vibeframe
Enables agentic coding tools to maintain a living, in-repo model of intent, architecture, and code, with deterministic indexing, bidirectional linking, and continuous drift detection for task-scoped context and reconciliation.
README
VibeFrame
A control plane for agentic coding. VibeFrame maintains a living, in-repo model of what your software is supposed to be — business intent, system design, and the code itself — keeps it deterministically synchronized with reality, and shows you where they diverge.
Coding agents write code faster than anyone can read it. What evaporates is the theory: what exists, why it exists, which decisions are load-bearing, and whether the code still matches what was decided. VibeFrame captures that theory at the moment it is cheapest to capture (planning time, through the agent's own MCP protocol), stores it as plain files in your repo, verifies it mechanically forever, and pays the agent back with task-scoped context briefs that beat re-grepping the codebase.
- Three planes, one graph — capabilities & workflows (intent), components & constraints (system), files & symbols (code, 100% derived — never guessed).
- Trust tiers on every fact —
Aderived /Benriched /Casserted; staleness is visible, never silent. - Drift detection with evidence — a live reflexion model: divergence, absence, broken bindings, orphans, cycles, locked-decision touches. Zero LLM calls; the framework is deterministic and the coding agent is the intelligence.
- In-repo model —
.vibeframe/YAML/Markdown versioned by git: branch-safe, PR-reviewable, survives any tool. A gitignored SQLite index is derived and rebuildable. - One daemon per project — Streamable-HTTP MCP endpoint, REST + SSE, and a live browser UI (Map, Focus, Inspector, Activity, Drift inbox, Decisions) in a single process.
- Claude Code first — a plugin bundles the MCP server, deterministic hooks, and skills; the model also serves any agent through an AGENTS.md digest.
Install
npm i -D vibe-frame-cli
Node ≥ 22.12. This installs the vibeframe command and everything it needs.
(Careful: the unhyphenated vibeframe on npm is an unrelated project — don't install
that. Tarballs are also attached to each GitHub release
if you prefer registry-free installs.)
Quickstart
npx vibeframe init # scaffold .vibeframe/ — commit it
claude mcp add vibeframe -- npx vibeframe mcp # or install the Claude Code plugin (below)
npx vibeframe start # daemon + browser UI (prints the URL)
For the full Claude Code experience (MCP plus deterministic hooks and the
/vibeframe:* commands), install the plugin from this repository inside Claude Code —
/plugin marketplace add jubayertalha/vibeframe then /plugin install vibeframe — and
skip the claude mcp add line above (the plugin brings the server).
Then ask your agent to plan a feature. It orients with one vibe_context call, declares
the workflow before coding, binds implementations after — and everything it forgets, the
deterministic layer catches: violations come back as same-turn warnings, unexplained
changes queue for reconciliation, and the drift inbox shows exactly where reality left
the plan, with file:line evidence. Full walkthrough:
docs/getting-started.md.
Documentation
| Getting started | install → first feature → the safety net |
| Architecture | the daemon, ground-truth asymmetry, package boundaries |
The .vibeframe/ format |
capabilities, workflows, components, constraints, decisions |
| Schema versioning | the three version domains and the migration contract |
| Validation & drift | every check, its gradations, and its documented false negatives |
| MCP interface | the seven vibe_* tools, resources, prompts |
| Claude Code integration | plugin, hooks, AGENTS.md digest |
| CLI · Browser UI | command and lens reference |
| Contributing · Security | working on VibeFrame; the trust model |
Status
v0.1.0 — the complete loop on greenfield TypeScript/JavaScript/Python projects: init → declare → index → link → validate → reconcile, with the daemon, MCP tools, hooks, and the six-lens UI. Deferred by design (see the Discovery): other-agent adapters, brownfield adopt-mode, symbol-level call graphs, teams, anything hosted.
Development
npm ci && npm run build && npm run check # lint + typecheck + unit tests
npm run test:dogfood # the full-loop acceptance suite
See docs/contributing/development.md and CLAUDE.md.
License
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.