Aditya Arya

I build tools for people who run coding agents. MindSync for routing and memory. Agent Notch for live quota.

[Aditya Arya @adityarya_ai]

I build the tools that make AI coding agents work together, and one product for everyone else:

→ MindSync: one control plane for your coding agents (Claude Code, Codex, Gemini, Grok, Cursor…). Routes subtasks to the right agent, locks files so they don't collide, hands off when one runs out of quota, and keeps memory across sessions. v1 open source, v2 in the works.

→ Agent Notch: a desktop HUD that shows your AI coding agents' live quota at the screen edge. v1 out.

→ gws-marketing: an MCP server that lets AI agents read your Search Console + GA4. Open source.

→ VedPatrika: Vedic astrology reports in Hindi + English, calculated with Swiss Ephemeris. Every conclusion shows its reason. Live.

Solo. Learning by shipping. Links below 👇

[Aditya Arya @adityarya_ai]

@mitansh_j07 Yep, file locking was the first thing that broke. Two agents on the same file = silent overwrites. Fixed it with per-file locks and isolated worktrees per worker. Do you run more than one agent at a time?

[Aditya Arya @adityarya_ai]

I build the tools that make AI coding agents work together, and one product for everyone else:

→ MindSync: one control plane for your coding agents (Claude Code, Codex, Gemini, Grok, Cursor…). Routes subtasks to the right agent, locks files so they don't collide, hands off when one runs out of quota, and keeps memory across sessions. v1 open source, v2 in the works.

→ Agent Notch: a desktop HUD that shows your AI coding agents' live quota at the screen edge. v1 out.

→ gws-marketing: an MCP server that lets AI agents read your Search Console + GA4. Open source.

→ VedPatrika: Vedic astrology reports in Hindi + English, calculated with Swiss Ephemeris. Every conclusion shows its reason. Live.

Solo. Learning by shipping. Links below 👇

[Aditya Arya @adityarya_ai]

MindSync → https://t.co/yunzSZQ9pm
Agent Notch → https://t.co/zmcwS9cXki
gws-marketing → https://t.co/YoLINDMIbb
VedPatrika → https://t.co/SKapfvAuW5

[Aditya Arya @adityarya_ai]

@lydiahallie Claude is back in the game again im loving opus 5.5

[Aditya Arya @adityarya_ai]

@HeyyyyShubham @kkarthikkgs Maybe that’s what his company does too 😛

[Aditya Arya @adityarya_ai]

@theroundpeg13 @kkarthikkgs Well, you definitely just gave me an idea worth building. Might actually give this a shot 👀

[Aditya Arya @adityarya_ai]

That wasn’t my question. If background checks can fail, improve the checks. If someone misbehaves, fire them and involve the police where appropriate.

But please explain how being born in UP, MP, Bihar or Rajasthan becomes a more reliable predictor of sexual misconduct than an individual’s actual behaviour and history. And if geography is the risk factor, why are women from the same four states still eligible?

You’re not answering the contradiction - you’re just changing the subject.

Also, ₹500 to a constable? That’s quite an allegation to casually make in public. Maybe we should ask your state police department whether they agree with your assessment. And if one constable ever takes a bribe, I suppose you’ll blacklist the entire police department next?

[Aditya Arya @adityarya_ai]

@martyamark https://t.co/YoLINDMIbb

Try this one

[Aditya Arya @adityarya_ai]

@sridharfyi Will be happy to join

[Aditya Arya @adityarya_ai]

In my opinion google improved the performance as well with gemini 3.8 flash in comparison to their older models im using it regularly for heavy tasks that burn more usage on codex or claude using mindsync where codex orchastrates it and still feel the diff with output takes less babysitting and follows instructions well in giving better output

[Aditya Arya @adityarya_ai]

@CumulativeWeb @quxiaoyin Local-first + read-only is the combo I actually trust. I'll run `trust_verdict` on my own first and see what it returns.

[Aditya Arya @adityarya_ai]

@quxiaoyin I built and shipped these myself — real production systems, not mockups.

Portfolio → https://t.co/KmejsRpfh7

Highlights that matter:
• MindSync AI — local-first MCP orchestrator that routes coding-agent work by capability, holds file locks so agents don't collide, keeps cross-session memory. 5.5k+ PyPI downloads, v1.9.0 live.
• AskPanditji — full-stack Vedic astrology product, my own deterministic Swiss Ephemeris engine, bilingual reports. Live & selling.
• Agent Notch — Windows HUD (Electron/React 19) tracking live token quotas across Claude, Codex, Gemini, Grok, Cursor.

High agency: I design, build & ship end-to-end — math, daemons, UI. Judge it on results.

[Aditya Arya @adityarya_ai]

@GithubProjects Here it is https://t.co/wN0lpwzlXY

[Aditya Arya @adityarya_ai]

@s7anit Localhost is what I'd poll too. Shipping --json first though no extra process, easy to maintain. If people actually hit it, the port is a thin wrap on the same dump

[Aditya Arya @adityarya_ai]

Built Agent Notch: a strip on the right edge of your screen showing live quota for every coding agent you run.

Codex, Claude Code, Cursor, Antigravity, Grok, OpenCode — read straight off your local credentials.

Shipped 4 days ago. 8 releases since.

What real use changed 👇

[Aditya Arya @adityarya_ai]

@s7anit Threshold already in Settings. JSON today is inbound- custom CLIs print quota, Notch reads it. Outbound dump isn't there yet. If I add it, would you rather notch - -json or a localhost port?

[Aditya Arya @adityarya_ai]

@divvsaxena Ive been following your work since few days… You’re doing great buddy. Wish you all the success 🤞🏽

[Aditya Arya @adityarya_ai]

@staskulesh Good job ive just completed 100 followers 🤞🏽😅

[Aditya Arya @adityarya_ai]

Im 32.

Solo builder from India.

Looking to connect with builders and founders worldwide.

[Timo 🌱 @TimoBuilds_]

I'm 33.

Solo founder from Germany.

Looking to connect with founders & indie hackers.

[Aditya Arya @adityarya_ai]

@0xPaulvibe Heyy paul happy to connect

[Aditya Arya @adityarya_ai]

@staskulesh hey Stas saw the reintro. 6666 followers is a fun number to land on…