Back to directory

Emulo

A tool that mines a developer's own past messages to their coding agents into a personal profile of working rules, each backed by a dated quote from an actual session, then installs that profile where the agent can read it.

Emulo

Emulo reads through a developer's existing session logs from Claude Code, Codex, Copilot CLI, OpenCode and Google Antigravity to extract rules the developer never wrote down explicitly, such as what counts as done, what gets rejected on sight, and how the developer debugs, designs interfaces or writes when actually working. Every extracted rule keeps a dated quote from the developer's own messages as its evidence, and the mining step writes separate profile files for work, design, writing and video so a host loads only the one relevant to its task.

Whether a host actually reads an installed profile varies: in the project's own host-by-host test, the profile reached the model through a project AGENTS.md file in Codex and was loaded into OpenCode's instructions, while in Claude Code the corresponding skill installs but is not reliably opened without being explicitly invoked; support in several other hosts is stated as unverified. The project describes its own evaluation methodology as not yet executed pending a separately approved evidence release.

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.