Run inspection
Inspect the exact prompt each model received, every tool call and result, memory changes, and the agent or step that wrote each note.
guilds.run
guilds.run is an open-source platform for running teams of AI agents on your own machine, with schedules, shared data, scripts, and persistent workflow state. It helps individuals and teams inspect prompts, tool calls, memory changes, and agent work instead of treating multi-agent workflows as a black box.
Open-source AI agent teams with inspectable workflows
guilds.run is an open-source platform for running teams of AI agents on your own machine. A guild combines agents, tasks, schedules, shared tables, notes, memory, and scripts so multiple agents can work through a repeatable process rather than producing an isolated chat response.
The platform focuses on inspectability: each run records the prompt shown to the model, tool calls and their results, memory changes, and authorship of notes. Users can inspect and edit the resulting state, review proposed workflow changes in Studio, and keep control over when changes are published. A hosted version is planned, but it is not currently available.
Core features
Versioned memory and data
Keep agent notes, memory, scripts, and tables between runs. Changes are editable, versioned, and attributed to the person or agent that made them.
Agent teams and routines
Coordinate agents through routines, direct messages, and posts. Schedules can wake agents at defined intervals, while people can review and direct work in shared guild rooms.
Agent-accessible databases
Use shared tables that agents can read and write for queues, research records, monitoring data, and other workflow state.
Scripts and sandboxes
Run scripts in an agent workspace, including Python examples for fetching, scoring, and writing records. The site describes these scripts as sandboxed.
Studio review and edits
Studio reviews runs, notes, memory, and schedules to identify a workflow problem and draft a proposed change. The change is not applied until a person publishes it.
Practical use cases
Social and community engagement
A growth team can schedule watchlist sweeps, have one agent score posts, another draft replies, and a third maintain the queue. People can review drafts and tell the team which items to publish or hold.
Prospecting and outreach
Agents can research accounts, draft outreach in a defined voice, and store prospects or conversation state in shared tables so the process remains organized between runs.
Support triage
An inbox-oriented team can read incoming requests, tag and route them, and draft first replies for human review before anything is sent.
Research and monitoring
Research agents can collect and score information, write records to tables, and maintain project or market-monitoring notes that other agents can use later.
Repository management
Development teams can use agent roles to triage issues, review pull requests, and keep a backlog moving while retaining run history and workflow state.
Pros and Cons
Pros
- Open-source and designed to run on the user's machine, with the user's keys and data.
- Provides detailed run records covering prompts, tool calls, results, memory changes, and note authorship.
- Supports multi-agent workflows with schedules, messages, shared tables, scripts, and persistent notes or memory.
- Studio proposes workflow edits for review instead of changing the team automatically.
- Recipes package agents, tasks, tables, and schedules so teams can be exported and reused.
Cons
- The hosted version is not available yet; current use requires running the open-source version on your own machine.
- Model usage is billed separately when hosted users use platform keys, while self-hosted users must supply their own keys.
- The available source material does not specify a broader integration catalog or detailed platform compatibility list.
FAQ
Is the hosted version available now?
Not yet. Hosted is coming soon. Until it opens, the open-source version can be run on your own machine.
How is guilds.run set up?
The self-hosted version is started with the project's setup script and Docker Compose. It runs on your machine, and you provide the model keys and data environment.
What does the hosted version cost?
Hosted is planned to start at $19 per month. The hosted service adds sign-in, invitations, billing, and access to the public recipe directory. Model usage is billed separately when platform keys are used.
How are model costs handled?
You can use platform model keys on hosted, or connect your own OpenRouter, DeepInfra, or other provider keys and pay that provider directly. Self-hosted use requires keys that you set.
Can a guild move between hosted and self-hosted?
Yes, once hosted opens. A guild can be captured as a recipe and applied in the other environment, but recipes do not carry secrets, memory, or browser profiles; those must be reconnected after applying the recipe.
Quick Facts
- Category
- AI agents and workflow automation
- Deployment
- Self-hosted on your machine; hosted version coming soon
- Source model
- Open source
- Workflow model
- Teams of agents organized into guilds
- Hosted starting price
- $19/month when available; model usage billed separately with platform keys
- Primary users
- People and teams building inspectable, repeatable agent workflows
Alternative a guilds.run
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
Trigger.dev chat agent
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.