developer kit. zero LLM, local-first, multi-vault. one command from git clone to working dev environment.

Context engine for Pi — zero-LLM compaction, session continuity, sandbox execution, and tool display optimization

Headroom context compression plugin for OpenClaw — 70-90% token savings with zero LLM calls

Deterministic metrics from your local Claude Code / Codex agent logs — parallelism, flow, throughput, projects. Zero LLM, zero server, on-device.

Deterministic YAML flow runner for tapflow — replay UI test flows against simulators/emulators with zero LLM calls

Deterministic AI coding harness-maturity scanner for AI-assisted repositories (Cursor-flagship, expanding to other AI-first dev tools). Zero LLM calls, zero network, zero runtime dependencies.

Local zero-LLM PR QA designer that maps commit intent and diffs to traceable behavior lifecycles, QA scenarios, and optional automation drafts.

Enforceable, zero-LLM-cost repo self-documentation: an annotated directory tree with a cross-ecosystem dependency graph.

A navigation memory for AI agents — zero-LLM web-navigation CLI that recalls cheap, reliable routes instead of re-exploring.

Deterministic, install-free structural extraction for TypeScript, Python, Ruby, Elixir, PHP, Kotlin, Dart/Flutter, and Swift codebases — AST → communities → god-nodes → framework & infra adapters, zero LLM/DB/network.

SMA (Shared Memory & Automation) - the accountability layer for AI coding agents: layered memory, multi-terminal coordination without a server, and verified claims. Plain files + git, zero LLM in the hot path. Install: npx -y sma-framework@latest init

Test that a bot only ever does the job it was given — and never deletes, sends or approves the thing it must not. Record it doing the job once, then replay on every commit with zero LLM calls. For the agents and RPA already running real work in production

Local, zero-LLM, evidence-first website audit CLI for AI-assisted landing-page development.

3–5 tools instead of 1,800. ≤65ms. Zero LLM calls in the router.

Zero-LLM-dep cryptographic evidence primitives for AI decision attestation and agent session recording. Ed25519 signing, hash chain, batch attestation, and streaming session bundle API. Independently verifiable offline.

MCP server - offline EU compliance corpus (EUR-Lex full text via local SQLite FTS5, verbatim zero-LLM) for GDPR, AI Act, DORA, NIS2, eIDAS 2.0, CRA

Process-scoped oh-my-opencode-slim preset switcher for OpenCode: TUI picker (/omopreset) + server-side model rewrite, zero LLM turns.

Vibehub workbench core — library-first, zero-LLM (decision-project-013/025). All real logic lives here; cli/mcp are thin adapters.

Add a referee to your coding loop — a deterministic, zero-LLM CLI that checks every change earned its win: it re-proves the result from the agent's source edits alone, so a green that came from editing the test reads GAMED. Runs in CI or a Claude Code hoo

Native TypeScript AI memory system and MCP server (Zero-LLM, LanceDB)