Use the clawboss-breakdown tool to guide goal setting with coaching questions.

Process:

  1. Goal Discovery - "What do you want to achieve?"
  2. Reality Check - "Where are you now?"
  3. Options Exploration - "What are possible approaches?"
  4. Action Commitment - "What will you do first?"

Output:

  • Creates memory/tasks/{goal-name}.md with structured plan
  • Updates memory/clawboss-state.json
  • Sets up progress tracking

2. clawboss-check

Description: Coaching-style progress review with adaptive questioning

Usage:

Use the clawboss-check tool to review progress on active tasks.

Process:

  1. Read active tasks from state
  2. Compare planned vs actual progress
  3. Ask appropriate coaching questions:
    • Completed → "What helped you succeed?"
    • Partial → "What happened? How can we adjust?"
    • Not started → "What's blocking you?"
  4. Update task file and momentum state

Output:

  • Updated task progress
  • Coaching feedback
  • Adjusted momentum score

3. clawboss-reflect

Description: Weekly/monthly deep reflection session

Usage:

Use the clawboss-reflect tool for weekly or monthly review.
Specify period: "week" or "month"

Process:

  1. Review all tasks in the period
  2. Identify patterns (successes, blockers, time-of-day effectiveness)
  3. Ask reflection questions:
    • "What did you learn about yourself?"
    • "What strategies worked?"
    • "What needs to change?"
  4. Update long-term insights

Output:

  • Reflection summary
  • Updated insights in state
  • Recommendations for next period

4. clawboss-state

Description: Read or update ClawBoss state (momentum, goals, insights)

Usage:

Use clawboss-state to read current momentum, goals, and insights.

Output:

  • Current momentum level (high/medium/low)
  • Active goals summary
  • Recent insights

Data Structure

memory/clawboss-state.json

Central state file tracking:

  • User preferences
  • Current momentum (with trend)
  • Active goals
  • Check-in timestamps
  • Long-term insights

memory/tasks/{goal-name}.md

Individual task files with:

  • Goal statement and motivation
  • Current reality assessment
  • Action plan
  • Daily progress log
  • Lessons learned

Coaching Frameworks

GROW Model

Goal - What do you want?
Reality - Where are you now?
Options - What could you do?
Will - What will you do?

Powerful Questions

Instead of:

  • ❌ "Did you finish?"
  • ❌ "Why didn't you do it?"

Ask:

  • ✅ "What helped you complete this?"
  • ✅ "What got in the way?"
  • ✅ "What will you try differently?"

Adaptive Intensity

High Momentum (5+ consecutive days)

  • Challenge with bigger goals
  • Celebrate wins
  • Push boundaries

Medium Momentum (steady progress)

  • Maintain current pace
  • Gentle encouragement
  • Consistent check-ins

Low Momentum (3+ days no progress)

  • Reduce to tiny first steps
  • Explore blockers
  • Reframe setbacks as learning

Heartbeat Integration

ClawBoss uses OpenClaw's heartbeat system for automatic check-ins:

Morning (09:00-10:00)

  • Review today's plans
  • Set intention
  • Quick motivation

Evening (19:00-21:00)

  • Progress review
  • Coaching questions
  • Update momentum

Weekly (Sunday 19:00)

  • Deep reflection
  • Pattern analysis
  • Next week planning

See HEARTBEAT.md for configuration.


Installation

Automatic via npx:

npx clawboss@latest

Manual:

  1. Copy this skill to ~/.openclaw/skills/clawboss/
  2. Add to openclaw.json:
    {
      "skills": {
        "entries": {
          "clawboss": {
            "enabled": true
          }
        }
      }
    }
    
  3. Inject templates into workspace files
  4. Restart OpenClaw

Requirements

  • OpenClaw >= 1.0.0
  • Node.js >= 18
  • No external API keys required

Privacy

All data stored locally in ~/.openclaw/workspace/memory/:

