Claude Local Bridge Playground banner
alankatanoisi alankatanoisi

Claude Local Bridge Playground

Development community

Description

Playground harness experiments for claude-local-bridge

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

Claude Local Bridge Playground

**This is the active repository** ([claude-local-bridge-playground](https://github.com/alankatanoisi/claude-local-bridge-playground), branch **`main`**). Canonical [claude-local-bridge](https://github.com/alankatanoisi/claude-local-bridge) is archived (tags `archive-2026-05-main` and `archive-2026-05-runner-clean-pr`); do not open new PRs there.

This repo is now the active lab for the **cc bridge runner**: a small local coding-agent loop that uses the bridge as its model transport, then focuses on prompts, tools, permissions, transcripts, archives, and command-line ergonomics.

The bridge still matters. It exposes a local Anthropic Messages API on `http://localhost:11437` and injects Claude Code OAuth credentials so the runner can call models without an Anthropic Console API key. But for new work in this repo, treat bridge/OAuth/interceptor code as plumbing and treat the runner as the product surface.

Who this repo serves

This is Alan's personal, long-horizon research playground. The deliverable is understanding — questions answered, systems mapped, ideas tested — never shipped software: nothing here is headed to production, a release, or a customer, and prototypes are disposable by design. Alan is a strong systems thinker and a deliberate novice at programming, so agents and docs here must not assume standard-developer background, conventions, or goals. The full agent-facing owner profile is [`docs/working-with-alan.md`](./docs/working-with-alan.md) (browser version: [`docs/working-with-alan.html`](./docs/working-with-alan.html)).

Current Direction: Minimal Runner Lab

The goal is to build, optimize, and customize a compact local runner inspired by minimalist-but-extensible agent design philosophies such as pi:

  • Keep the default prompt small and repo-agnostic.
  • Keep startup context minimal until a flag, profile, or template asks for more.
  • Keep the tool surface understandable by grouping capabilities: read, write, r