Back to directory

J-Space Cognition Suite

A skill package of thirteen modules that gives a coding agent a structured, evidence-backed workflow for deep reasoning, long tasks, verification, and recovery.

J-Space Cognition Suite

J-Space Cognition Suite is an inference-time skill package for AI coding agents, aimed at complex reasoning, repository engineering, coordinating multiple agents, and authorized security analysis. It is installed as a single skill and organizes thirteen modules around one shared routing entry point, so a host agent loads only the modules a task needs.

Its controllers are plain Python scripts that persist state across a long task, reread the actual source files rather than trusting a cached summary, detect when a map or piece of evidence has gone stale, and block progress when a required condition is missing. It needs Python 3.10 or newer for its executable checks, though lower-effort agents can fall back to following the documented instructions in prose. It is based on and references Anthropic's J-space global-workspace research.

It is released under the Apache-2.0 license.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.