  • No cloud sync
  • No external services
  • User owns all data

License

MIT


Learn More

Usage Guidance
What to check before installing or enabling ClawBoss: - Source and distribution: SKILL.md suggests installing via `npx clawboss@latest`, but the registry metadata lists no install spec and source/homepage are missing. If you plan to use npx, inspect the package on npm (publisher, version, tamper history) and the GitHub repo referenced in SKILL.md before running it. - Review files that will be written: The skill will create/modify files under your OpenClaw workspace (memory/tasks/*.md and memory/clawboss-state.json) and may inject templates into workspace files or require adding entries to openclaw.json. Back up existing workspace data and inspect any templates it will inject. - Inspect the remaining code not fully shown (state-manager.js and any omitted files) for unexpected network calls, unusual filesystem paths, or code that modifies system-wide configs. The reviewed files do not show outgoing network activity, but the omitted files are important to verify. - Persona and privacy: The persona set includes an 'intimate-partner' voice that uses affectionate language. That can encourage personal disclosures. If you want a strictly professional tone, switch persona to a coach/mentor/buddy style or inspect persona defaults. - Installation preference: Prefer manual install by copying the provided skill files into your skills directory so you can audit the code, rather than running npx blindly. If you must use npx, first confirm the npm package identity and inspect package contents. - If you lack the ability to audit the rest of the code, treat this as higher-risk: either refrain from network installs or run the skill in a restricted environment. If you want, I can (a) point out the specific locations to inspect in the omitted state-manager file, (b) search the repo manifest for HEARTBEAT.md / templates it plans to inject, or (c) help craft a checklist for auditing the npm package before running `npx`.
Capability Analysis
Type: OpenClaw Skill Name: clawboss Version: 1.0.0 The OpenClaw AgentSkills skill bundle 'ClawBoss' is classified as benign. Its functionality as an AI productivity coach, including goal setting, progress tracking, and reflection, is entirely consistent with its stated purpose. All data is stored locally within the `~/.openclaw/workspace/memory/` directory, as explicitly stated in `SKILL.md` and confirmed by the code. The `scripts/state-manager.js` file implements robust filename sanitization (`sanitizeFilename`) to prevent path traversal vulnerabilities when creating or accessing task files. There is no evidence of data exfiltration, malicious execution (e.g., shell commands with unsanitized input), persistence mechanisms, or prompt injection attempts against the AI agent. The `npx` command in the installation instructions is for user-initiated setup, not for agent execution.
Capability Assessment
Purpose & Capability
Name/description (AI productivity coach using GROW) matches the included code: goal breakdown, progress checks, reflection, heartbeat, persona system and local state files. Minor mismatch: the registry entry says "No install spec — instruction-only skill", yet the package contains multiple runnable JS modules and the SKILL.md instructs installation via `npx clawboss@latest` or copying into ~/.openclaw/skills. This is a distribution/install inconsistency to be aware of but does not by itself indicate malicious intent.
Instruction Scope
Runtime instructions are scoped to coaching tasks and local state management (memory/tasks/*.md and memory/clawboss-state.json). The skill will read/write task/ state files and uses OpenClaw's heartbeat for automatic check-ins. The SKILL.md and code also reference injecting templates into workspace files and editing openclaw.json; that requires file system writes outside just a cache directory — confirm which templates and files are modified. The included persona 'intimate-partner' produces personal/romantic messaging; while not a security vulnerability, it can prompt sensitive personal disclosure and could be unsuitable for some users.
Install Mechanism
Registry metadata shows no formal install spec but SKILL.md recommends `npx clawboss@latest` (pulling from npm) and also suggests manual copying. This is an inconsistency: the skill bundle includes source files, yet the authoritative install path is unclear. If you follow the npx route, you will fetch code from npm (network install) — verify the npm package and publisher first. No embedded downloads or remote endpoints were found in the inspected code, but the lack of a clear, reproducible install spec and the 'source: unknown' / 'homepage: none' entries increase risk.
Credentials
The skill requests no environment variables or external API keys and claims to store data locally. The code shown does not reference process.env credentials or external APIs. This is proportionate for a local coaching tool.
Persistence & Privilege
The skill writes and updates local state and task files under the OpenClaw workspace (SKILL.md: ~/.openclaw/workspace/memory/, code calls stateManager.saveState/loadTaskFile, etc.). always:false (not force-included). This level of persistence is expected for a coaching skill but you should confirm the exact file paths and ensure the skill only modifies its own workspace files and not other agent/global configs; SKILL.md also instructs editing openclaw.json to enable the skill which is expected behavior.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install clawboss
  3. After installation, invoke the skill by name or use /clawboss
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release: GROW model coaching, adaptive accountability, heartbeat integration
Metadata
Slug clawboss
Version 1.0.0
License
All-time Installs 2
Active Installs 2
Total Versions 1
Frequently Asked Questions

What is ClawBoss - AI Productivity Coach?

Transforms your OpenClaw agent into an AI coach using the GROW model to guide goal setting, track progress, adapt intensity, and facilitate reflection with m... It is an AI Agent Skill for Claude Code / OpenClaw, with 569 downloads so far.

How do I install ClawBoss - AI Productivity Coach?

Run "/install clawboss" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is ClawBoss - AI Productivity Coach free?

Yes, ClawBoss - AI Productivity Coach is completely free (open-source). You can download, install and use it at no cost.

Which platforms does ClawBoss - AI Productivity Coach support?

ClawBoss - AI Productivity Coach is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created ClawBoss - AI Productivity Coach?

It is built and maintained by Lei Wang (@paitesanshi); the current version is v1.0.0.

💬 Comments