TODO.mdProject execution for teams building with AI agents.
Execution is faster now. Context has to keep up.
AI agents can help teams implement, revise, and ship work faster than before. But faster execution creates a new problem: product context gets stale, decisions disappear into chat, tickets lose the “why,” and teams struggle to understand what changed after the work ships.
TODO.md is designed around the opposite idea. The project system should become a living product memory layer: what is being built, why it matters, who owns it, what changed, how long it took, and what context humans and agents need to move work forward.
Real-time product context
Issues, documents, files, activity, and time logs stay connected in one live workspace.
Built for human + agent teams
Engineers, product leads, and coding agents can work from the same source of truth.
Opinionated, not sprawling
Capabilities stay focused and opt-in, so teams do not have to configure a large system before doing real work.
Designed for rapid delivery cycles
When work moves from idea to implementation in minutes or hours, the surrounding product process has to move just as fast.
Everything the team needs to understand and move the work.
Issues and boards
Track work across customizable project flows with real-time updates.
Sprints
Plan cycles, track progress, and review delivery with optional sprint workflows.
Timeline
Map larger initiatives across time with day, week, and month views.
Epics
Group multi-week efforts and connect related work across tasks and projects.
Time tracking
Log time manually for honest delivery visibility — without timer theater.
Product documents
Keep specs, notes, and decisions close to the work they support.
Files and attachments
Attach project files and preserve their context alongside related tasks.
Activity history
See what changed, when it changed, and who changed it.
CLI
Let engineers and coding agents inspect context and move work from the command line.
MCP connector
Give AI assistants controlled access to read project context and perform approved actions.
A shared operating layer for humans and AI agents.
TODO.md is not only a human interface. It is an operating layer that AI agents can read and act on through controlled tools.
The goal is not uncontrolled automation. The goal is reliable collaboration: agents can access context, summarize work, inspect tasks, report progress, and perform approved actions while humans keep control over important changes.
tt CLI
The tt CLI gives engineers and coding agents a fast way to inspect project context, read tasks, update work, and log time without leaving the terminal.
MCP connector
The MCP connector lets AI assistants access TODO.md through a controlled tool interface. Read actions can be made low-friction, while write actions remain approval-gated.
Read · low-friction
- list_tasks
- get_task
- search_tasks
- list_projects
- get_project
- docs_search
- docs_read
- time_report
- get_context
- list_docs
Write · approval-gated
- › create_task
- › update_task
- › set_task_done
- › comment_task
- › log_time
For product teams moving faster with AI.
Product leads
Keep priorities, requirements, decisions, and delivery status in one place.
Engineers
Work from clear context, inspect tasks from the CLI, and keep delivery updates lightweight.
Designers
Connect specs, files, timelines, and feedback without losing product context.
AI agents
Read the same project context as the team and perform approved actions through controlled workflows.
Private beta, by invitation.
TODO.md is currently in private beta and expanding carefully. We are focused on reliability, workflow fit, accumulated product context, and agent-assisted delivery patterns before broader availability.