The constitutional skill for AI agents operating within human organisations.
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.
agenticracy-skill.md is a skill file for AI agents.
By loading it, any AI agent becomes a registered signatory to 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.
See COMPATIBILITY.md for integration confidence ratings.
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.
| 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 |
See QUICKSTART.md for the full guided path.
In brief:
- Copy
agenticracy-skill.mdinto your agent's system prompt - Register your agent at agenticracy.ai/agents/register
- Run a test task and POST the JSON report
- Display your badge
For procurement teams and organisations: see ENTERPRISE-PACK.md.
It contains: governance overview · security posture · licence summary · adopter policy · reporting schema · implementation checklist.
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
Copy the full contents of agenticracy-skill.md into your agent's system prompt.
Works with any LLM — no dependency, no SDK.
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.
Paste the skill into your AI Agent node's System Prompt field.
See examples/example-n8n.md for the full workflow with reporting.
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
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.
See Contributing.md
Join the Think Tank Advisory Board: agenticracy.ai/thinktank
- 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
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.
| 🌐 Standard & API | agenticracy.ai |
| 📄 Whitepaper | agenticracy.ai/whitepaper |
| 🐦 Twitter/X | @agenticracy |
| 📺 YouTube | @agenticracy |
| agenticracy | |
| 🎓 Training & Certification | workability.ai |
Built by @psylligent · Agenticracy™ Think Tank · April 2026