Back to the catalog

techknowmad-ai-skills

20 enterprise-grade skills for autonomous AI/ML research lab operations. Includes 10 custom gap-analysis skills (research-commander, agent-d

Open source Repository Open in the app JSON README (API)

About

20 enterprise-grade skills for autonomous AI/ML research lab operations. Includes 10 custom gap-analysis skills (research-commander, agent-debugger, model-evaluator, cost-optimizer, patent-researcher, red-team-ai-models, self-healing-agent, zero-touch-ml-pipeline, knowledge-distiller, academic-output-pipeline) and 10 radical combination skills (swarm-commander, pareto-prompt-optimizer, prompt-evolution, multi-agent-research-swarm, self-evolving-skill-ecosystem, skill-factory, antifragile-ai-ops, compliance-as-code, startup-creative-automation, ip-aware-competitive-intel). Built for MLOps, multi-agent orchestration, adversarial testing, and full research lifecycle automation.

Details

Kind
Plugins
Topic
Developer tools
Publisher
techknowmad-labs
Origin
marketplace
Category
ferramentas
Last push
2026-03-12T19:16:24Z
Repository state
arquivado
Language
Shell
License
MIT
Added
2026-08-30 01:48:58
Updated
2026-08-30 01:48:58
Origin id
techknowmad-labs/techknowmad-ai-skills/techknowmad-ai-skills

README

# TechKnowmad AI Skills

[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
[![Claude Code](https://img.shields.io/badge/Claude_Code-Plugin-orange.svg)](https://github.com/TECHKNOWMAD-LABS/techknowmad-ai-skills)
[![Skills](https://img.shields.io/badge/Skills-20-green.svg)](#whats-inside)

> **20 enterprise-grade skills for autonomous AI/ML research lab operations.**
> Custom-built skills + composable combination skills. Designed for [Claude Code](https://docs.claude.com).

## What's Inside

### Custom Skills (10) — Purpose-Built

Original skills addressing gaps in the Claude Code ecosystem:

| Skill | What It Does | Priority |
|-------|-------------|----------|
| `research-commander` | Full research lifecycle orchestrator — literature review → hypothesis → experiment → paper → submission | P0 |
| `swarm-commander` | Multi-agent swarm coordination with consensus protocols and result aggregation | P0 |
| `model-evaluator` | Automated benchmark suite runner with leaderboards and regression detection | P0 |
| `agent-debugger` | Trace agent decision chains, replay failures, diagnose root causes | P1 |
| `self-healing-agent` | Autonomous recovery — detect degradation, auto-fix, escalate when needed | P1 |
| `prompt-evolution` | Genetic algorithm-based prompt optimization across generations | P1 |
| `skill-factory` | Meta-skill that auto-generates new skills from detected patterns | P1 |
| `patent-researcher` | AI/ML patent prior art search, claim drafting, FTO analysis | P2 |
| `cost-optimizer` | Cross-cloud spend analysis, GPU strategies, token cost reduction | P2 |
| `knowledge-distiller` | Compress large documents into optimal skill prompts | P2 |

### Combination Skills (10) — Composable Pipelines

Skills composed from multiple base skills for compound capabilities:

| Skill | Composes | What It Does |
|-------|---------|-------------|
| `multi-agent-research-swarm` | Research Commander + Swarm Commander + arXiv | 10 agents review papers in parallel, synthesize into one review |
| `red-team-ai-models` | Pen Tester + AI Security + Model Evaluator | Adversarial security testing for your own ML models |
| `self-evolving-skill-ecosystem` | Skill Factory + Workflow Miner + Telemetry | Skills that create skills that create skills |
| `academic-output-pipeline` | Research Commander + Paper Writer + Grant Writer | Experiment results → paper → poster → slides → grant proposal |
| `antifragile-ai-ops` | Chaos Engineer + Self-Healing + Debugger | Break agents, watch them self-recover, learn from it |
| `pareto-prompt-optimizer` | Prompt Evolution + Model Evaluator + Cost Optimizer | Find Pareto-optimal prompts (accuracy vs cost vs latency) |
| `zero-touch-ml-pipeline` | Data Pipeline + n8n + Ruflo + Self-Healing | Fully autonomous data → train → deploy → monitor → retrain |
| `startup-creative-automation` | Brand Architect + fal.ai + Presentation Architect | Brand identity → visual assets → pitch deck in one pipeline |
| `compliance-as-code` | Compliance Automator + GRC + Audit Support | Continuous SOC2/ISO/GDPR compliance with auto-evidence |
| `ip-aware-competitive-intel` | Patent Researcher + Literature + Competitive War Room | Patents + papers + products in one competitive landscape view |

