Agent Plan Manager

Structured plan management for OpenClaw Agent's tasks and execution workflows. Use when the OpenClaw Agent needs to create, track, or break down execution pl...

Security Scan

This is an instruction-only skill that manages Markdown plan files for agent workflows; its requested actions and artifacts are consistent with its description and it does not ask for credentials, installs, or external network endpoints.

Purpose & Capability

Name/description match the actual instructions: creating and updating Markdown plan files and templates for agent task planning. No unrelated binaries, credentials, or services are requested.

Instruction Scope

SKILL.md confines behavior to creating/updating files under a 'plans/' directory and using agent tool sequence names (e.g., browser_navigate, WriteFile) to describe workflows. It does not instruct the agent to read unrelated system state, exfiltrate data, or contact unexpected endpoints.

Install Mechanism

No install spec and no code files — nothing is downloaded or written to disk by an installer. The skill is instruction-only, which minimizes install risk.

Credentials

No environment variables, credentials, or config paths are requested. The templates and instructions do not justify additional secrets or unrelated access.

Persistence & Privilege

always:false (default) and the skill does not request persistent system-wide changes or modify other skills. Autonomous invocation is allowed by platform default but is not elevated here.

Assessment

This skill is internally coherent and limited to creating and maintaining Markdown plan files for the agent. Before installing, confirm you are comfortable with the agent writing/modifying files under a plans/ directory in its working environment. Do not store secrets or credentials in plan files. Also verify which agent tools (e.g., browser_navigate, WriteFile) your agent can execute — those tool executions are implied by the templates and will run with whatever privileges your agent already has. If you want tighter control, restrict the agent's filesystem permissions or review plan files before execution.

Manage OpenClaw Agent's execution plans and task workflows in a structured, trackable way.

Who is it for? OpenClaw Agent that needs to organize tasks, track execution progress, or document step-by-step implementation plans.

What makes it different?

• Agent-Centric: designed for AI agent execution workflows with clear entry/exit points
• Structured: separates high-level goals from detailed tool-calling steps
• Trackable: uses checkboxes and status fields for execution state visibility
• Lightweight: just Markdown files, integrates with OpenClaw's file system tools

Quick Start

1. Create plans/plans-overview.md from the overview template
2. Add execution items to the table with appropriate status
3. If a plan needs detailed steps, create plans/detailed/XX-{plan-name}.md
4. Update plans/detailed/index.md to link the new plan
5. Update statuses as execution progresses

Directory Structure

plans/
├── plans-overview.md      ← from templates/plans-overview.md
└── detailed/
    ├── index.md           ← from templates/index.md
    └── XX-{plan-name}.md  ← from templates/detailed-plan.md

When to Use

• Creating a new execution plan for complex tasks
• Tracking multi-step tool calling workflows
• Breaking down user requests into actionable steps
• Recording detailed execution documentation
• Structured agent task management

Example: Web Scraping Task

Step 1 — Add to plans/plans-overview.md:

| 1 ||***|| Web scraping for product prices ||***|| In Progress ||***|| High ||***|| TBD ||***|| TBD ||***|| Extract prices from example.com |

Step 2 — Create plans/detailed/01-web-scraping.md:

# Detailed Plan - Web Scraping for Product Prices

## Plan ID: #1

## Objective
Extract product prices from example.com and save to a CSV file.

## Schedule
TBD

## Deadline
TBD

## Tool Sequence

1. `browser_navigate` → example.com/products
2. `browser_snapshot` → extract product list
3. `browser_click` → navigate to each product
4. `WriteFile` → save results to products.csv

## Implementation Steps

### 1. Navigate to target URL
- [x] Use browser_navigate to load example.com/products

### 2. Extract product data
- [ ] Capture page snapshot
- [ ] Parse product names and prices
- [ ] Handle pagination if present

### 3. Save results
- [ ] Format data as CSV
- [ ] Write to products.csv

## Estimated Token Usage
~2K tokens

Step 3 — Update plans/detailed/index.md:

| #1 | Web scraping for product prices | TBD | TBD | [01-web-scraping.md](./detailed/01-web-scraping.md) |

Templates

All templates live in templates/:

• plans-overview.md — High-level execution board with status table
• index.md — Index of all detailed execution plans
• detailed-plan.md — Step-by-step execution plan with tool sequences

Tips

• Keep plan names concise; use kebab-case for filenames (01-web-scraping.md)
• Update last_updated in frontmatter whenever the overview changes
• Use checkboxes (- [ ]) in detailed plans to track execution progress
• Include tool sequences to clarify the execution flow
• Archive completed plans by moving them to the "Completed Items" section
• Estimate token usage when possible for complex multi-step tasks