Trakr
Projects contain tickets. That's the whole system.
A self-hosted issue tracker and work-management platform built to be unceremonial, including what the incumbent makes you buy separately: a query language, a customer support desk, ITSM and ITIL service management, SLA policies, timesheets and billing, ten drillable reports, on-premise AI, and a pipeline that takes a ticket to a pull request.
The tracker became the work.
Somewhere between the first board and the third year, the tool meant to record work started generating it instead: fifteen ticket types, a configuration nobody dares touch, page loads measured in seconds, a support desk, a service desk, a timesheet tool and a reporting add-on each on their own licence — all hosted somewhere you did not choose, at a per-user price that rises every time you hire.
One application, one database, and a data model you can hold in your head.
Three decisions worth reading
Capabilities
- Types, priorities, statuses and services are configuration rows, not fixed lists — each organisation, and optionally each project, defines its own with colour and icon.
- A purpose-built query language over core fields, people, dates, collections, time, links, full text and SLA state, with fourteen operators and historical operators replayed from the activity log.
- Design guarantees: a negation reaches empty rows, a bare date means the whole day, every sort is settled for deterministic paging, and an unparseable value is refused with the clause quoted back.
- An AI mode returns a query, never results, so paging, columns, grouping, export and bulk selection keep working and the query stays visible, editable and savable.
- Ticket templates at user, project and organisation scope, optionally bound to a type; bulk update, transition and link with values resolved by name across projects.
- List, board, roadmap, calendar, timeline, backlog and sprint views, with cycles as sprints, phases or program increments.
- Delegations — 'I'm away until the 19th' — redirect assignment and make routing skip you.
- Three project worlds — standard, helpdesk and service — separated server-side, not by whichever page is asking.
- A public customer portal with emailed guest access codes, internal versus public comments, canned responses and per-project settings.
- ITSM and ITIL as opt-in switches: everything is a ticket, so service requests, incidents, problems and changes share comments, search, bulk edit, merge, SLA, automation and the API unchanged.
- Problems carry root cause, workaround and a known-error flag; changes carry type, risk, planned window, implementation and backout plans and an outcome.
- SLA policies with per-priority targets, business calendars and holidays, pause-on-status and breach notification; a running timer keeps the deadlines it started under.
- Approvals enforced by the workflow condition — a review board is simply an approval asked of a group.
- Email-to-ticket across five providers with reply threading by ticket key.
- Work logs in familiar duration notation with a billable flag — no approval workflow, no sign-off, no locking period.
- A personal timesheet as a day-tile grid including empty days; exports as CSV, spreadsheet, PDF, including a multi-month single PDF.
- Statistics spanning up to ten years with day-through-year buckets, year-over-year comparison, and both person-days and full-time-equivalent reporting, clearly labelled.
- Billing and budget reporting in mandays at a price per manday.
- Ten reports: ad-hoc generator, billing, budget, SLA, resolution metrics, time in status, team performance, statistics, timesheets and AI spend, plus a personal dashboard with eighteen widget types.
- Time in status replays the activity log so it can never drift, and reports the median as well as the mean.
- Every figure drills through to the underlying tickets as an ordinary, editable, savable search.
- Rules at platform, tenant, organisation and project scope, triggered by schedule, events, or both — never a cron expression to mistype.
- Twelve quality checks, an open-ended condition builder over twenty-six fields and twenty-four operators, and nineteen actions from field-setting to webhooks.
- Guarantees: a rule-authored write never triggers another rule, the workflow is never overruled, one failing action costs only that action, and dry run reports what it would do without doing it.
- Assignment routing with round-robin or least-busy modes over member, group, skill or named pools, with a live preview of every candidate and why.
- Over twenty-five AI providers, sixteen self-hostable — the whole AI feature set runs on-premise with nothing leaving the network.
- Twelve AI features across search, suggestion, translation, generation, merging, templates, assistance, bulk editing and reporting, with tiered cost control, budgets, rate limits and a circuit breaker.
- A Model Context Protocol server with nine tools, eight resources and five prompts; ticket-to-pull-request runs a coding agent on a branch under a budget cap.
- Guided organisation and project setup wizards that leave nothing behind if abandoned.
- Organisation restructuring: a drag-and-drop board for moving, merging and splitting, rewriting ownership across roughly thirty tables in one transaction.
- Configuration transfer exports an organisation or project as a portable bundle with no identifiers, people or secrets, and import never deletes.
- Eighteen email templates resolved per field down four scopes, with a refusal to publish a customer subject that drops the ticket key.
- Audit forwarding to ten sinks per organisation over signed HTTP or syslog, plus backups, self-update, migration management and nightly housekeeping.
- A cluster with front-end, back-end and scheduler roles and a designated backup node that takes over within ninety seconds, without a restart.
- Import from CSV, and from the incumbent's CSV and XML exports, for a clean migration.
What it does
- 01One application replaces four separate licences: support desk, ITSM, timesheets and reporting are all included.
- 02On-premise AI across 25+ providers, 16 self-hostable, so nothing leaves the network — plus a pipeline that turns a ticket into a pull request.
- 03Licensed per deployment, not per seat, with unlimited users.