Entire retweeted
Git was built for developers. What does it look like when AI agents start writing the code?
@ashtom, former GitHub CEO and now co-founder and CEO of @EntireHQ, is taking the AGNTCon + MCPCon NA stage with “Building the Git Forge of the Future.”
Live now: bit.ly/4dVkjdO
Your agent shipped the feature. Who has the context to review it?
Joseph Katsioloudes and Peyton Montei of @EntireHQ join AI DevCon NYC with a hands-on workshop on keeping context across agent sessions.
You can use TX20 to knock 20% off your ticket: luma.com/aidevcon-nyc2026?co…
Leave your agents running and come have an entire espresso with us. ☕️ Or a pragmatic donut. 🍩
Join @ashtom and @GergelyOrosz for a pop up sit down to talk about source control, code review, and agents.
San Francisco, Thursday October 29, 9am.
Space is limited: luma.com/0m8oub77
Starting today, you can use an external agent plugin to tie your commits to your @grok history, so you can review sessions right alongside your code.
Then ship away with grok @bot as part of your team.
entire.io/blog/the-entire-cl…
Entire retweeted
is that $45 SF Restaurant Week dinner a good deal?
w/ Codex + Claude Code, i built an app to compare SF Restaurant Week menus, prices, and ratings without opening 20 tabs and PDFs.
you can even see the agent sessions behind the build w/ @EntireHQ
find the best set meal 👇
Entire retweeted
Replying to @lizziepika @EntireHQ
They're the best, can't live without them now 💯
If anything, they're too good!
With Entire. One API for code + context. entire.io
Entire retweeted
I see a tweet, I make a video.
Track your Grok sessions (prompts, outputs, tool calls) w/ @EntireHQ!
Setup + connecting Codex/Claude Code in 🧵
hey @EntireHQ, any plans to hook Grok? i turned Entire on for my repo, but today's commits came from Grok, so the recap shows 0 sessions and 0 checkpoints.. would love to see those sessions in there too..
Entire retweeted
Celebrate Walktober🚶♀️ w/ a tour of San Francisco’s staircases
Video edited by Codex, see prompts, outputs, tool calls, etc used w/ @EntireHQ entire.io/gh/elizabethsiegle…
Our #neatHack submission
Lognac is an agent built to catch medicine discrepancies when a patient visits and leaves the hospital.
It compares medication history, conditions and allergies with hospital orders, discharge prescriptions and take-home instructions.
It flags dose changes, missing medicines and safety concerns, linking each finding to its source.
A clinician documents the decision; Lognac then rechecks the documents and prepares a timestamped report, follow-up reminder and optional evidence graph. The prototype uses fictional UK records.
each and every run traced in neatlogs, Entire mapped code affected by Lognac’s fixes and the business plan mapped out in CFOai with codex.
Built with @arithecfo @EntireHQ @neatlogs
xitter.cf/bettersayAJ/status/210…
building agents is easy
getting them to production is hard
today we’re launching @neatlogs to fix that
Entire retweeted
our #neatHack submission 🧪
i built parallax because i kept seeing the same problem with ai research. it sounds confident even when it’s wrong.
so i wanted to see what happens when two agents research the same question independently, then have a third settle disagreements and another check every quote against the original source.
7 agents handle the whole job, from pricing to delivery.
building it taught me a lot. @neatlogs helped me catch bugs i missed, including rate limits breaking jobs in production. fixed those, and we went from 39% guesswork to 0%, with 14/14 known facts correct.
@EntireHQ checkpointed the build, and @arithecfo helped turn my actual costs into a business plan with ~83% gross margins.
built this with @myacxt
try it: useparallax.xyz
code: github.com/neromtoobad/paral…
building agents is easy
getting them to production is hard
today we’re launching @neatlogs to fix that
Entire retweeted
Jangkau: an AI agent that fixes web accessibility, not just audits it. Every edit is re-checked in a real browser, rolled back if it breaks.
186 failures → 0 on 4 real pages, ~$0.09/page.
