Issues
Subtasks, links, custom fields, attachments and a full history on every issue. Soft delete with a restore window, so a mistake is not a loss.
ACME-12 Bug Major Severity: High depends on ACME-4
Issue trackingOpen-source project tracking for product and engineering teams
Track issues, plan sprints and keep project decisions together. Connect your AI agent to find and update work with the permissions you choose.
Free during beta. Your first 20 members stay free forever. AGPL-3.0, source on GitHub
Alex Lee assigned ACME-12 to Alex Leejust now
Alex Lee changed Status: Open to In Progressjust now
Alex Lee logged 20m, Developmentjust now
›Find the unassigned high-priority bugs in ACME, take the newest one, and log the 20 minutes I just spent on it.
ACME-12 is yours and In Progress, with 20m logged. ACME-7 and ACME-9 are still unassigned, and ACME-7 is Critical.
Watch an AI agent update a real sprint through MCP, then follow the work to Done in the app.
Nothing Left keeps the issue, the plan and the reasoning in one place, and keeps your team and your agents working from the same project context. Open an issue and the decision that shaped it is one click away. Ask your agent and it reads the same page.
It covers issues, sprints, time, reports and project knowledge. It does not review code or run deployments.
One issue, the whole way through. Pick a step to see what the team sees at that point.
Steps. Open an invitation link that is older than seven days.
Expected. A message that the link has expired and a way to request a new one.
Actual. A blank page.
Casey Kim assigned ACME-12 to Alex Lee3 Oct 2026
Casey Kim set Estimate: 2h, Due: 9 Oct 20263 Oct 2026
Decision record, published to project members
Invitation links expire after seven days (INVITE_TTL_DAYS). An expired link must explain itself and offer a new one. A blank page or a redirect to sign-in is a bug.
Decided 2 Oct 2026. The expiry message is shown on the invitation landing page itself, before any sign-in prompt, because the person may not have an account yet. See ACME-12 for the first instance.
On ACME-12, under Referenced by
Invitation links and expiryKnowledge base
Mentioning ACME-12 links the issue, and the issue lists this article under Referenced by. Your agent can read either.
Alex Lee changed Status: In Progress to Fixed8 Oct 2026
Alex Lee logged 1h 25m, Development8 Oct 2026
Alex Lee logged 20m, Development6 Oct 2026
Hand-drawn product views with fictional data. Nothing here changes code or deploys anything; the agent only reads and updates the tracker.
People use the app. Agents use MCP. Both reach the same product API with the same identifiers, the same Markdown and the same permission rules. A compatible MCP client can manage issues, sprints, time and project knowledge through it.
claude mcp add --transport http nothingleft \
https://api.nothingleft.dev/mcp \
--header "Authorization: Bearer nl_pat_…"
# or locally over stdio: npx -y @nothingleft/mcpWorks with MCP clients that support HTTP token authentication or a local stdio setup. Read the setup guide
The MCP server is a client of the public REST API. If the app can do something, the agent can, credentials aside. If the app cannot, neither can the agent.
Agents connect with a personal access token you scope to projects and actions. Every change is recorded under you. A missing scope is refused by name, not silently ignored.
ACME-12, @alex, In Progress, 1h 30m. Markdown is stored and Markdown is returned. There is nothing to parse before a model can reason about the work.
Triage an issue, find duplicates, write the standup, draft release notes, turn a resolved incident into a runbook. Each one shows its plan and applies it when you agree.
Statuses, priorities and types are lists you edit. Five built-in roles keep permissions straightforward. There is no workflow engine to configure before the first bug can be filed.
Subtasks, links, custom fields, attachments and a full history on every issue. Soft delete with a restore window, so a mistake is not a loss.
ACME-12 Bug Major Severity: High depends on ACME-4
Issue trackingOrder the backlog, plan a sprint with a goal and points, follow it on a board by status, priority or type. A sprint ends with a decision about what is left, never a silent rollover.
Sprint 08 21 of 34 points 3 added after start
Sprints and boardsOne query language for the search bar, saved views, report scopes and agents. Time is logged against the issue it explains. Nine reports, each leading back to the issues it counted, all exportable as CSV.
type: Bug #Unresolved assignee: me sort by: priority
Search and reportsRunbooks, decisions and onboarding notes beside the issues they explain, private to the project. Reference an issue from an article and the issue shows the article back.
article://ACME/invitation-links-and-expiry referenced by ACME-12
Knowledge baseSmall enough that setup is an afternoon. Complete enough that you are not back in a spreadsheet by the second sprint. Team size does not define the fit; try it on real work and decide.
Why teams try it
Decisions and runbooks sit beside the issues they explain, in the project, found by the same search.
Same API, same permission rules, every change recorded under you. Nothing to wire up beyond a token.
AGPL-3.0, self-host the same build, and the first 20 members of every organisation free for good.
Limits, and what is deliberately not built, are answered in the questions below
Sign up, name your organisation, create a first project and you are on the issue list. Invite people when you are ready, or try it alone first. Then tell us whether it fits.
Invite everyone who does the work. Nothing is withheld past a headcount.
When billing arrives, the first 20 members of every organisation remain free. The promise is stated in the product, not only here.
No seat is priced today. You will hear about any future charge well before it takes effect.
The questions that decide whether a trial is worth your week. Short answers here, details in the docs.
No. It is a complete web app for issues, sprints, time, reports and project knowledge. Agent access is there when you want it, through a token you create and can revoke.
You sign up with Google, GitHub or an email address, name your organisation and create a first project; the key is derived from its initials and you can change it. Inviting people is a step you can skip. You land on the project's issue list, usually within a few minutes. There is no approval queue.
Yes. Create an organisation and a single project, invite people later, connect an agent later or never. Issues export to CSV at any time, so a trial does not lock anything in.
No subscription charge for any team size during beta. The first 20 members of every organisation stay free forever once billing exists. We will give notice before charging for seats beyond that; nothing is priced today. Pricing
Each organisation has 5 GB of platform storage for attachments; avatars do not count. You see usage at 80% and 95% before anything is refused. Connect your own Cloudflare R2 or AWS S3 bucket and the limit lifts the moment the bucket is verified. Limits and storage
Yes. The code is AGPL-3.0 and the hosted service runs the same build: PostgreSQL, S3-compatible storage, the web app, the API and the MCP server. Seat and storage limits are environment settings you can change. One Docker Compose command runs all of it, and a second guide covers AWS. Self-hosting guide
MCP for any agent client, over HTTP or locally over stdio. GitHub and GitLab: branches, commits and pull requests linked to issues by key, with review and CI state on the issue, and status rules that move an issue on a branch, a pull request or a merge. Google and GitHub sign-in. CSV export and generic CSV import.
Not built yet: deployment status on issues, outgoing webhooks, native importers from other trackers, Bitbucket, and general workflow rules beyond those status rules. An agent connected to both your old tracker and this one can help carry issues across, with you reviewing the mapping. GitHub and GitLab
A workflow or transition-rules engine: any status may follow any status, and automations only set a status when git work or subtasks move. Custom roles and permission schemes: there are five built-in roles. Per-comment visibility: comments follow the issue. These keep setup short and the agent surface predictable.
@alexyou touched the invite flow last. Could you take this one?