Everything you need to integrate Ejentum's Ejentum API into your agent: from a 30-second curl call to the theoretical foundations behind suppression-driven reasoning.
These are the same documents rendered at ejentum.com/docs. Read them here or on the website: same content, your choice.
Most AI tools give agents more facts to work with (documents, data, context). Ejentum harnesses the reasoning power agents already have: structured injections that channel the model's existing capability into disciplined execution, preventing the shortcuts and decay that degrade output over long chains.
RA²R (Reasoning Ability-Augmented Retrieval) retrieves reasoning abilities. A structured injection that governs how the agent thinks: what to focus on, what failure modes to block, and how to verify its own output. Where RAG retrieves facts, RA²R retrieves ways of thinking.
679 abilities across four harnesses. One REST endpoint. No SDK.
| I want to... | Read |
|---|---|
| Try it in 30 seconds | Quickstart |
| Test whether it works on my own tasks | Evaluate |
| Understand the mechanism | Concepts, then The Method |
| See actual injection payloads | Injection Examples |
| See before/after output quality | Response Examples |
| Integrate with my framework | Integrations |
| Set up an agentic IDE (Cursor, Windsurf, Claude Code) | Skill Files |
| Read the request/response schema | API Reference, then Architecture |
| See benchmark results | Benchmarks |
| Understand one specific product layer | Reasoning · Code · Anti-Deception · Memory |
| Look up a specific term | Glossary · FAQ |
| Document | Description |
|---|---|
| Quickstart | Add a cognitive injection to your agent. One API call. No SDK. |
| Evaluate | Measure whether Ejentum improves your agent. One afternoon. |
| Concepts | How the Ejentum API works. Four product layers. Why suppression matters. |
| Document | Description |
|---|---|
| Reasoning Harness | 311 abilities across 6 cognitive dimensions. The original product. |
| Anti-Deception Harness | 139 abilities. Blocks sycophancy, hallucination, prompt injection, social engineering. |
| Code Harness | 128 abilities across 13 engineering disciplines. Code generation, refactoring, architecture. |
| Memory Harness | 101 abilities. Perception sharpening, state tracking, behavioral calibration. |
| Document | Description |
|---|---|
| API Reference | Complete technical specification for the Ejentum API v1. |
| Integrations | n8n, LangChain, CrewAI, Claude SDK, Cursor, Make.com. |
| Injection Examples | Real, complete injection payloads for all 7 modes. |
| Response Examples | Real before/after outputs from blind benchmarks. |
| Agent Tool Guide | Developer reference with code examples. |
| Builder's Playbook | 28 screenshots from real work sessions. |
| Document | Description |
|---|---|
| Ejentum Skill File (all modes) | Autonomous routing across all 4 harnesses. Drop-in for any agent. |
| Reasoning Skill File | 311 abilities, 6 dimensions, blind-evaluated. |
| Code Skill File | 128 abilities, 13 disciplines, blind-evaluated. |
| Anti-Deception Skill File | 139 abilities, 6 domains, blind-evaluated. |
| Memory Skill File | 101 abilities, two-pass protocol, blind-evaluated. |
For MCP-speaking clients (Claude Code, Cursor, Cline, Windsurf, Continue), the same harnesses are also available as a packaged MCP server: github.com/ejentum/ejentum-mcp. Install with npx -y ejentum-mcp plus EJENTUM_API_KEY.
| Document | Description |
|---|---|
| The Method | Theoretical foundations. Failure dimensions. Suppression asymmetry. |
| Benchmarks | Eight benchmark suites across four product layers. Full results. |
| Architecture | Request lifecycle. What the API returns. How to inject. |
| Document | Description |
|---|---|
| FAQ | Straight answers. Compatibility, reliability, pricing, usage. |
| Glossary | Key terms: Negative Gate, Reasoning Topology, Injection, Harness, Ability. |
| Changelog | API version history and stability guarantees. |
- Product:ejentum.com
- Benchmarks:github.com/ejentum/benchmarks
- Examples:github.com/ejentum/examples
- Blog:ejentum.com/blog
Documentation is released under CC BY 4.0.