Who meridian is for#
Developers tracking work across multiple tickets in a sprint
Meridian's day timeline shows exactly how time split across tickets and context switches, including unplanned work that derailed the sprint plan. At standup time the draft is already written from what the developer actually worked on, not what they remember at 9am the next morning.
Skip if:
Skip if your team requires billable-hour reporting by client or project. Meridian produces a work journal and standup drafts, not a billable-hours export, so it does not replace Harvest or Toggl Track for invoice-level time reporting.
Individual contributors building a performance review record
Meridian saves every workday as a searchable journal entry, so when performance review season arrives you have a concrete, dated record of what was shipped, what was fixed, and what decisions were made. The journal is local and permanent, independent of any PM tool or employer-controlled system.
Skip if:
Skip if your employer requires all work to be logged in a centrally managed system. Meridian's local journal does not integrate with corporate reporting dashboards or feed into org-wide time tracking.
Freelancers reconstructing weekly timesheets from actual work
For freelancers who forget to run a timer, Meridian's automatic capture provides a ground-truth record of what was worked on each day. The journal entries can be used to reconstruct timesheet entries for client billing, even when the work happened across scattered sessions without a timer running.
Skip if:
Skip if you need client-facing invoice generation or a billable-hours breakdown by project that exports directly to accounting software. Meridian does not produce those reports.
Developers explaining estimate overruns in retrospectives
Meridian logs the actual sequence of what happened during a task: what apps were in use and how long each phase took. When a two-day estimate becomes five days, the journal provides the concrete account of what occupied the extra time, ready to reference in a retrospective or post-mortem.
Skip if:
Skip if your retrospectives are team-level and aggregate across multiple people's work. Meridian is a personal tool with no multi-user view or team dashboard.
The problem it solves#
Most time-tracking tools assume you will remember to start a timer when you switch tasks and stop it when you are pulled into something else. In practice, you forget. The result is a worklog fabricated at the end of the week or a gap in the record that makes sprint retrospectives, performance reviews, and client billing all harder than they need to be.
The deeper pain is reconstruction. When a manager asks why a feature took three days instead of one, or when you need to explain an estimate overrun in a retrospective, you are usually working from memory and git timestamps rather than an accurate account of what happened. Tools like Toggl Track and Harvest put the burden on you to log accurately; if you did not log it at the time, it is gone. Meridian inverts this: it watches what happened and lets you approve the record, instead of asking you to build it from scratch.
How it solves it#
Automatic screen capture and day timeline
Meridian captures the apps and windows you work in continuously in the background, using approximately 5 to 10% CPU on average. It builds a timeline of your day from what actually appeared on screen, so the record reflects what you did rather than what you remembered to log. Capture runs in-process in the tray; no separate daemon needs to be registered.
AI-generated daily summary and standup draft
At the end of each workday Meridian passes your activity timeline through an AI provider you connect (Google Gemini, OpenAI, or Claude) to produce a summary of what was completed, which estimates held, and where time actually went. It also writes a standup draft ready to paste or post directly to your PM tool, with no manual typing required.
PM tool integration with approval gate
Meridian connects to Jira, Linear, GitHub Projects, Azure DevOps, and Trello. Connections are read-only until you approve a specific update; Meridian never writes to your board on its own. When you approve a post, it submits the worklog or ticket update directly with the full context of what actually happened on that task.
100% local encrypted storage
All captured activity, analysis, and journal entries are stored in an encrypted database on your device. Nothing is uploaded to any server, and no activity data is transmitted to the AI provider beyond what is needed to generate a specific summary. Older capture is pruned automatically after 30 days; active storage runs to approximately 20 GB per month.
App exclusion and pause controls
Any app can be excluded from capture from the menu bar at any time. Capture can be paused instantly and resumed when ready. Individual journal entries, days, or captured sessions can be deleted. These controls mean you decide exactly what goes into your work record and what stays private.
Searchable work history across any past day
Every day is saved as a complete, searchable journal entry the moment it is summarized. You can pull up any past day to see what was worked on, what decisions were made, and what ran over estimate. The journal is stored locally, so it is available offline and persists independently of any cloud service or employer system.
Strengths and trade-offs#
Strengths
- No manual time entry or timer discipline requiredUnlike Toggl Track and Clockify, which depend on you remembering to start and stop timers, Meridian captures time passively from screen activity. The record is always complete regardless of whether you remembered to log. For developers who context-switch frequently or get pulled into unexpected work, this is the material difference between an accurate record and an approximate one.
- All data stays on device with fully auditable MIT codeEvery line of code is MIT licensed and available on GitHub, so any privacy concern can be settled by reading the source. The capture database is encrypted at rest on your machine. No activity data is stored on Meridian's servers; the only thing that ever leaves your device is a worklog you explicitly approved, sent directly to your PM tool.
- PM updates drafted from observed work, not recollectionMeridian writes PM updates from the observed timeline of what you did, not from a manual description you type at the end of the day. The update reflects which tasks ran over estimate, what unexpected work appeared, and what actually shipped, with the correct task context matched automatically. You review and approve; the tool does not post without your sign-off.
