For solo builders
From zero to one, with the lights on.
021 is one dashboard for everything you're building: projects, ideas, store reviews, revenue and the work in between. Your AI agents can read all of it and pick up tasks across every project, not just the repo they're open in.
Free, bring your own keys. In private use while it takes shape; opening to more builders soon.
Tuesday 21 October
3 things need you
- Paperplane 2.3 was rejected in review
- Ledgerly MRR is down 12% week on week
- 2 one-star reviews waiting for a reply
Portfolio
Which project deserves your next hour?
Every other tool sees one app. 021 sees all of them, so the overview can answer the only question that matters at 8am: a rejected build, a revenue dip, reviews waiting on a reply, sorted by what needs you first. Revenue from the stores and RevenueCat sits next to the commits you put in, so effort and return finally share a screen.
Ideas
A home for “what should I build next?”
- 1 / 3
Capture
Press i anywhere and type one line. Title is the only required field, so the thought survives the moment it arrives.
- 2 / 3
Shape
Attach evidence as it piles up: the fourteen reviews asking for offline mode, a GitHub issue, a link. Evidence is a count, not a stage.
- 3 / 3
Commit
Promote a feature idea to a task, or a new-app idea to a planned project with its pitch as the first brief. Or park it, on purpose.
Agents
Agents do the work. You keep the decisions.
021 runs an MCP server, so Claude Code or Claude Desktop can read your attention list, reviews, metrics and ideas, claim tasks and propose specs, on your own Claude account. What an agent can't do is decide: nothing moves to Ready or Done, no idea gets promoted, without you. Those verbs aren't on the wire at all.
> what should I work on this morning?
⏺ 021 · get_attention()
⎿ Paperplane 2.3 rejected: Guideline 4.3 (spam)
⎿ Ledgerly: 2 unanswered one-star reviews
⎿ Ledgerly MRR −12% week on week
⏺ 021 · get_ideation_context(project: "Ledgerly")
⎿ 14 reviews mention "offline"; 3 of the 2-star ones churned
⏺ 021 · propose_task_spec("Offline mode for receipts")
Spec proposed. It waits in Inbox for you to mark it Ready.
What's new
See what everyone else shipped. Then stop scrolling.
Launches from Show HN, GitHub and Hugging Face. Announcements from the AI labs. New models and price changes. Releases of the packages your own projects depend on. It remembers what you've scrolled past, and when you've seen everything, it says so.
Your stackRelease
@acme/payments 3.0, a major release
Used in Ledgerly and Quietbook
ModelsPrice change
Price cut on a model Ledgerly calls
Input $0.25 → $0.10 per million tokens
Show HNLaunch
A receipt scanner that runs fully on-device
212 points · 48 comments
- You've seen everything new.
Free
No plan to pick and no card on file. 021 exists because I needed it, and it costs little to run.
Bring your own keys
Integrations use keys and connections you create and can revoke at any time. AI runs on your own Claude account over MCP, never on a key of ours.
Your data stays yours
Every row is scoped to your account in the database itself, not just in the app. Credentials are encrypted at rest.
Built in public
Everything that ships is listed in the changelog, dated, with what changed and why.
Connects to
Pulled from the tools your projects already live in.
GitHub
Repos, issues and PRs, plus the packages you depend on
App Store Connect
Listings, review state, ratings and reviews
Google Play
Listings, installs, ratings and reviews
RevenueCat
MRR, subscribers and trial conversion
Claude Code & Desktop
The whole portfolio over MCP
StripeSoon
Web revenue