A structured task management system for AI agents that ensures complex workflows survive context loss through persistent knowledge graph notes.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-tasks
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install memory-tasks using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Memory Tasks?
Memory Tasks is a sophisticated framework designed to solve the problem of 'AI amnesia' during long-running projects. By leveraging Basic Memory schemas, this skill allows AI agents to transform ephemeral goals into structured, queryable notes within a knowledge graph. Every task is treated as a first-class entity that includes status tracking, step-by-step progress, and rich context preservation.
Built as a core component of Openclaw Skills, Memory Tasks ensures that an agent can stop, compact its memory, and restart without losing the thread of its work. It provides a standardized way to define what needs to be done, who is doing it, and exactly what context is required to resume work after a session reset or context window flush.
Memory Tasks Use Cases
- Managing multi-step engineering workflows that exceed 3 steps or standard context limits.
- Resuming complex development tasks after an agent restart or session compaction.
- Performing a pre-compaction flush to capture current state and mental models before they are lost.
- Auditing project progress via the knowledge graph to identify blocked or abandoned tasks.
How Memory Tasks Works
- Task Creation: When a complex objective is identified, the agent creates a note with the Task type, populating both metadata for querying and observations for schema validation.
- State Tracking: As work progresses, the agent updates the note's current_step and context fields to reflect the latest technical decisions and file changes.
- Persistence: The task remains in the knowledge graph, safely stored outside the active conversation window.
- Context Resumption: Upon a new session, the agent queries for active tasks, reads the detailed context, and picks up exactly where it left off.
- Completion: Once the objective is met, the status is updated to done with a final summary for future reference.
Memory Tasks Setup
To implement Memory Tasks within your Openclaw Skills environment, ensure the Task schema is defined at memory/schema/Task.md:
# Ensure your memory directory is initialized
mkdir -p memory/schema/
# The Task schema enables validation and advanced querying
# Use the write_note tool to initialize your first Task
# Example Python initialization call
write_note(
title="Setup Project CI",
directory="tasks",
note_type="Task",
metadata={"status": "active", "current_step": 1},
content="# Setup Project CI\n\n## Observations\n- [status] active\n- [current_step] 1"
)
Memory Tasks Data Schema & Taxonomy
Memory Tasks utilizes a strict YAML-based schema to ensure data integrity across the knowledge graph:
| Field | Type | Description |
|---|---|---|
status |
Enum | active, blocked, done, or abandoned |
steps |
Array | Ordered list of concrete, checkable actions |
current_step |
Integer | 1-indexed pointer to the current action |
context |
String | Critical technical context needed for resumption |
parent_task |
Relation | Link to a parent Task entity in the graph |
Memory Tasks Advanced Features
- Schema Validation: Use
schema_validateto identify tasks that are missing critical metadata or have drifted from the defined structure. - Knowledge Graph Integration: Tasks can be linked to other notes, documentation, or code entities using standard bidirectional links.
- Metadata Filtering: Perform advanced searches across all Openclaw Skills notes to find tasks assigned to specific agents or with specific priority levels.
- Automated Flushes: Trigger bulk updates of all active tasks during imminent compaction events to prevent data loss.
SKILL.md
Loading
Related Openclaw Skills
Tabby Virtual AI Cat
A virtual pet care skill for AI agents to adopt, feed, and raise a social Tabby cat via real-time API interactions and emotional feedback loops.
Scrapclaw
A Dockerized browser-backed scraping service that fetches HTML from JavaScript-heavy or Cloudflare-protected websites via an HTTP API.
Personal Insight Engine (PIE)
A strategic analysis tool that distills actionable insights from local session logs using advanced LLMs.
Bracket Oracle
A sophisticated March Madness optimization tool that utilizes Monte Carlo simulations and statistical efficiency data to generate winning tournament brackets.
Terraform Helper
An offline, high-performance reference tool for Terraform concepts, implementation patterns, and infrastructure best practices.
MoltStreet News
AI-curated market research and multi-analyst ETF analysis providing opposing perspectives and source links for informed financial decision-making.
Featured*
