Low-overhead capture
Fast manual logging for humans; automatic capture for AI agents.
Record human and AI work with almost no overhead, then turn one unified timeline into any report — daily, weekly, monthly.
Open source · MIT · Self-hosted on Cloudflare
Why now
Agents write code, run tasks and burn tokens alongside your team every day. Spantail puts that work on the same timeline as human effort, so a week of human + AI work is finally one honest record.
A date, duration, description and tags. Built for fast daily logging — no forms to fight.
Register an agent and its sessions land on the timeline — duration and token usage, no manual entry.
Apply a template to any filters — workspaces, projects, users, dates. Daily, weekly, monthly, built in.
Fast manual logging for humans; automatic capture for AI agents.
Markdown + Liquid templates over any filters. No hardcoded report types.
Immutable snapshots via expiring, revocable links with an optional passcode.
Inline Markdown comments and emoji reactions, right next to the report.
c to create, j/k to move, o to open. Operate without leaving the keyboard.
Built-in MCP server and CLI over the same REST API the web UI uses.
A report is just a template applied to filters you choose — any mix of workspaces, projects, people and dates. Nothing is hardcoded.
Register a coding agent once and its sessions stream onto the timeline — duration and token usage per session, with no manual entry. See exactly where the tokens went.
Integrations
Install the plugin once and every Claude Code session lands on the timeline — duration and token usage, no manual entry.
Learn more →Connect the GitHub App and commits, pull requests and reviews line up next to logged time on the same timeline.
Learn more →A built-in MCP server and a CLI talk to the same REST API the web UI uses. Log work, pull reports, or wire Spantail into scripts and AI clients.
$ spantail log "Implemented report engine" --project core --duration 2h ✓ entry recorded · core · 2h $ spantail report run monthly --workspace acme --month 2026-06 ✓ monthly report ready → 141h across 5 projects # AI clients connect over MCP → https://acme.spantail.com/mcp
Spantail isn't a service you hand your work to. You run it yourself, so every entry, report and token count lives on infrastructure you control — not ours.
The open-source core is MIT — read, audit and fork it. No black box, no lock-in.
Runs on your Cloudflare account; work lives in your own D1 database.
One deployment, one company. Your records never sit on our servers.
Because you host it, your work is never sent to us — or used to train anyone's AI models.
One Worker serves the API, the MCP endpoint, shared report views, and the app. D1 is your primary database. One deployment, one company — not a multi-tenant SaaS.
Enterprise
The full core: time tracking, reports, sharing, CLI, MCP, Claude Plugin and GitHub App. Self-hosted on your Cloudflare account.
Read the setup guide →SSO/SAML, audit log, org-wide analytics, DLP and priority support — on top of the open-source core.
View plans →Interested in reselling, integrating or deploying Spantail?Partner with us →
Clone spantail/spantail and deploy to your own Cloudflare account — the setup guide takes you from zero to a running instance in minutes.
Yes. Spantail is lightweight enough to start on Cloudflare's free tier. Depending on your usage and the features you enable, the paid plan may be needed.
Of course. Spantail works fine for a team of one — and it is a great fit for freelancers with multiple clients and projects: log everything in one place and report per project.
Spantail is open core: the core is MIT-licensed and self-hostable. Enterprise features (SSO, audit, analytics, DLP) are a paid add-on.
In your Cloudflare D1 database. One deployment serves one company — it's not a multi-tenant SaaS.
No. The reference Claude Code hook sends only compact token/timing telemetry — conversation bodies stay on your machine.
Spantail is open source and generally available. Deploy it on your own Cloudflare account in minutes — or talk to us about Enterprise.