Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AgentTools

License: MIT Python

AgentTools is AIML Solutions' lane for deterministic, CLI-first agent skills and MCP-adjacent tooling.

The goal is to make agent capabilities inspectable: each skill should have a stable command surface, clear inputs, reproducible outputs, and a small test path.

AgentTools also includes lightweight AI harness artifacts for turning agent runs into auditable episode packages: task specs, tool registries, verification reports, failure attributions, entropy audits, and intervention logs.

Who This Is For

  • Engineers who want small, inspectable agent tools instead of opaque prompt piles
  • Teams evaluating coding agents and needing evidence artifacts around each run
  • Recruiters or clients looking for proof of CLI tooling, repo-quality automation, and AI harness discipline

Proof Points

Area Current evidence
CLI skills repo-health, python-project-audit
Harness artifacts 7 public-safe JSON templates plus a project metrics example
Tests python3 -m unittest discover -s tests
Example outputs examples/repo-health-report.example.md, examples/python-project-audit.example.json
Public boundary examples avoid secrets, client data, private task details, and live financial data

Implementation Status

This repository currently contains three implemented CLI skills and a lightweight CI smoke test. Additional skills should be added only when they have a runnable command, documented input, and example output.

Current Skills

Skill Purpose
repo-health Repo hygiene scoring and Markdown reporting
python-project-audit Python packaging and test-readiness checks
harness-audit AI harness readiness scoring for repos or agent-run folders

Quick Start

Run a repository health check:

bash skills/repo-health/run.sh .

Run a Python project audit:

python3 skills/python-project-audit/run.py .

Run an AI harness audit:

python3 skills/harness-audit/run.py . --markdown

Review example outputs:

How It Fits MultiClaw

This repo should become the bridge between public MultiClaw workflows and agent-executable maintenance tasks. It complements mcp-fastmcp-starter, which demonstrates a FastAPI-style tool service surface.

The harness/ directory provides public-safe artifact templates for model-harness-environment workflows. Use them when a task needs evidence beyond a final patch or prose answer.

Commercial Use Cases

  • Repo quality audits before technical interviews, client delivery, or open-source release
  • Agent-run evidence packages for coding-agent evaluation and technical review
  • Lightweight harness templates for teams adopting AI-assisted software workflows
  • Portfolio proof for Python automation, verification discipline, and agentic systems engineering

Release Discipline

Skills in this repo should inspect repository structure and metadata with predictable, reviewable output. Keep examples generic, reproducible, and suitable for technical review.

Next Improvements

  • Add example outputs for each skill
  • Add tests for each CLI entry point
  • Add a comparison note showing when to use this repo vs mcp-fastmcp-starter
  • Add additional skills only after they have runnable implementations
  • Add validation tests for harness JSON artifacts

About

MCP integration lane for deterministic tool adapters, protocol-first connectivity, and cross-system agent interoperability.

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages