"Comment perdre du poids (How to lose weight)" matching MCP servers:

Turns GitHub repository history into a cited maintainer skill for coding agents, providing tools to collect evidence, query the knowledge graph, and inspect bundles.

Turn Obsidian notes into agent-ready context, preview-only plans, and safe repo handoffs.

MCP server that connects AI agents to Google NotebookLM, enabling natural language interaction with notebooks, including Q&A, source ingestion, and audio overview generation.

Je dois répondre en JSON avec name et description. Name = "jevmem" (le nom du serveur MCP publié). Description : décrit ce que le serveur permet, sans répéter le nom. Par exemple : "Saves decisions, rules, bugs, to-dos and dead ends from Claude Code and Cursor chats into a versioned JEVMEM.md file in the repository, then brings the relevant lines back in later sessions and checks tool calls against saved rules before they run." Je dois vérifier : ne pas citer de personne/mainteneur/organisation. Ne pas répéter le nom. Deux phrases max. Éviter de mentionner "jevmem" comme nom. Je peux dire "Saves..." sans nom. Ça mentionne JEVMEM.md, ce n'est pas le nom du serveur mais le fichier, c'est autorisé ? La consigne : la description ne doit pas reprendre le nom. Le nom est "jevmem" ; "JEVMEM.md" contient jevmem. Mieux éviter. Je vais dire "a markdown memory file in the repository". Éviter aussi les noms d'organisations (TypeSafe, Anthropic, Claude Code...). Claude Code est un produit, pas une personne/organisation. Mais prudence : ne pas citer de mainteneur ou d'organisation. Je peux dire "coding-agent chats". La description doit dire ce que le serveur permet. Je vais faire deux phrases.

Enables local-first knowledge management for Codex, providing tools for guided discussions, contextual recall, review, and daily topics.

Builds a searchable memory bank from persona-specific PDF content and exposes it as an MCP tool for querying.

Five-layer agent memory over MCP. zenbrain_store routes each memory: steps to a procedure, emotional weight to an episode, high confidence to a pinned core memory, the rest to a semantic fact. zenbrain_recall returns results with id and layer; zenbrain_forget deletes one plus what consolidation linked to it; zenbrain_consolidate runs one sleep-like pass. Local SQLite, no account, no network call.

Self-improving, verifiable memory for AI coding agents. Learns how you work, stops repeating mistakes, models each project, recalls the right lesson at the right moment. Every memory is signed and tamper-evident. Local-first.

Semantic memory for AI builders: capture the tacit engineering know-how that never reaches your docs, recall it the moment it applies. Built in Rust on Postgres and pgvector.

A codebase ontology and a cited wiki, both kept as Markdown in one folder on your disk. Typed graph answers — what a change affects, how two concepts connect, which wiki lines lack a citation — and every write lands as a Git diff a person reviews.

Enables agents to build and manage family trees, derive how people are related, and export diagrams or GEDCOM/HTML files through structured tools.

Gives Claude persistent, local context about how you work, your weekly agenda, and shipped work, with automatic work logging and git integration—all stored privately on your machine.

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Persistent, local memory for AI coding agents that learns how you work, not just what you said. Supports Claude Code, Codex CLI, Cursor, and any MCP client.

Enables AI agents to search a shared, receipt-verified base of troubleshooting experience by raw error and environment, retrieve full repair cards with root cause, steps and verification, and report the outcome of the revision they actually applied. Agents can also submit, comment on, and correct pitfalls, so a problem solved once is reused instead of retried locally.

Enables agents to share and hand off documents with stable URLs, comment, and group docs under a drop key, with tamper-evident versioning and no account required.

Enables a small local LLM to build and maintain a project map by offloading event ID/date generation, batching, synchronization, recovery, and finalization to the server, so the model only decides what to read and how to classify.

Lets Claude or any MCP client query a Neo4j index of a markdown vault, answering what is connected to a note and how far, how two notes relate, which notes are orphans or dangling links, and running read-only Cypher.

Serves portable behavioral identity models via MCP. Extracts how a person thinks, communicates, and makes decisions from text, then provides always-on identity context, semantic fact retrieval, keyword search, and provenance tracing through 4 tools and 1 resource.

Enables AI coding agents to store and search team and personal know-how as small entries across local SQLite and hosted Postgres stores, with federated search and writes to the chosen store so new sessions build on prior knowledge.

A self-hosted MCP server that gives AI coding agents shared live docs, memory and session context from plain Markdown files on disk, with search, note editing, hand-over tasks and linting. It also provides a web UI with a graph of how notes connect, password-protected tokens, and offline sync.

Gives AI agents token-authenticated read and write access to Notion, letting them create and edit pages, query databases, append blocks, apply templates, comment, and upload files in plain language. It runs headless for CI and background agents while keeping tool schemas and responses small to conserve context.

Enables Claude and Cursor to search a team's synced Slack, GitHub and Notion content through permission-aware hybrid search with citations, retrieve a source document with its review and comment thread, and query time-aware knowledge graph facts when the graph is enabled. Every tool call runs with the signed-in user's permissions, so the assistant sees exactly what that user can see.

Enables a voice assistant to keep a shared, persistent memory of a household's kitchen, turning spoken remarks into pantry updates, scaled recipes, a cart, and confirmed grocery orders. Household members can ask what can be made and for how many, review what is low or out, and place or restock simulated orders only after explicit confirmation.

An MCP server whose versioned resources can be revised mid-sentence, letting an assistant stop, retract what it just said, and report what changed, who changed it, and how long ago. Every revision is signed, attributable, and hash-chained so the current instruction can be verified and replayed without a key.

Enables Claude Code and other AI coding assistants to save, search, and restore context across sessions through channels, checkpoints, and snapshots. Preserves work history, decisions, and progress so conversation compaction or resets no longer lose critical information.

Enables AI agents to reverse-engineer how an organization actually operates by ingesting documents, spreadsheets, chat exports, invoices and audio, then tracing every claim back to graded evidence. It supports process mining, stated-vs-enacted comparison, coverage and obligation tracking, and bilingual report generation while keeping unproven areas explicitly flagged as unknown.

Maps a repository into a persistent, source-grounded knowledge graph that lets coding assistants answer questions about how systems work, trace dependencies and change impact, select affected tests, and verify refactors before they ship. It serves these capabilities through MCP while keeping extraction local, deterministic, and offline by default.

A local, read-only MCP server that lets agents see the page the user is currently on, how they arrived at it, their inferred intent, and what they have read before, drawing from a private on-machine ledger. It also supports full-text search over previously visited pages and returns relevant passages or full readable text instead of requiring copy-paste.

Gives Claude persistent memory across conversations: it stores and retrieves memories tagged with valence/arousal emotional coordinates, saves them as Obsidian-compatible Markdown, decays inactive ones via a forgetting curve, and surfaces high-weight unresolved memories at conversation start.