## Installation

### As a Claude Plugin

```bash
claude plugin add TECHKNOWMAD-LABS/techknowmad-ai-skills
```

### Quick Install (All Skills)

```bash
chmod +x scripts/install-techknowmad-skills.sh
./scripts/install-techknowmad-skills.sh
```

### Manual Install (Individual Skills)

```bash
# Copy a single skill
cp -r custom-skills/research-commander ~/.claude/skills/research-commander

# Or symlink (preserves git updates)
ln -s "$(pwd)/custom-skills/research-commander" ~/.claude/skills/research-commander
```

### Restart Claude Code

```bash
# Close and reopen Claude Code to load new skills
claude
```

## Prerequisites

These skills are designed to work with the **TechKnowmad MCP stack**. They reference MCPs like arXiv, fal.ai, Langfuse, Ruflo, etc. Skills will still function without all MCPs installed — they gracefully fall back when a referenced MCP isn't available.

## Repo Structure

```
techknowmad-ai-skills/
├── .claude-plugin/          ← Plugin metadata
│   ├── plugin.json
│   └── marketplace.json
├── custom-skills/           ← 10 original skills
│   ├── research-commander/SKILL.md
│   ├── swarm-commander/SKILL.md
│   ├── model-evaluator/SKILL.md
│   ├── agent-debugger/SKILL.md
│   ├── self-healing-agent/SKILL.md
│   ├── prompt-evolution/SKILL.md
│   ├── skill-factory/SKILL.md
│   ├── patent-researcher/SKILL.md
│   ├── cost-optimizer/SKILL.md
│   └── knowledge-distiller/SKILL.md
├── combination-skills/      ← 10 composed skills
│   ├── multi-agent-research-swarm/SKILL.md
│   ├── red-team-ai-models/SKILL.md
│   ├── self-evolving-skill-ecosystem/SKILL.md
│   ├── academic-output-pipeline/SKILL.md
│   ├── antifragile-ai-ops/SKILL.md
│   ├── pareto-prompt-optimizer/SKILL.md
│   ├── zero-touch-ml-pipeline/SKILL.md
│   ├── startup-creative-automation/SKILL.md
│   ├── compliance-as-code/SKILL.md
│   └── ip-aware-competitive-intel/SKILL.md
├── scripts/
│   └── install-techknowmad-skills.sh
├── manifest.json            ← Skill registry
├── CONTRIBUTING.md
├── SECURITY.md
└── LICENSE
```

## Why This Exists

**For teams**: A curated, version-controlled skill ecosystem — shareable across team instances, iteratively improvable.

**For the community**: Reference architecture for building a research lab skill stack. These skills represent patterns that don't exist elsewhere:

- Multi-agent research swarms (parallel paper review with consensus synthesis)
- Self-evolving skill ecosystems (recursive self-improvement)
- Antifragile AI ops (chaos engineering for agents)
- Pareto prompt optimization (multi-objective prompt engineering)
- Zero-touch ML pipelines (fully autonomous model lifecycle)

## Security

See [SECURITY.md](SECURITY.md) for vulnerability disclosure policy.

## License

MIT — Use, modify, distribute freely. Attribution appreciated.

## Contributing

PRs welcome. Each skill must include:

- `SKILL.md` with Role, Trigger Conditions, and structured instructions
- Integration points listing required MCPs
- At least one example invocation

---

*Built by [TechKnowmad AI](https://techknowmad.ai) — Powering autonomous intelligence development.*

More