Podiom
A local-first workspace that gives Claude Code and Codex agents durable names, persistent memory, schedules, and long-term goals to work on.
🔗 Visit Podiom
Description
Running an AI coding assistant in a terminal usually means each session starts from a blank slate, you lose context, scheduled work, and history the moment you close the window. Podiom is built to fix that: it turns your local Claude Code or Codex agents into persistent 'team members' with their own identity, memory, and ongoing responsibilities, instead of disposable one-off sessions.
Podiom is an open-source, local-first workspace that orchestrates Claude Code and OpenAI Codex agents, giving each one a durable name and persistent context across sessions. You can hand an agent a one-time task, a recurring scheduled job, or a long-term 'goal' that it plans and executes autonomously over time, with every session, decision, and tool call remaining fully auditable. It runs cross-platform (macOS, Linux, Windows), supports Home Assistant and container deployment, and even ships native iOS/Android apps to check in on agent progress remotely.
💬 Our review
The short version: Podiom addresses a real annoyance, AI coding agents forgetting everything between sessions, with a thoughtful local-first design, worth trying if you already run Claude Code or Codex locally and want persistence instead of disposable terminal sessions.
Most people today just re-explain context to their agent every session or build ad-hoc note files to fake persistence; Podiom's structured approach, named agents, durable state, scheduled jobs, and autonomous goals with a full audit trail, is a more deliberate answer to the same problem. Its local-first design (data stays on your machine, configurable root) is a genuine plus for anyone wary of hosted agent runtimes holding their project context. At 9 stars it's very early, so expect some rough edges, but the mobile apps for remote check-ins are a nice, unusual touch for a project this young.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Gratuit, open-source — coût des abonnements Claude Code/Codex à part
Pros
agents avec identité durable et mémoire persistante entre sessions
gère tâches ponctuelles, jobs récurrents et objectifs autonomes long-terme
audit trail complet des sessions, décisions et appels d'outils
local-first, données restent sur votre machine
apps mobiles natives iOS/Android pour suivre les agents à distance
Cons
projet très jeune (9 stars)
nécessite déjà d'utiliser Claude Code ou Codex localement
gestion de goals autonomes à surveiller de près au début