Shared project archive
Sessions from team members are collected in one archive associated with a project, allowing agents to access prior project work.
ResumeContext est un service de mémoire partagée pour les équipes d’agents de programmation. Il enregistre les sessions en local, les synchronise avec une archive de projet et la met à disposition des agents via MCP, afin qu’ils puissent retrouver les travaux antérieurs et poursuivre entre agents, comptes ou machines.
ResumeContext is a shared memory service for coding-agent teams. A local daemon reads the session history that supported coding agents already write to disk, syncs those sessions to a shared project archive, and makes the archive available to agents through the Model Context Protocol (MCP). The product is designed to preserve project decisions, investigations, and unfinished work beyond an individual agent session.
Its main workflows are cross-agent continuity and shared team recall. An agent can search previous sessions and return answers with references to the originating session and line. Because the archive is associated with a project rather than a particular agent or account, a developer can switch tools, accounts, or machines without starting the task context over.
Sessions from team members are collected in one archive associated with a project, allowing agents to access prior project work.
The archive is provided back to coding agents as an MCP server, so users can query shared history from the agent session they are already using.
Responses can identify the source session and relevant line range, making it possible to trace an answer back to the recorded work.
A small daemon reads agent history from local disk and syncs it without requiring an installation inside the agent or changes to prompting.
The `resumecontext agents` command can list supported agents, inspect their history locations, and choose which agent histories to sync, including custom paths.
Months of sessions remain searchable, with the agent retrieving relevant passages instead of requiring all history to remain in the active context window.
Point another agent at the same project archive when the current tool reaches its limit, allowing work to continue without manually reconstructing the previous session.
A teammate can ask what another developer discovered about a bug and receive an answer grounded in the recorded investigation, including session and line references.
Project members can ask what changed, who is blocked, or whether work is overlapping, using the team’s recorded sessions instead of separate status updates.
The project archive supports continuity when work moves between agents, accounts, or machines, provided those environments are connected to the same project.
The product reads session history from supported coding agents. Its agent-management command can list supported agents, show their normal history directories, report whether history is present locally, and configure a custom history path where needed. The supplied documentation examples include Claude Code, Cursor, and Codex, but the complete supported-agent list should be checked in the current documentation.
No. ResumeContext reads the history agents already write to disk and returns the shared archive through MCP. The documentation states that nothing is installed into the agent and prompting does not need to change.
The quickstart describes a setup of about two minutes: install the CLI, sign in, run `resumecontext init` in the project directory, and connect the coding agent. After setup, it syncs automatically. The setup is performed on each machine that will contribute or read project history.
History is stored on ResumeContext’s servers within the project it was synced to and is readable by members of that project. The pricing page states that agent access is read-only and enforced by a database role without access to other data.
A teammate can be removed from the project page, and their access ends immediately. Sessions they already recorded remain in the archive and retain attribution to that teammate.
Les données de trafic sont fournies à titre indicatif uniquement.
superqode.dev
SuperQode est une couche de harness open source pour les agents de programmation. Elle permet de découvrir, créer, exécuter, évaluer et optimiser des harnesses d’agents depuis un seul terminal. Elle assure l’interopérabilité entre A2A, ACP, MCP et UHP, afin que les équipes puissent conserver leurs agents de programmation et modèles existants.
valpere.github.io
session-indexer est un outil pour développeurs hébergé sur GitHub qui indexe les transcriptions de sessions Claude Code dans SQLite et permet de les rechercher par sens. Il est conçu pour retrouver l’historique JSONL par projet, avec FTS5 en solution de repli lorsque la recherche basée sur des embeddings n’est pas disponible.
build.reflex.dev
Reflex Build transforme les prompts en applications Python déployables sur Reflex Cloud, votre cloud ou en local, avec des modèles prêts pour la production et des intégrations pour les agents et outils de programmation.
soloterm.com
Solo est un espace de travail de bureau pour exécuter des agents de programmation IA avec votre environnement de développement local.
www.cybedefend.com
CybeDefend est une plateforme de sécurité du code conçue pour les workflows de programmation avec l’IA. Sa couche VibeDefend examine les prompts des agents et les diffs, applique les règles du projet et de l’entreprise, protège contre les actions risquées et regroupe les analyses du code, des dépendances, des secrets, de l’infrastructure et du CI/CD avant une pull request.
million.dev
Vérification pour agents de programmation IA afin de prouver que le code fonctionne