ejentum.com
@ejentum
The Reasoning Harness Tool that sharpens llm agents. Dynamic & Adaptive Reasoning Injections per task. Engineered and optimized for agentic inference.
12
Published Tools
27
Total Stars
362
Weekly Downloads
14
GitHub Followers
128
Public Repos
Published Tools
6 MCP Servers6 Agentsacross 4 categoriesRAG augments knowledge. Ejentum augments reasoning. 4 harnesses, 679 abilities, runtime injection.
Ejentum
MCP server for Self-Inspect: one keyless tool that returns a single self-inspection metathought for what your agent is currently doing. Calls the public api.ejentum.com/self-inspect endpoint, or runs the published heuristic fully offline with SELF_INSPECT
...moreejentum
Zed editor extension wrapping the Ejentum MCP server. 8 cognitive operations (4 harnesses × dynamic + adaptive) available to Zed's agent.
...moreejentum
LlamaIndex tool spec for the Ejentum Reasoning Harness. Wraps the hosted Ejentum MCP server and exposes 8 tools (4 harnesses × dynamic + adaptive).
...moreejentum
Two-pole terrain maps that counter premature collapse in agent reasoning. An agent states three POVs on its task and gets back one deterministic map to locate which reading it collapsed onto. Keyless, no LLM. MCP server + zero-dependency Python + HTTP endpoint.
...moreEjentum <[email protected]>
CrewAI tool for the Ejentum Reasoning Harness. Each call retrieves a task-matched cognitive operation engineered in two layers: a natural-language procedure plus an executable reasoning topology (graph DAG with gates, parallel branches, and meta-cognitive exits). Injected before the LLM step to harden reasoning against decay on complex tasks and long agent loops.
...moreEjentum <[email protected]>
Agno Toolkit for the Ejentum Reasoning Harness. Each call retrieves a task-matched cognitive operation engineered in two layers: a natural-language procedure plus an executable reasoning topology (graph DAG with gates, parallel branches, and meta-cognitive exits). Injected before the LLM step to harden reasoning against decay on complex tasks and long agent loops.
...moreEjentum <[email protected]>
AutoGen tools for the Ejentum Reasoning Harness. Four agent-callable async functions (harness_reasoning, harness_code, harness_anti_deception, harness_memory) returned by ejentum_tools(api_key=...). Each call retrieves a task-matched cognitive operation engineered in two layers: a natural-language procedure plus an executable reasoning topology.
...moreEjentum <[email protected]>
LangChain integration for the Ejentum Reasoning Harness. Four BaseTool subclasses (reasoning, code, anti-deception, memory) an agent calls before generating. Each call returns a task-matched cognitive operation engineered in two layers: a natural-language procedure plus an executable reasoning topology (graph DAG with gates, parallel branches, and meta-cognitive exits).
...moreEjentum <[email protected]>
smolagents Tool subclasses for the Ejentum Reasoning Harness. Eight tools: four dynamic (reasoning, code, anti-deception, memory) and four adaptive (adaptive-reasoning, adaptive-code, adaptive-anti-deception, adaptive-memory) that pre-fit the operation to the task via an adapter LLM. Each call returns a structured cognitive injection: a natural-language procedure plus an executable reasoning topology.
...moreLlamaIndex tool spec for the Ejentum Reasoning Harness. Wraps the hosted Ejentum MCP server and exposes eight cognitive harness tools as LlamaIndex FunctionTool objects: four dynamic (reasoning, code, anti-deception, memory) plus four adaptive variants (adaptive-reasoning, adaptive-code, adaptive-anti-deception, adaptive-memory) that pre-fit the cognitive operation to the caller's task via an adapter LLM.
...moreEjentum <[email protected]>
PydanticAI toolset for the Ejentum Reasoning Harness. EjentumToolset subclasses pydantic_ai.FunctionToolset and registers eight tools: four dynamic (reasoning, code, anti-deception, memory) plus four adaptive variants (adaptive-reasoning, adaptive-code, adaptive-anti-deception, adaptive-memory) that pre-fit the cognitive operation to the caller's task via an adapter LLM.
...more