Plan and run coding sessions from Obsidian with Claude Code
Author drives Claude Code from Obsidian, with plan prompts, handoff docs, per-plan tmux sessions and an Efforts page tracking open PRs and progress. ·
Done withClaude CodeClaude
- What they did
- Runs Claude Code inside separate Linux user accounts on Arch, one per major project, each with its own Obsidian vault. QuickAdd commands create a plan prompt, then a plan workflow skill writes the plan and a handoff doc as subpages. A "Send to Claude Session" command opens a tmux session named for the plan.
- How it went
- Gives separate per-plan sessions and a global Efforts page showing open PRs, percent complete, next step, and a color-coded executing, idle or closed status. No timings or quantified results are given.
- Worth knowing
- The author pays for a $20/mo personal and $100/mo work Claude subscription, and found OpenClaw and Hermes unproductive and local Qwen via OpenCode slow.
Try it yourself with Claude Code
Work from my Obsidian vault at [vault path]. For the task [describe feature or bug] in [repo path], write a plan note, then a handoff doc listing the steps, files involved and how to test it, and start a tmux session named for the plan. Keep an Efforts page in the vault with each plan's status and open PR links, and update it as you go; you're done when the plan, handoff doc and Efforts entry all exist. Ask me before pushing code or opening a PR.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Nov 5, 2026
Related use cases
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Research and draft cold outreach with Claude Code
Claude Code handles research and drafting for cold outreach while a separate platform sends at scale.
Publish local news posts to WordPress automatically
Claude Code agents research local news, write HTML posts, upload them to WordPress, set featured images, verify and log results, and also check Airtable and Gmail.
Submitted by from search gap: marketing
Review Linux kernel patches
Google engineers launched Sashiko, an agentic AI code reviewer for the Linux kernel.
Custom agent
Upgrade site dependencies and frameworks with Cursor
Kent C. Dodds had Cursor's agent look up migration docs, plan and upgrade Remix v2 to React Router v7 and xstate in his site.
Build an AI app over ten days
The author spent about 10 days building an AI-based app with Replit Agent; it worked well but sometimes forgot instructions and reintroduced unwanted behavior.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.