Trade-offs
- -Heavy local storage footprint (~20 GB per month)Screen capture generates approximately 20 GB of data per month, even with automatic pruning of capture older than 30 days. Users on machines with limited SSD space will need to monitor available storage, particularly during the first month of use when no pruning has occurred yet.
- -macOS prebuilt is Apple Silicon only; Intel not supportedThe downloadable app does not support Intel Macs. Intel users must build from source, which requires Rust 1.93.1, Node 20, and bun, plus running the install and dev scripts manually. The build process works on Intel per the README, but there is no prebuilt Intel binary.
- -Daily summaries require an external AI provider API keyThe journal summarization and standup drafting features require a connection to Google Gemini, OpenAI, or Claude. You need an API key from one of those providers and will incur their per-token costs for each daily summary. The capture timeline works without a provider, but the journal drafting and PM posting features do not.
meridian vs alternatives#
Meridian vs Toggl Track
Both tools track where your time goes during the workday, but they take opposite approaches to data capture. Toggl Track requires you to start and stop timers manually or use its browser extension to log activity; every entry reflects what you told it. Meridian watches what is actually on screen and reconstructs your day from observed activity, so the record reflects what happened rather than what you remembered to log.
| Feature | Meridian | Toggl Track |
|---|---|---|
| License | MIT | Proprietary |
| Data storage | On-device, encrypted | Cloud |
| Time capture | Automatic (screen activity) | Manual timers / browser extension |
| PM tool integration | Jira, Linear, GitHub Projects, Azure DevOps, Trello | Via third-party integrations |
| Pricing | Free, build from source | Free tier; paid plans above that |
Meridian wins for developers who want passive, accurate logging without changing their workflow. Toggl Track is still the better choice when you need team-level time reports, a polished mobile app, or client billing integrations that Meridian does not currently offer.
Meridian vs RescueTime
RescueTime also runs in the background and categorizes your time automatically, making it the closest conceptual parallel to Meridian in the paid category. The difference is in the output: RescueTime produces productivity scores and category breakdowns; Meridian produces a work journal, a standup draft, and PM ticket updates. If you want to know whether you spent too much time in Slack, RescueTime answers that. If you want a searchable record of what you actually built and why a task ran long, Meridian is more useful.
RescueTime sends all activity data to its cloud servers. Meridian keeps everything on-device. For developers in environments with strict data policies, or who simply do not want a third-party service receiving their screen activity, that distinction is material.
Meridian vs Harvest and Clockify
Harvest and Clockify are manual time-tracking and billing tools aimed at freelancers and agencies tracking billable hours across clients. Meridian does not produce billable-hours reports or invoice exports, so it is not a direct replacement for teams that use Harvest for client billing. Where Meridian is relevant for freelancers is in producing an accurate day-level record of what was actually worked on: the automatic capture can serve as the raw log that feeds manual billing entries, rather than relying on memory at the end of the week. If client invoicing is the primary need, Harvest or Clockify remain the better choice.
Install and self-host#
Build Meridian from source: install Rust 1.93.1, Node 20, and bun before running the setup scripts.
```bash
git clone https://github.com/Meridiona/meridian
cd meridian
cp .env.example .env
bash install-dev.sh
bash dev-start.sh
```What it's built on#
- Languages
- JavaScriptRustTypeScript
- Frameworks
- Next.jsReact
FAQ#
Does Meridian send my screen activity to the cloud?
No. All capture, analysis, and journal storage happen on your device. The only data that ever leaves your machine is a worklog you explicitly approved, sent directly to your PM tool. If you connect an AI provider for daily summaries, only the minimal context needed to generate that summary is sent to the provider; raw screen capture is never transmitted.
Which PM tools does Meridian connect to?
Meridian currently connects to Jira, Linear, GitHub Projects, Azure DevOps, and Trello. All connections are read-only until you approve a specific post; the tool never writes to your board on its own. You review each drafted update and submit it with one click.
Is Meridian useful without a PM tool connected?
Yes. Without a PM tool connection, Meridian works as a private, searchable record of your work: daily timeline, completed tasks, decisions, and the reasons tasks ran long. The standup draft and PM posting features require a connection, but the journal and timeline are fully functional as standalone features.
Does Meridian work on Intel Macs?
The prebuilt download is Apple Silicon only; Intel is not listed as a supported platform for the binary distribution. Intel Mac users can build Meridian from source (requirements: Rust 1.93.1, Node 20, and bun), but there is no prebuilt Intel download. Windows 10/11 is fully supported with a prebuilt binary.
Is Meridian free to use?
Yes. Meridian is MIT licensed and free to download, build, and run. There is no paid plan, no trial, and no credit card required. The only ongoing cost is the API usage from the AI provider you connect for daily summaries (Google Gemini, OpenAI, or Claude), billed directly by that provider based on your daily usage volume.
Similar open-source tools#
rakazo
AI teammates you own: your keys, your model, your machine.
showmd
The most readable way to preview any .md file on macOS.
Kanri
Local-first Kanban board that stores all data in a JSON file
Taiga
Agile project management for Scrum, Kanban, and hybrid teams
Zettlr
Open source Markdown editor for researchers and academics
OpenLogi
Local-first Logitech control without accounts or telemetry