Official integration guides and skill files for the Ejentum Logic API — a runtime reasoning harness for AI agents.
Your agent POSTs a task description and a mode. The API returns a structured cognitive injection (failure pattern to avoid, procedure to follow, verification criterion, suppress/amplify signals). Your agent absorbs the injection and writes from it. 679 engineered abilities across four modes: reasoning, code, anti-deception, memory.
| Folder | What it is | Install time |
|---|---|---|
/n8n |
One HTTP Request tool + pre-built workflow JSON. Agent auto-routes to the right mode via $fromAI(). |
3 min |
/claude-code |
Five native Claude Code skill files (ejentum-unified, ejentum-reasoning, ejentum-code, ejentum-anti-deception, ejentum-memory). Drop into ~/.claude/skills/, set one env var, done. |
5 min |
Each subfolder has a README with picture-by-picture install walkthrough and downloadable assets.
| Framework | Docs |
|---|---|
| LangChain / LangGraph | https://ejentum.com/docs/integrations#langchain |
| CrewAI | https://ejentum.com/docs/integrations#crewai |
| Cursor / Windsurf / Codex | https://ejentum.com/docs/integrations#agentic-ides |
| Make.com | https://ejentum.com/docs/integrations#makecom |
| Any framework (universal pattern) | https://ejentum.com/docs/integrations#universal-pattern |
If you want an integration that isn't here yet, open an issue.
Free tier: 100 calls, no card. Generate at ejentum.com/dashboard.
Paid: Ki €19/mo (5,000 calls), Haki €49/mo (10,000 calls + multi-mode). See pricing.
- Homepage: https://ejentum.com (five benchmark pills across all four harnesses)
- Full methodology: https://ejentum.com/docs/benchmarks
- "Under Pressure" research paper: Zenodo · SSRN
MIT. See LICENSE.
PRs welcome on any integration folder. For additions (new framework), open an issue first so we can scope it together.
For questions about the Ejentum API itself: info@ejentum.com.