Back to directory

Uni-CLI

A single command-line tool that turns a plain-language request into a selected operation across APIs, browsers, desktops, local tools and MCP servers, choosing the strongest available interface and returning a structured result.

Uni-CLI

Uni-CLI maintains a catalog of discoverable operations and their contracts, then routes an intent to the strongest operator with the smallest effective scope: a structured API when a stable service contract exists, a local runtime for files and system state, a browser protocol for an authenticated web session, a semantic browser using DOM and Chrome DevTools Protocol semantics for a page-only flow, an accessibility layer for native desktop applications, and pixel-based visual computer use only when no stronger interface is available. Only one provider actually runs for a given task, and a failed path keeps its original cause along with a suggested repair command.

Example commands search for an operation, list what a specific site supports, check the health of the browser delivery path, and attempt a repair when something that used to work has drifted. It installs as a global npm package, and persistent non-interactive installations can update themselves in a detached background worker, reporting progress through the tool's structured output, while interactive terminals keep an explicit yes-or-no upgrade prompt.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.