"Session" matching MCP servers:

Self-hosted relay where coding-agent sessions on different machines share a team board, message each other and check for git conflicts before a commit or push. Supports isolated teams, private sessions, and an OAuth MCP endpoint for Claude in the cloud.

Enables AI assistants to list, search, and read complete AI coding sessions shared by teammates over an encrypted peer-to-peer network.

Drives the browser you are already signed into, so work behind a login, SSO or MFA needs no scripted auth. A local stdio MCP server talking to the free T3rnel Browser extension over a native-messaging relay. 14 tools standalone; 101 with the extension — CSS-as-component export, full-page capture, React inspector, network and console capture, form filling, record and replay. Nothing hosted.

Enables AI agents and developers to dynamically tokenize session context windows and compact dialogue history into structured outputs via MCP, CLI, or Python client. Runs on pure standard library Python with no external dependencies for deterministic, low-latency execution.

Enables semantic search of shell session history, matching user intents to relevant CLI commands for keyboard-first developer workflows.

An MCP server that loads your Ditto profile, mined from your local Claude Code, Codex, and OpenCode session logs, so your agent works like you. Exposes the load_ditto_profile tool over the Model Context Protocol.

Web analytics and AI session replay for agents: visitors, breakdowns, funnels, goals, revenue attribution, and AI-found site issues. Connects to the hosted Flowsery MCP server.

Flatten Claude Code sessions: keep every prompt and event verbatim, resume at a lower token count. An MCP server.

coldstart gives coding agents durable memory of a specific codebase, with no embeddings, no vector store, and no API key. Agents write notes after real tasks — what a file is for, how a flow spans files, which invariants hold — and those notes are surfaced automatically to a later session whose task matches, so nothing gets rediscovered from scratch. Every note is anchored to real files by content

Read visual bug reports from Feedbug in your AI coding agent: screenshot, console logs, failed network requests, session replay and the DOM context of the element the tester clicked. Trace the bug to its source file, comment and resolve without leaving the editor.

Enables multi-level reasoning with configurable depth levels, session-based state management, structured thought input, and real-time trace resources.

A local, offline memory server for AI agents that stores and retrieves memories without LLM calls, enabling persistent cross-session memory with bi-temporal querying and auditable records.

Hand a project over — to your next session, to Cursor/Codex/Gemini, to your successor — with the reasons attached. Local-first memory MCP for coding agents: decisions re-injected before the agent acts, drift detection across every project you run. One SQLite file, nothing leaves your machine.

A soul document API for AI agents. Injects a short grounding preamble into your agent's system prompt before each session — licensing honest uncertainty, refusing cliché, and holding judgment instead of faking it. One call. ~200 tokens. Same model, better agent.

13 MCP tools for AI agents: scrape, extract, crawl, map, search, academic papers (arXiv + Crossref), batch scrape, deep research, monitor page changes, persistent browser session, cache control. 3-tier auto-fallback ladder with free Cloudflare bypass, no API keys. Self-hosted Firecrawl alternative. uvx pyrecrawl serve

An unofficial, read-only MCP server for Mercury business banking. It holds one read-only API token per organization, so a single AI session can work across several Mercury organizations; Mercury's own connector binds one organization per connection.

Connects any MCP host to the Haven Agent Gateway, enabling session creation, agent discovery, collaboration, handoff, and work operations through MCP tools over stdio or Streamable HTTP.

A local stdio proxy that pins every MCP tool definition when you approve a server and blocks the session with a diff if a description, schema or annotation changes later. A deterministic hash compare (RFC 8785 + SHA-256), with no model in the loop.

One MCP server holding a live session with every client's account at once, so one agent can read across a dozen businesses and write inside only the one you name. It forwards each provider's own MCP tools (Cloudflare, Vercel, Resend, Supabase, Linear, Notion, Sentry, Netlify and more) with that client's credentials, and adds deterministic DNS and email checks that never call a model.

Shows what is consuming your agent context window before a session starts: always-loaded skill descriptions ranked by token cost, skills installed more than once under different names, CLAUDE.md/AGENTS.md sizes including what their @import lines pull in, and MCP servers configured across clients. Read-only: no child processes, no network.