A specialized skill for AI agents to transform raw, unstructured information into a disciplined taxonomy of durable knowledge.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-taxonomist-clarkchenkai
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-taxonomist-clarkchenkai 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 Taxonomist?
Memory Taxonomist is a high-precision protocol designed for turning messy input, such as meeting transcripts or raw agent logs, into a structured and stable knowledge base. Instead of treating every input as a generic string of text, this skill enforces a strict taxonomy to ensure that high-value information is preserved accurately while noise is filtered out.
By utilizing this skill within the Openclaw Skills ecosystem, developers and AI practitioners can significantly improve retrieval quality. The system prioritizes the integrity of stored data over mere volume, ensuring that an agent's long-term memory remains actionable, consistent, and free of contradictions.
Memory Taxonomist Use Cases
- Processing new notes or transcripts that contain a mix of different information types.
- Performing agent memory design or periodic memory cleanup tasks.
- Organizing meeting outputs that combine decisions, user preferences, and open questions.
- Storing user context safely to ensure high-quality retrieval in future sessions.
- Managing complex datasets where differentiating between durable facts and provisional data is critical.
How Memory Taxonomist Works
- Atomic Decomposition: The skill breaks down the input into atomic claims, ensuring that single paragraphs containing multiple info types are split into individual units.
- Classification: Each atomic unit is sorted into specific categories: facts, preferences, procedures, unresolved questions, or exceptions.
- Durability Assessment: The protocol evaluates whether information is durable truth or provisional, preventing recent mentions from overriding established facts without verification.
- Conflict Resolution: It identifies contradictions, overrides, and edge cases before any data is committed to memory.
- Action Recommendation: The process concludes by recommending specific storage actions such as storing, updating, deprecating, or holding for further clarification.
Memory Taxonomist Setup
To integrate Memory Taxonomist into your environment, ensure your agent is configured to recognize Openclaw Skills protocols. Add the skill definition to your project:
# Clone the skill repository or copy the SKILL.md into your skills directory
mkdir -p skills/memory-taxonomist
cp SKILL.md skills/memory-taxonomist/
Configure your AI agent's system prompt to trigger this protocol when handling memory-intensive tasks.
Memory Taxonomist Data Schema & Taxonomy
The skill enforces a strict six-part output contract to organize data effectively. This ensures that every processed memory follows a predictable taxonomy:
| Section | Description |
|---|---|
| Facts | Verified, objective truths and historical data. |
| Preferences | User-specific inclinations that are not universal rules. |
| Procedures | Step-by-step methods or workflows identified in the text. |
| Unresolved Questions | Gaps in knowledge or pending clarifications treated as first-class objects. |
| Exceptions | One-off cases or overrides that deviate from standard procedures. |
| Storage Action | Technical instructions (Store, Update, Deprecate, or Hold). |
Memory Taxonomist Advanced Features
- Granular Classification: Moves beyond generic summaries to treat unresolved questions and exceptions as distinct data points.
- Conflict Flagging: Built-in logic to detect contradictions between new input and existing memory before data corruption occurs.
- Contextual Guardrails: Prevents the 'recency bias' by distinguishing between ephemeral mentions and durable knowledge.
- Multi-Agent Compatibility: Can be used as a pre-processing layer for other agents within the Openclaw Skills framework to ensure clean data ingestion.
SKILL.md
Loading
Related Openclaw Skills
Dual-Mode Reasoner
A disciplined reasoning protocol that protects against high-stakes errors by intelligently switching between rapid execution and deep, risk-aware deliberation.
Feedback Controller
A disciplined, closed-loop protocol for AI agents to identify, localize, and correct execution drift through a structured feedback system.
Goal Clarifier
A purpose-driven skill designed to translate ambiguous requests into structured, executable briefs by focusing on outcomes rather than mechanisms.
OpenUI Workspace Delivery
A local-first MCP workflow for generating, reviewing, and shipping shadcn UI components via a grounded proof loop.
Polymarket Value Trader
A disciplined prediction market trading skill for AI agents that utilizes value-based analysis to trade on Simmer and Polymarket venues.
Shopflow Read-only Packet
A technical integration packet designed to teach AI agents how to interface with Shopflow as a read-only surface using MCP servers.
Featured*
