A persistent memory and monitoring system that creates structured task notes in Obsidian to prevent AI agent amnesia and silent failures.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-keep-alive-for-obsidian
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-keep-alive-for-obsidian 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 Keep-Alive for Obsidian?
Memory Keep-Alive for Obsidian is a specialized skill designed to bridge the gap between AI agent execution and long-term knowledge management. By leveraging the power of Openclaw Skills, this tool ensures that every task assigned to an agent is automatically documented within a structured folder hierarchy in your Obsidian vault. It addresses the common problem of state loss during restarts or context window overflows by maintaining a hard copy of task progress, decisions, and next actions.
Beyond simple documentation, it introduces a sophisticated monitoring loop for long-running operations. This system acts as a watchdog for your AI workflows, ensuring that tasks don't just stall silently. By using Openclaw Skills to manage these persistent state files, developers can build more reliable agents that can resume work across sessions with full context and verified integrity.
Memory Keep-Alive for Obsidian Use Cases
- Managing multi-day research or coding projects that require context to survive agent restarts.
- Automating the creation of a progress trail in Obsidian for every task assigned to an AI agent.
- Monitoring background AI processes to detect and recover from stalls or ambiguous next steps.
- Ensuring structured handoffs between different agent sessions using a standardized RESUME.md format.
- Maintaining a searchable history of AI-driven decisions and discoveries within a personal knowledge base.
How Memory Keep-Alive for Obsidian Works
- Upon task initialization, the skill creates a dedicated directory within the Obsidian vault under the Tasks folder.
- It populates this directory with essential Markdown files—RESUME.md, CHECKLIST.md, and DOCS.md—based on a predefined template.
- The agent is instructed to update these files at every major milestone, ensuring the heartbeat and next action are always current.
- When a user activates the keep-alive loop, a series of scheduled jobs (Watchdog, Replayer, Escalator) begin monitoring the task folders.
- The Watchdog job checks for heartbeats older than 24 hours or missing next actions to identify stalled tasks.
- Integrity layers run periodically via Openclaw Skills to repair missing notes and refresh the master workflow index.
Memory Keep-Alive for Obsidian Setup
To configure the skill, ensure your Obsidian vault path is accessible. The skill will look for an environment variable or default to a standard path.
# Set your custom vault path (Optional)
export OBSIDIAN_VAULT_PATH="/path/to/your/Obsidian Vault"
# The skill uses a canonical template at:
# {baseDir}/templates/TEMPLATE.md
Ensure that the agent has write permissions to the vault directory to manage the Session-Resume-Workflow and individual task folders.
Memory Keep-Alive for Obsidian Data Schema & Taxonomy
The skill organizes data within the Obsidian vault using a standardized hierarchy to ensure compatibility across different Openclaw Skills implementations:
| File/Folder | Purpose |
|---|---|
Tasks/Session-Resume-Workflow/ |
Contains the global TEMPLATE.md, WORKFLOW-INDEX.md, and LOOP-STATE.md. |
<task-name>/RESUME.md |
Core state file: status, last heartbeat, next action, and restart notes. |
<task-name>/CHECKLIST.md |
A numbered list of progress steps and verification status. |
<task-name>/DOCS.md |
Detailed goals, technical decisions, and session notes. |
<task-name>/WATCHDOG.md |
Temporary file generated when a stall is detected, containing blocker details. |
Memory Keep-Alive for Obsidian Advanced Features
- Automatic task heartbeat tracking to distinguish between active, stalled, and complete work.
- Multi-layered monitoring loop with Watchdog, Replayer, and Escalator jobs for autonomous recovery.
- Standardized stall tagging (e.g., blocked-on-external, ambiguous-next-step) for easier debugging.
- Self-healing validator that backfills missing documentation from canonical templates.
- Smoke test integration to verify the health of all scheduled monitoring jobs within the Openclaw Skills framework.
- Command-based loop control using /loop-start and /loop-stop to manage resource usage.
SKILL.md
Loading
Related Openclaw Skills
DeepRead Legal
An AI-powered skill for extracting structured data and redacting PII from legal documents with high accuracy.
Hugging Face Mirror Trending Skill (English)
A zero-dependency Python skill that enables AI agents to fetch and report real-time trending data from Hugging Face via the HF-Mirror API.
Memorial
A sophisticated digital legacy toolkit for building memorial archives with personality simulation and voice cloning capabilities.
Word Jumble Generator
A high-fidelity puzzle engine that generates scrambled word games, literal cartoon hints, and printable rendered images.
Meeting Secretary AI Skill
A professional AI meeting secretary skill designed to perform deep semantic analysis on transcripts and generate high-quality, structured meeting minutes.
bw-openclaw-boost
An efficiency-focused toolkit for OpenClaw that provides cost tracking, memory management, and advanced permission controls.
Featured*
