← Back to Skills Marketplace
Smith Matrix
by
Chen Yijun
· GitHub ↗
· v0.1.3
· MIT-0
381
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install smith-matrix
Description
This skill should be used when the user asks to "create a multi-agent system", "spawn agents for parallel tasks", "decompose task recursively", "set up agent...
Usage Guidance
This skill is coherent and self-contained, but it performs filesystem reads/writes — review the files before installing. Because the write/read rules are guidance only (not automatically enforced), run it in an isolated workspace or sandbox first, and avoid pointing tasks at sensitive directories. Verify that you trust the skill source before copying it into your ~/.claude/skills/ directory. If you need stronger guarantees, restrict the agent's filesystem permissions or inspect the generated .smith-matrix contents after each run.
Capability Analysis
Type: OpenClaw Skill
Name: smith-matrix
Version: 0.1.3
The 'Smith Matrix' skill bundle is a framework for recursive task decomposition and multi-agent orchestration using a directory-based isolation protocol. It consists entirely of documentation and markdown instructions (SKILL.md, smith.md) that guide the AI agent to manage sub-tasks within a structured '.smith-matrix/' workspace. The bundle includes safety features such as recursion depth limits (max level 3) and write-access constraints to prevent resource exhaustion and file conflicts. No malicious code, data exfiltration logic, or harmful prompt injections were identified.
Capability Assessment
Purpose & Capability
Name/description describe creating and coordinating multiple local 'Smith' agents; the SKILL.md and supporting docs only require creating and managing a local .smith-matrix directory structure and template files, which is consistent with that purpose.
Instruction Scope
The runtime instructions direct the agent to create directories, read and write template and task files under a .smith-matrix tree and to copy smith.md into child directories. This stays within the skill's purpose. Note: these are instructions to perform filesystem I/O — the protocol constrains writes to the skill's own directories, but that constraint is procedural (guidance) not technically enforced by the platform.
Install Mechanism
No network install or executable downloads are present — the skill is instruction-only and the 'installation' is a simple local copy (cp -r) into a Claude skills directory. This is low-risk compared with arbitrary remote downloads.
Credentials
The skill requests no environment variables, credentials, or config paths. All required operations are local file reads/writes within its own directory tree, matching the declared requirements.
Persistence & Privilege
The skill is not marked always:true and does not request elevated or cross-skill configuration. It instructs creating files under .smith-matrix and copying into the skills folder (normal for a local skill). Autonomous invocation is allowed but is the platform default and not a unique concern here.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install smith-matrix - After installation, invoke the skill by name or use
/smith-matrix - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.3
No file changes detected between versions 0.1.3 and 0.2.0.
- No visible changes or updates in this release.
- All documentation and skill metadata remain unchanged.
v0.1.2
- Updated conceptual references and protocol documentation.
- Improved clarity and detail in the core concepts and usage guidance in docs.
- No changes to the user-facing workflow or API.
v0.1.1
- Added recursive safety limits: agents may only nest 3 layers deep (LEVEL 0–3), and each level can have up to 5 sub-agents.
- Enforced terminal rule: agents at LEVEL ≥ 3 must execute tasks directly, without further decomposition.
- Clarified core concepts by describing new agent hierarchy constraints in SKILL.md.
- No functional or breaking changes to the skill’s usage or core workflow.
v0.1.0
Smith Matrix v0.2.0 introduces a detailed protocol for building conflict-free, recursively self-similar multi-agent systems.
- Added clear triggers and use cases for when to activate the skill (multi-agent, parallelism, matrix setup, recursive tasks).
- Introduced Smith agent concept with hierarchical structure and unique directory isolation for conflict-free parallel processing.
- Documented full initialization workflow, including root agent setup, directory structure, and onboarding files.
- Defined rigorous read-write rules for agent subdirectories to ensure parallel safety.
- Provided step-by-step child agent creation protocol.
- Included best practices, example use cases, and installation instructions.
Metadata
Frequently Asked Questions
What is Smith Matrix?
This skill should be used when the user asks to "create a multi-agent system", "spawn agents for parallel tasks", "decompose task recursively", "set up agent... It is an AI Agent Skill for Claude Code / OpenClaw, with 381 downloads so far.
How do I install Smith Matrix?
Run "/install smith-matrix" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Smith Matrix free?
Yes, Smith Matrix is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Smith Matrix support?
Smith Matrix is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Smith Matrix?
It is built and maintained by Chen Yijun (@cyijun); the current version is v0.1.3.
More Skills