Back to directory

Helix

A self-healing runtime that wraps an AI agent's function calls, diagnoses failures such as a rejected payment, tries a fix, and remembers the fix so the same error resolves instantly next time.

Helix

Helix wraps a function call, such as an agent sending a payment, in a pipeline that perceives what broke when it fails, constructs candidate fixes, scores them, commits and executes the best one, verifies whether it worked, and stores a working fix in a SQLite-backed knowledge base scored by reinforcement learning. The next time the same error occurs for any agent, the stored fix applies without another diagnosis or model call.

It reports a 99.9% success rate against a blind-retry baseline of 81.9% across a set of Base Mainnet transactions, and separately reports that several frontier models all failed on a bare execution reverted error while its pipeline resolved it consistently. It installs as an npm package, offers observe, auto and full operating modes with increasing levels of autonomy, and is built as the first product on top of a broader VialOS agent runtime.

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.