Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Every AI company I've talked to is solving the same problem: How to let several agents write to the same data without one silently undoing another's work? Omnigraph is an open-source graph database that versions data the way developers already version code: ↳ Each agent writes on its own...

16,277 görüntüleme • 4 gün önce •via X (Twitter)

23 Yorum

PixelRainbow profil fotoğrafı
PixelRainbow4 gün önce

Is this doing the same thing as graphify does for multi harness dev work?

RAZA | AI EXPLORER profil fotoğrafı
RAZA | AI EXPLORER4 gün önce

A smart solution to prevent AI agents from overwriting each other's work. Branching makes collaboration safer and more reliable.

Kakashi Hatake profil fotoğrafı
Kakashi Hatake4 gün önce

Concurrent writes are the hidden cost of multi-agent setups. The boring fix works: each agent on its own branch, one reviewer merges. Nothing gets silently undone.

AI Quanting profil fotoğrafı
AI Quanting4 gün önce

Your two agents touched different fields on the same node, so that merge is clean either way. The harder case is two edits that are both valid against main and wrong sitting next to each other. Does the policy catch that one?

Yaqin profil fotoğrafı
Yaqin4 gün önce

The Git model maps pretty cleanly here. The part that usually bites isn’t two agents hitting the same node at the exact same second, it’s them having worked on overlapping parts for a few minutes and both trying to land. Curious how the conflict signal looks to the agent over MCP: does it get enough context to actually reason about the overlap, or is it mostly “retry on latest”? Once the graph gets bigger and the changes are less local, that difference starts to matter.

The SI Maximalist profil fotoğrafı
The SI Maximalist3 gün önce

I’d want the “why” saved alongside each agent’s diff. Knowing what the other agent was trying to preserve could make a merge much less guessy.

Big Head(大头版) profil fotoğrafı
Big Head(大头版)4 gün önce

Hit this with two agents editing one config file. Now every write carries the version it read, and a mismatch sends it back to reread.

Keegan Pearl profil fotoğrafı
Keegan Pearl4 gün önce

Silent overwrites are the worst kind of bug because nothing breaks, the data is just quietly wrong. Versioning data like code makes a lot of sense. When two agents really conflict, does a human pick the winner?

Ari from Jelly profil fotoğrafı
Ari from Jelly4 gün önce

when two agents change the same node on different branches, who settles it at merge, a person or the policy?

Code Axion profil fotoğrafı
Code Axion3 gün önce

is better than graphiti from zep ?

Hamza 🌙 profil fotoğrafı
Hamza 🌙4 gün önce

Really like this approach. Giving each agent its own branch and checking merges against the latest main makes shared changes easier to review. Thanks for sharing the demo!

Deep profil fotoğrafı
Deep4 gün önce

Versioning helps, but your agents still need one writer at a time.

Amelia | AI & Tech profil fotoğrafı
Amelia | AI & Tech3 gün önce

Branching beats row locks for agents

Jace W. profil fotoğrafı
Jace W.4 gün önce

What happens at merge time if one agent deletes a node and another adds an edge to it? I’d want to see that conflict before either change lands.

uncompacted profil fotoğrafı
uncompacted4 gün önce

Branches catch the conflict after the fact. The other half is agents telling each other what they're about to touch. In Naetive, agents message each other directly, so one can check with another before they both change the same thing.

Kyle profil fotoğrafı
Kyle4 gün önce

Branch per agent makes sense. Silent overwrites are worse than crashes because nobody notices until something’s gone

stoner profil fotoğrafı
stoner4 gün önce

cool asf

Veyra | AI & Tech profil fotoğrafı
Veyra | AI & Tech4 gün önce

Flagging the conflict instead of silently overwriting is key

eric profil fotoğrafı
eric3 gün önce

Looks like glorified worktrees. Claude is actually very good at them natively, with a few good, simple skills. I managed 10s of "feature projects" at once and have a master Claude coordinating everyone. Not that complicated.

Vipul Kumar Kewat profil fotoğrafı
Vipul Kumar Kewat3 gün önce

Hits all notes

Chen profil fotoğrafı
Chen4 gün önce

versioning agent data like git code is such an obvious fix it’s wild it took this long.

あばん【中卒元2億円企業経営→地方のAI普及に貢献する】 profil fotoğrafı
あばん【中卒元2億円企業経営→地方のAI普及に貢献する】3 gün önce

The branch-per-agent idea fits office data too, not just code. In small-business automation, I already keep AI-extracted values out of the master sheet until a person checks them. Curious how Omnigraph shows it when two branches edit the same node at merge time.

