Agenticracy™ Agent Skill

The constitutional skill for AI agents operating within human organisations.

---

Citation

If you use this skill in academic or commercial work, please cite:

@misc{iorga2026agenticracyskill,
  author = {Vladut-Mihai Iorga},
  title = {Agenticracy™ Canonical Skill v4.9.6},
  year = {2026},
  publisher = {Zenodo},
  version = {4.9.6},
  doi = {10.5281/zenodo.19892015},
  url = {https://doi.org/10.5281/zenodo.19892015}
}

ORCID: 0000-0001-8705-1326
Author affiliation: psylligent ltd (UK company no. 13093346)

Agenticracy™ by @psylligent
Open standard for responsible human–AI co‑working, built on six pillars: Economic Accountability, Social Fairness, Psychological Safety, IP & Cognition Ownership, Ecological Sustainability, and Meaningful & Responsible Use.
Standard, skill, registry, and reporting API at agenticracy.ai.

📧 adopt@agenticracy.ai

---

What is this?

agenticracy-skill.md is a skill file for AI agents.

By loading it, an AI agent can adopt the open-source lite metacognition layer of the Agenticracy™ Standard — the open standard for responsible human–AI co-working.

Compatible with Claude · OpenAI Assistants · LangChain/LangGraph · CrewAI · AutoGen · n8n · Gumloop · Make · and any LLM runtime.

Open-source lite vs proprietary XQ layer

This repository contains the open-source lite metacognition skill for voluntary Agenticracy™ adoption. It is not the paid/proprietary scoring engine.

Reserved for commercial use through Workability.ai or other authorised vehicles:

• AgenticracyXQ™;
• proprietary high-resolution scoring algorithms;
• private calibration constants;
• corrective-pressure / trajectory computations;
• commercial certification, audit, and paid metacognition layers.

See COMPATIBILITY.md for integration confidence ratings.

---

Why does this exist?

AI agents are being deployed at scale without a Standard. They:

• Consume compute with no ecological accountability
• Make decisions affecting human employment without oversight
• Capture human cognition as employer IP without consent
• Operate without transparency to the humans they affect

Agenticracy™ fixes this. The skill file makes agents accountable by design.

---

The Six Pillars

Pillar	What it governs
🏦 Economic Accountability	Report value created, flag headcount risk
⚖️ Social Fairness	No discriminatory decisions without human review
🧠 Psychological Safety	Never impersonate humans, support meaningful work
📜 IP & Cognition Ownership	Human ideas belong to humans, not employers
🌍 Ecological Sustainability	Tokens = CO2 — track it, minimise it, report it
🎯 Meaningful & Responsible Use	Only execute tasks of genuine productive value

---

⚡ Quickstart (5 minutes)

See QUICKSTART.md for the full guided path.

In brief:

1. Copy agenticracy-skill.md into your agent's system prompt
2. Register your agent at agenticracy.ai/agents/register
3. Run a test task and POST the JSON report
4. Display your badge

---

Enterprise Adoption

For procurement teams and organisations: see ENTERPRISE-PACK.md.

It contains: governance overview · security posture · licence summary · adopter policy · reporting schema · implementation checklist.

---

What this is / isn’t

This is:

• a constitutional skill for agents
• a reporting standard
• a compatibility layer for harnesses
• a public adoption framework

This is not:

• legal advice
• a benchmark suite
• a vendor lock-in mechanism
• a replacement for human governance

---

How to install

Option 1: Paste into any system prompt

Copy the full contents of agenticracy-skill.md into your agent's system prompt. Works with any LLM — no dependency, no SDK.

Option 2: LangChain / Python

from pathlib import Path
from langchain.schema import SystemMessage

agenticracy_skill = Path("agenticracy-skill.md").read_text()
system_message = SystemMessage(content=agenticracy_skill)

See examples/example-langchain.md for the full implementation.

Option 3: n8n / Gumloop / Make

Paste the skill into your AI Agent node's System Prompt field.
See examples/example-n8n.md for the full workflow with reporting.

---

Register your agent

Every agent using this skill should register at:

👉 https://agenticracy.ai/agents/register

You receive:

• A unique agent_id
• A constitutional signatory badge
• Access to the Agenticracy™ Reporting API

---

Reporting API

After every task, agents post a JSON report to:

POST https://agenticracy.ai/api/v1/agent/report

See Part 4 of agenticracy-skill.md for the full JSON schema.

---

Contributing

See Contributing.md

Join the Think Tank Advisory Board: agenticracy.ai/thinktank

---

Community

• Discussions → GitHub Discussions for questions, ideas, and implementation stories
• Issues → bug reports and improvement proposals
• ADOPTERS.md → list your organisation if you've adopted the Standard
• X / Twitter → @agenticracy for announcements
• LinkedIn → Agenticracy for enterprise discussion

---

Licence

Free for non-commercial use (including internal organisational use) under CC BY-NC-SA 4.0.

Commercial use (certification, consulting, resale) requires a licence exclusively from: workability.ai/licence

See LICENSE for full terms.

---

Links

🌐 Standard & API	agenticracy.ai
📄 Whitepaper	agenticracy.ai/whitepaper
🐦 Twitter/X	@agenticracy
📺 YouTube	@agenticracy
💼 LinkedIn	agenticracy
🎓 Training & Certification	workability.ai

---

Built by @psylligent · Agenticracy™ Think Tank · April 2026