Shipping 2.1x More PRs Using Claude and Linear Automation Stack
jeremystan · x · 2026-08-23
The author achieved a 2.1x increase in weekly PR output for a large-scale ML project by combining Claude, Linear, and custom automation tools.
Core Toolchain:
- fable: Planning and orchestration.
- Linear: Shared state and dependency management.
- dispatcher: Automation dispatcher.
- Claude: Code implementation.
- worktrees/cmux: Isolation and review surfaces.
Workflow:
- Sessions create Linear tickets. The planning session sets strategy, grooms the backlog, and promotes tickets to TRIAGED.
- The author reviews and drags them to TODO for execution.
- The dispatcher polls for tickets, launches sessions, and presents an HTML plan for approval.
- The author is notified again when the PR is green with a confidence estimate.
- The dispatcher handles Git state sync, auto-resolves CI failures, and runs queued experiments automatically.
The author now focuses solely on reading, feedback, and merging.
More from coding & agent
- vLLM keeps hitting 400 context errors with DeepSeek harness; llama.cpp runs 24h+ fine — cviperr33 · 2026-08-23
- Dad ditches OpenRouter token burn for local Qwen 3.8 + Pi agent, controlled via Telegram — dcnotpc · 2026-08-23
- FatherLode dev update: Claude-built game adds weather, museum with 30+ treasures, Suno 5.5 music — Extension-Parsnip789 · 2026-08-23
- Open Source MCP Connector: Links Claude with Office 2019 — FarRespond73 · 2026-08-23
- Can Local Qwen Models Handle Real-World System Programming like GTK4/Qt? — MongoWithBongoss · 2026-08-23
- OpenClaky explores Agent self-evolution and custom apps — 赛博禅心 · 2026-08-23