@neatlogs @entirehq @arithecfo
github.com/PugarHuda/jangkau #neatHack xitter.cf/neatlogs/status/210900…
Entire retweeted
Our #neatHack submission
We built Wincheck: an agent that takes a Windows-only Python bug report, reproduces it on a real Windows machine, fixes it, and proves the fix with the maintainers' own tests.
Built with @arithecfo @EntireHQ @neatlogs
The problem: most Python maintainers don't run Windows, so Windows-only bugs can sit open for months. Nobody on the team can even see them fail.
How it works. Three phases, and the harness enforces each one, not the prompt:
1. Read: the agent reads the issue and the code
2. Reproduce: it can only write test files, and every test it submits must fail
3. Fix: tests are frozen, and the fix only counts if the reproduction passes and nothing that passed before breaks
No failing test, no code change.
Results, all on GitHub's Windows runners:
• 6 real upstream bugs from click, tox, pytest and build
• 4 fixed, including 3 of 4 held-out bugs the agent never saw while I tuned it
• Scored with the maintainers' own tests from their fix commits
• Run again for consistency: 3/3 on the held-out pytest bug, 2/3 on click
• 27 harness tests on Windows and Linux CI. Disable any gate and a test fails
• The misses are logged with the reason, not hidden
How the partner tools fit in:
@neatlogs: every run is one trace, with setup, baseline, the agent's three phases, every Claude and tool call, and the score. It surfaced three incidents while I built:
• Each run was split into 4 traces. One missing parent span, fixed
• "Slow Project Preparation" on every run. Switched installs to uv, and time outside the agent went from ~50s to ~34s
• "Execution Failed" on runs where the gates refused out-of-phase writes. The refusals worked
Zero open incidents now.
@EntireHQ: 9 public checkpoints, so you can see the agent sessions behind the commits. Emails and phone numbers are redacted before anything is published.
@arithecfo: a 12-month plan built from measured run costs. $49 per successful fix or $199/month for 10 fixes per repo. Both clear a 70% margin, and 23 paying repos cover a founder's salary.
Try it yourself: fork the repo and run it on GitHub's Windows runners, or add it to Claude as an MCP server.
Repo: github.com/nftkingiii/winche…
Every claim links to a run you can open: docs/evidence.md
Demo below 👇
#neatHack
building agents is easy
getting them to production is hard
today we’re launching @neatlogs to fix that
Entire retweeted
Introducing GAUNTLET: an adversarial chaos engineering platform that tests AI agents under failure scenarios and scores their resilience.
Built solo in 2 days at #neathack. Thanks
@neatlogs
@EntireHQ
@arithecfo
and my AI agent for the chaos help. Really thanks to all mentors!
building agents is easy
getting them to production is hard
today we’re launching @neatlogs to fix that
Entire retweeted
Entire is everything you probably think Github should be...but better! Trails are the future 👏
Going live with @blackgirlbytes in 20min
She’s walking us through Entire (the new platform from GitHub’s former CEO) and taking your toughest AI questions live.
Join us and 70k of your new best friends: leonnoel.com/discord
Entire retweeted
AI code generation needs better context!
Got to tune in for @blackgirlbytes demo of Entire: The bottleneck is reviewing the generated code. Connecting diffs directly to the AI agent sessions behind them gives us the why, not just the lines changed.
Entire changing the game!
Going live with @blackgirlbytes in 20min
She’s walking us through Entire (the new platform from GitHub’s former CEO) and taking your toughest AI questions live.
Join us and 70k of your new best friends: leonnoel.com/discord
Independence Day is coming: log into Entire without needing a GitHub account 👀
entire.io/blog/entire-dispat…
RT @Patrick_Beninga: I just released the demo of my game!
Super happy with how it turned out, Check it out and download the demo here: htt…
Code review used to mean reading the diff. Now it means reading the diff and guessing what the agent was thinking. 🤔
The fix is sharing sessions. In this blog, @HaimantikaM shares how she writes our docs with her coding agent, then shares the whole session with reviewers. Prompts, responses, tool calls and file changes, all tied to the commit.
Don’t worry, she redacts the prompts where she yells at her agent. 🤖
Sharing sessions is the new social coding 👇
entire.io/blog/sharing-sessi…






