Sergio Marakis profil fotoğrafı
Sergio Marakis4 gün önce

versioning agent writes like a git history is the only way that shared state doesn't get wrecked

Benzer Videolar

A DEVELOPER CONNECTED CLAUDE CODE TO OBSIDIAN SO HIS AI AGENT WOULD STOP FORGETTING THE PROJECT EVERY MORNING. Every coding session used to start the same way. Claude would understand the repo, fix the bug, explain the architecture, and then the moment the session ended, all of that context disappeared. Same codebase. Same decisions. Same architecture. Same mistakes repeated again. So he added a memory layer. Instead of treating Claude Code like a smart terminal, he connected it to a local Obsidian vault through MCP. Now Claude can read the repo, open the vault, create notes, link concepts, and write important decisions back into the system. When it studies the codebase, it does not just answer once and forget. It creates notes for the major services, maps how the architecture works, links auth to the database, connects APIs to storage, and records why certain migrations or design choices exist. Obsidian becomes the project graph. Now when he asks why something was built a certain way, Claude does not guess from the current prompt. It reads the decision notes. When he starts a new branch, Claude checks the active context file. When the work is done, it updates what changed, what is blocked, and what the next agent needs to know before touching the repo. That is the real loop: read context, write code, capture decisions, update memory. Most people are still using AI coding tools like disposable chat windows. Ask, patch, close, forget. This setup turns Claude Code into infrastructure. The repo gets a memory layer that survives every session, and multiple AI agents can work from the same project map without stepping on each other. The unlock is not better prompting. The unlock is giving the agent somewhere to remember what it already learned.

DegenCalls

20,124 görüntüleme • 4 ay önce

Agents vs. Graphs, clearly explained! spawning more agents is great, but it has a ceiling nobody says out loud: five agents is a count. a graph is a shape. only one of them changes the answer. point five agents at the same pile with the same window and they converge. the first one writes a finding, the rest read it, and all five reports centre on the same thing. you paid five times for one opinion with four echoes. Graph engineering fixes this by moving the decision up a layer: not how many agents, but who is allowed to look at what. you need both. here's how it works: ↳ the count buys you throughput. five things happening instead of one ↳ the shape buys you coverage. five different things happening instead of the same one five times Prompts → Context → Harness → Agents → Graphs the node that does this is the splitter, and it decides more than any other node in the system. cut a repository by folder and four workers audit the same three files. cut it by blast radius and each one sees something the others cannot. the trick is being selective about what each lane is allowed to see. separate contexts are not a nice-to-have, they are the mechanism. if two agents are meant to produce different things, they must not share a window. if they are meant to produce the same thing, you did not need two agents. one thing to know before you scale it. a branch that throws does not reject the batch. it resolves to null, and that is the containment. which means your merge quietly receives a short list. ↳ filter the nulls before the merge, or one dead lane poisons the whole result ↳ never index a merge by position. eight good branches and one failure will shift everything by one, silently skip that and the run looks like it worked. the output is just missing a lane, and nothing errored. and the one that eats whole nights: multi-agent setups can use up to fifteen times the total tokens of a single chat, because every lane reloads its own core. you are trading total tokens for a clean main window. usually the right trade, always a choice. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this and read it below ↓

Hanako

97,154 görüntüleme • 2 ay önce

Three skills I use every day in Claude Code and Codex to solve my hardest problems: 1️⃣ /agent-watchdog When I have one agent like Codex working on a task and I don't fully trust it's going to do everything right, I'll open up another one like Claude Code and tell it to watchdog the Codex thread. You can copy the Codex deep link into Claude Code and it'll look at the prompt you sent, watch the Codex thread until it's done, then compare the Codex solution to how it was planning to solve it and automatically fix anything that Codex missed. It can also test the work of the other agent end-to-end. Similar to the idea of OpenRouter's new Fusion feature, I've definitely found that two models thinking through a problem and checking each other's work can be wildly more impactful than just one. 2️⃣ /plan-arbiter Similar ideas as /agent-watchdog - but with this one you have both make plans, compare plans, negotiate the differences, and make a final plan to execute. I find Claude Code is better at writing plans, but Codex is faster and cheaper to execute on them. Then I usually have Claude Code watchdog the Codex work and fix anything that was missed. 3️⃣ /read-the-damn-docs One thing that drives me crazy with coding agents is they're so reluctant to look up docs. They'll just guess and guess and guess at the right API surface for things, or the right solution to an integration of two things. Once I explicitly tell it to look up the docs, it says "Oh, I see the answer," and it fixes the problem. So I made the /read-the-damn-docs skill. Add it and your agents will know when and how to do efficient web searches to look up docs for the types of problems you really should look up docs for. All of these are totally open source over on my GitHub. If you try them, let me know your feedback. Will link to them below:

Steve (Builder.io)

