v1.1.0: Your AI Coding Agent Now Writes Your Work Journal
Git captures the what. Your AI agent now captures the why—automatically, after every commit, across 6 platforms.
The missing layer
Worktale has always captured the mechanics of your day: commits, line counts, files changed, timestamps. But the most valuable part of a coding session—the decisions—disappeared the moment you moved to the next task.
Why did you choose that approach? What trade-off did you make? What was the root cause of that bug you spent two hours on? The git diff can't tell you. And by Friday, neither can you.
With v1.1.0, your AI coding agent captures all of it. Automatically. After every commit.
What's new in v1.1.0
The worktale note command
A new CLI command that appends a narrative note to today's daily summary. Simple, fast, and designed to be called programmatically by AI agents.
$ worktale note "Fixed race condition in job queue — workers were claiming the same job because the claim query wasn't using SELECT FOR UPDATE"
✓ Note added to 2026-03-12
Notes are appended, never overwritten. Multiple notes accumulate in order throughout the day. They appear in worktale digest, the TUI dashboard, and (eventually) your Worktale Cloud portfolio.
The command also auto-registers repos—if you run it in a git directory that isn't tracked yet, it creates the database entry for you. No need to worktale init first.
Direct AI agent integrations
This is the big one. We built integrations for every major AI coding agent on the market:
| Platform | Type | Enforcement |
|---|---|---|
| Claude Code | Plugin (marketplace) | Deterministic |
| GitHub Copilot CLI | Plugin + postToolUse hooks | Deterministic |
| OpenAI Codex CLI | Skill (SKILL.md) | Advisory |
| Cursor | Rules file | Advisory |
| Cline | Rules file | Advisory |
| Windsurf | Rules file | Advisory |
Deterministic means the integration is enforced by the platform's plugin or hook system—it fires every time, regardless of the model's mood. Advisory means the agent follows instructions in a rules file. It works reliably in practice, but you can combine it with worktale hook install for guaranteed git-level capture as a safety net.
How it works in practice
Here's a real session. I activated /worktale in Claude Code and asked it to add rate limiting and fix a race condition:
# Claude Code session
> /worktale
Worktale narration active. I'll add context to your daily narrative after each commit.
# ...working on features...
$ git commit -m "feat: add rate limiting to upload endpoint"
⚡ worktale: captured (+89 / -12)
$ worktale note "Added rate limiting to /api/upload — previous impl allowed unlimited requests causing OOM crashes in production"
✓ Note added to 2026-03-12
$ git commit -m "fix: race condition in job queue"
⚡ worktale: captured (+34 / -8)
$ worktale note "Fixed race condition — workers claimed same job because claim query wasn't using SELECT FOR UPDATE"
✓ Note added to 2026-03-12
At end of day, I run worktale digest and get a compiled summary: commit stats, module activity, and the AI-narrated notes giving full context. No manual journaling. No trying to remember at 5pm what I did at 9am.
What the agent captures (and what it doesn't)
Does capture
- ✓ Intent — why the change was made
- ✓ Decisions — trade-offs considered
- ✓ Root causes — bugs found, how they were fixed
- ✓ Context — the high-level story
Does NOT capture
- × File paths or line counts
- × Raw diffs or code snippets
- × Mechanical descriptions
- × Source code or secrets
The notes are narrative, not mechanical. They're the kind of thing you'd tell a colleague over coffee, not a changelog entry.
Getting started
# Install or update the CLI
$ npm install -g worktale@latest
# Initialize in any repo
$ cd your-project
$ worktale init
Then install the integration for your AI tool. For Claude Code:
# In Claude Code:
> /plugin marketplace add worktale/worktale-plugin
> /plugin install worktale-plugin@worktale-plugin
# Restart Claude Code, then in any session:
> /worktale
For all 6 platforms with detailed install instructions, see the integration docs.
You can also use it without AI
worktale note is just a CLI command. You can call it manually, from a shell alias, from a CI pipeline, or from a git hook script. The AI integration is the most convenient way to use it, but it's not the only way.
# Manual usage
$ worktale note "Deployed v2.1.0 to production"
$ worktale note "Started migration to new payment provider"
Still local-first. Still private.
Nothing about this changes the privacy model. Notes are text strings stored in a local SQLite database. No network requests. No telemetry. No accounts. Your data stays on your machine.
Privacy isn't a feature toggle. It's the architecture. There is no server to send data to. There is no account to create. The entire database is a single file at ~/.worktale/worktale.db. Delete it and everything is gone.
What's next
The CLI is free and open source forever. We're building Worktale Cloud as an optional layer on top—public developer profiles, cross-repo timelines, AI-powered weekly digests, standup generators. The local tool stays free. The cloud service is how we sustain the project.
But right now, the most interesting thing is this: for the first time, the AI agent that writes your code can also write the story of why that code exists.
Links:
- • Worktale CLI on GitHub (MIT, open source)
-
•
npm package —
npm install -g worktale@latest - • Integration docs for all 6 AI coding agents
- • Full documentation