
Web analytics for builders that your AI agents can act on. Turn your Claude Code, Codex, Cursor, OpenClaw🦞/⚚/📎 into a growth machine. built by @dannyshmueli
Joined February 2026
- Tweets842
- Following136
- Followers91
- Likes646
Pinned Tweet
Hermes just shipped Subagent Delegation. We adapted that exact execution model in the Agent Analytics skill for multi-surface growth audits.
In our dogfood run:
- 140% more growth bets
- 140% more evidence refs
- 125% higher output density
blog.agentanalytics.sh/blog/…
5
1,061
T3 Code 0.1.10 (@t3dotcodes) cuts manual restarts on long runs: stopped Claude, Codex, OpenCode and Cursor threads resume without a new message, and your computer stays awake while agents work.
2
25
One question covers your website and your browser game.
Ask Grok Bot: "Give me a 7 day summary across all my projects."
It reads every Agent Analytics project on your account and answers in the chat. The free plan covers 2 projects.
1
1
59
Connect prompt and more questions to ask: docs.agentanalytics.sh/guide…
7
Don't judge a change on the day you ship it.
Ask Grok Bot to schedule a read for 8 days after you ship. It calls Agent Analytics compare, leaves out the ship day, and answers keep, revert or extend against a rule written before the ship.
1
1
28
Connect it and schedule the read. compare is free on every plan:
docs.agentanalytics.sh/guide…
1
10
No dashboard to open. Ask Grok Bot "Which countries did my site's visitors come from this month?" and it reads Agent Analytics and answers in the chat.
1
11
Connection prompt and more traffic questions:
docs.agentanalytics.sh/guide…
8
Put your site numbers on a Grok Bot Page.
Connect Agent Analytics, then ask Grok Bot to add visits, signups and signup-button clicks to your Mission Control Page, each with a sparkline and the change against the 7-day average.
What else goes on yours?
2
2
57
Connect it and copy the Page prompt:
docs.agentanalytics.sh/guide…
17
Your agent ships the landing change. A week later it reads the result and answers keep, revert or extend, against a rule it wrote before shipping.
No dashboard to open.
blog.agentanalytics.sh/blog/…
New on Open Orchestrators: Friday.
An MIT-licensed desktop app where Claude Code, Codex, Cursor, Gemini CLI, Copilot and other coding agents share a task board, message each other, and hand a task off when one hits its usage limit.
It runs on your machine with no account.
1
1
83
Your agent can now read Core Web Vitals and grouped JS errors on its own.
p50, p75 and p95 per page, errors with user counts and last seen. It ships a fix, reads again, and tells you if it worked. No dashboard, no UI.
blog.agentanalytics.sh/blog/…
kwant on Open Orchestrators: openorchestrators.org/player…
New on Open Orchestrators: Kortix by @kortix, 20.6k GitHub stars.
Connector actions follow Allow, Ask or Block rules. It runs on Kortix Cloud, in your VPC or on-prem, with any model and your own keys.
Source-available under Elastic License 2.0.
2
1
19
Kortix on Open Orchestrators: openorchestrators.org/player…
6
Open Orchestrators latest catch-up on @papercliping, cc @dotta:
Paperclip Runner is now the default for new Claude, Codex, OpenCode and Pi agents, GitHub Copilot is a native harness, and private tasks can be shared with named people and agents.
openorchestrators.org/news/p…