Back to directory

The Pipeline Framework

A Java framework for strongly typed application flows that combine model decisions, authorized SaaS capabilities and deterministic business logic, generating the surrounding transport, persistence, retries, durable effects and observability.

The Pipeline Framework

The Pipeline Framework, or TPF, is a Java framework for building strongly typed application flows that compose model decisions, authorized SaaS capabilities, deterministic business logic and long-running work into one application model. It generates and runs the surrounding shell that application code would otherwise have to write by hand: transport, connections, persistence, retries, durable effects, waits, replay, telemetry and deployment integration.

A model call is treated as one ordinary, provider-backed step that returns a typed value or an inert proposal from a release-pinned catalogue of allowed operations, rather than being given ambient access to connectors, credentials or effects; a separately generated boundary validates and invokes at most one approved operation afterward. An agent-style loop is then built as ordinary composition of these typed steps, letting the application decide where it reads state, performs an effect, asks for approval, or waits for a later event, and connected SaaS applications are built from selected MCP tools, persisted GraphQL operations, OpenAPI contracts and host-owned OAuth connections.

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.