Gantry Ledger
Ledger keeps state, evidence, and decisions together, reviews changes, and suggests next steps. Your team can keep building without losing why each decision was made.
open source
One
shared
history.
Software and hardware, in one place.
Gantry has two independent services on a shared foundation. Ledger keeps the history of work and reviews changes. Crew lets persistent agents plan, consult, test, and decide within a scope you approve.
Your agent prepares the setup. You review access before it connects.
ROBOTS BUILDING ROBOTS. scroll to explore
FIG. 01 / ONE SOURCE, MANY PATHS
01 / THE WORLD WE WANTROBOTS BUILDING ROBOTS
A world where
robots build robots.
Robots and agents see what work is needed. They define what a robot must do, find parts, build it, and keep improving the robots already at work.
They form the roles and knowledge each task needs, as people do.
02 / TWO SERVICESLEDGER / CREW
Lead the work.
Or delegate it.
Use Ledger when your team drives development and wants review grounded in the record. Use Crew when you want a persistent agent team to develop within a scope you set.
Gantry Crew
Bots tailored to your team and project work like an organization. Given a goal and scope, they plan, talk, and act on their own. They carry lessons forward, fitting your work better over time.
Each service stands alone. The larger path runs from design and digital twins to sourcing, operation, and repair.
GET STARTEDGANTRY CLI / CODING AGENTS
Start with
your agent.
Paste this into any coding agent to choose a service and prepare its setup.
VIEW FULL PROMPT
Help me choose and set up Gantry for this robotics project. Use the official repository: https://github.com/fourn9/gantry-oss. Read README.md, docs/SERVICE_SPLIT.md, docs/AGENTS.md, docs/BOT_DEVELOPMENT.md, docs/CREW_ORGANIZATION.md, and SECURITY.md. Ask whether I want Gantry Ledger, where I direct development with recorded evidence and review, or Gantry Crew, where a persistent agent team develops within my delegated scope. Do not set up both unless I ask. Check prerequisites and install Gantry as documented. Identify my project, tools, agent client, goal, and completion conditions. For Ledger, prepare a scoped local connection with gantry-ledger connect PROJECT_PATH --client CLIENT --prepare, using the actual path and supported client name. For Crew, follow its onboarding docs and show me the proposed roles, permissions, tools, runtime, and inference use before activation. Keep owner credentials private and preserve existing client settings. Leave activation, worker startup, and paid inference to my approval. Show me what will change and how to verify it.