Now on npm  ·  MIT License

Your workday,
synthesized.

Workweave pulls open PRs, assigned issues, and Slack mentions into a single, prioritized plan — so you start each day with clarity, not tabs.

Get started View on GitHub
workweave synth
──────────────────────────────────────────────────────────────── Workweave Plan · Tuesday, April 15 · github, linear, slack · AI · local ──────────────────────────────────────────────────────────────── 6 tasks · 5h 40m scheduled · 2h 20m remaining · 84 events fetched 1 0:00–1:30 REVIEW Auth token expiry — 3 PRs need sign-off 94 Blocking: @dana waiting on review since yesterday ↗ github.com/emin93/api/pull/204 2 1:30–2:30 IMPLEMENT Rate-limiter refactor — branch ready to close 88 api · workweave ↗ github.com/emin93/api/pull/198 3 2:30–3:00 RESPOND Slack: infra outage discussion thread 76 #incidents · 14 replies, 2 @mentions 4 3:00–4:00 IMPLEMENT Linear P1 — fix pagination bug in search API 71 api · assigned 2d ago 5 4:00–4:45 INVESTIGATE Flaky CI in monorepo — 4 failures this week 58 ↗ github.com/emin93/monorepo/actions/runs/9912 6 4:45–5:40 FOLLOW UP Vendor migration doc — waiting on PM reply 42 ████████████████████████████████░░░░░░░░ 5h 40m / 8h (71%) ────────────────────────────────────────────────────────────────

Everything you need. Nothing you don't.

Built for developers who want signal, not noise.

⚡

Three connectors, one plan

Pull signals from GitHub, Linear, and Slack. Correlate related items across sources automatically.

🤖

AI-powered or fully offline

Use a local Llama model with no API key, or plug in Anthropic or OpenAI. Rules mode works with no internet.

🗓

Time-boxed scheduling

Tasks are fit into your available hours with context-affinity nudging — related work stays together.

🔢

Priority scoring

Items are ranked by urgency, importance, social pressure, staleness, and blocking factors — not guesswork.

🖥

Rich terminal output

Color-coded plan with time slots, category badges, source URLs, and a time-budget bar.

📦

JSON mode for automation

Pipe --json output into dashboards, scripts, or your own tools with a clean, stable schema.

How it works

A five-step pipeline from raw events to a ready-to-execute plan.

Ingest
→
Normalize
→
Correlate
→
Prioritize
→
Schedule

1 · Ingest

Connectors fetch open PRs, assigned issues, DMs, and thread mentions in parallel.

2 · Normalize

Every raw event is mapped to a common Artifact model with source, type, timestamps, and metadata.

3 · Correlate

Artifacts sharing a repo, project, or linked ID are grouped into clusters so related work stays together.

4 · Prioritize

Each cluster gets a priority score from 0–100 based on urgency, age, blocking status, and social signals.

5 · Schedule

Clusters are assigned time slots within your workday budget. Overflow goes to the backlog.

AI that runs anywhere

Pick your provider once in workweave setup — it's saved and used on every run. Override anytime with --provider.

Anthropic Claude

Uses claude-haiku-4-5. Fast, hosted. Requires an ANTHROPIC_API_KEY.

OpenAI

Uses gpt-4o-mini. Fast, hosted. Requires an OPENAI_API_KEY.

Local model

Qwen 2.5 1.5B via node-llama-cpp. Runs on CPU, fully private, no API key needed.

Connectors

Workweave pulls signals from where your work actually lives. More connectors coming.

🐙

GitHub

Open pull requests, requested reviews, assigned issues, and CI status across your repos.

📐

Linear

Assigned issues and in-progress tickets from your Linear workspace.

💬

Slack

Direct messages, thread replies, and @mentions that need your attention.

＋

More coming

Jira, Notion, GitHub Discussions, and more. Open an issue to request a connector.

Get started in minutes

Node.js 18+ required.

# Install globally from npm $ npm install -g workweave   # Run the interactive setup wizard $ workweave setup   # Check your connectors are ready $ workweave detect   # Synthesize today's plan $ workweave synth

Available on npmjs.com/package/workweave  ·  Building from source? See the README.