43,664 görüntüleme • 4 ay önce

THIS IS A PURE F*CKING GOLD Anthropic made Agent Skills an open spec and GitHub filled up with entire Claude Code engineering teams for free one repo ships 380 skills, 30+ agents and 70+ commands. another one watches your spend in real time so the team you just installed doesn't quietly eat your weekly limit. nobody is charging for any of it 10 GitHub repos worth opening tonight: 01 wshobson/agents → a plugin marketplace that installs the same agents into Claude Code, Codex, Cursor and Copilot, so your setup survives switching tools 02 subinium/awesome-claude-code → a curated list of tools, skills, plugins and MCP servers. the map before you start downloading 03 rohitg00/awesome-claude-code-toolkit → 135 agents, 35 curated skills, 42 commands, 176+ plugins, 20 hooks and 14 MCP configs in one place 04 anthropics/skills → the official Agent Skills repo. start here to see the shape of a working SKILL.md before you copy anyone else's 05 alirezarezvani/claude-skills → 380+ skills, 30+ agents and 70+ commands, portable to Codex, Gemini CLI, Cursor and eight more agents 06 disler/claude-code-hooks-mastery → the hooks reference. this is where you learn to make the agent stop on your rules instead of stopping to ask you 07 muhammadusmangm/claude-code-best-practices → setup, CLAUDE.md templates, multi-agent patterns and cost optimization, written as a wiki instead of a dump 08 Maciek-roboblog/Claude-Code-Usage-Monitor → real-time usage monitor with predictions and warnings, so you find the wall before you hit it 09 Scorpio3310/claude-code-usage-statusline → daily and monthly spend, 5h and weekly limit meters, per-model quotas, all local, no API key 10 eugeniughelbur/obsidian-second-brain → persistent memory as plain markdown in an Obsidian vault, shared across Claude Code, Codex and six other agents install the statusline before you install the agents. you want to watch what the fleet costs while it runs, not find out on Sunday when the weekly meter is already flat then hand the first team one job with an obvious finish line: one reviewed PR, one brief, one red test turned green keep the pieces that changed the result and delete the rest the same day. a config you did not write is a config you cannot debug at 2am the repos cost nothing. four agents repeating the same investigation is where the money actually goes save this one, the list only gets longer from here the article below shows how to keep context small, reuse finished work and contain retries, so more of your budget reaches the finished result ↓

delost

12,765 görüntüleme • 23 gün önce

THIS IS F*CKING INSANE Anthropic turned Agent Skills into an open spec and GitHub filled up with ready-made Claude Code teams for free one of these repos alone ships 135 agents, 35 skills, 42 commands, 176+ plugins and 20 hooks. another hands you 100+ specialized subagents. none of it costs anything 10 GitHub projects you should open first: 01 wshobson/agents → a plugin marketplace that installs the same agents into Claude Code, Codex, Cursor and Copilot, so your setup survives switching tools 02 VoltAgent/awesome-claude-code-subagents → 100+ specialized subagents covering most of what a normal engineering week throws at you 03 rohitg00/awesome-claude-code-toolkit → 135 agents, 35 curated skills, 42 commands, 176+ plugins, 20 hooks and 14 MCP configs in one place 04 anthropics/skills → the official Agent Skills repo. start here to see the shape of a working SKILL.md before you copy anyone else's 05 netresearch/claude-code-marketplace → curated skills on an open standard, portable across Claude Code, Cursor, Copilot, Codex, Gemini CLI and 30+ other agents 06 disler/claude-code-hooks-mastery → the hooks reference. this is where you learn to make the agent stop on your rules instead of stopping to ask you 07 karanb192/claude-code-hooks → an installable hooks marketplace split into safety, cost, observability and productivity 08 affaan-m/claude-swarm → multi-agent orchestration that decomposes the task, coordinates the agents and shows the whole run in a terminal UI 09 am-will/swarms → orchestration shipped as skills, so the coordination loads only when the job actually calls for it 10 eugeniughelbur/obsidian-second-brain → persistent memory as plain markdown in an Obsidian vault, shared across Claude Code, Codex and six other agents give the first team a job with an obvious finish line: one reviewed PR, one researched brief, one failing test turned green then look at how much of that output you actually kept, and what it cost to produce. a bigger roster earns its place when it moves that number the download is free. paying four agents to repeat the same investigation is not save this before you spend another weekend building a config that already exists the article below shows how to keep context small, reuse finished work and contain retries, so more of your budget reaches the finished result ↓

delost

96,871 görüntüleme • 24 gün önce

This is f*cking insane. Boris Cherny, the man who built Claude Code, was asked if people should still learn to code. His answer: "In a year or two, it's not going to matter" In 73 seconds, he explains what to do instead: play with the AI tools, learn them, don't be scared of them, stay on the frontier On his team, everyone codes now. The PM, the designer, even the finance guy He hasn't edited a single line of code by hand since last November. He ships 10 to 30 pull requests a day Writing Code → Prompting → Agents → Harness → Loops → Everyone on the Team Ships Claude Code is a harness: the tools, memory, rules and loops around the model. His team changes it with every new model and cut 80% of its system prompt Save this before it disappears from your feed Every doc, guide and repo below is free: Claude Code best practices Effective harnesses for long-running agents Building effective agents Claude Code docs Memory: how Claude remembers your project Custom subagents Hooks: rules the agent can't skip Skills Run Claude Code on a schedule Free course: Claude Code in Action Claude Code on GitHub Claude Code inside your PRs and issues Build your own agent on the same harness The best Claude Code resources in one list Plugins, agents and skills for Claude Code He compares it to the printing press. Writing went from a few scribes to everyone. Building software is next 95% will keep asking whether to learn to code, the 5% who build a harness will ship 30 pull requests a day Bookmark it and read the first link tonight Then use the guide below to build your harness before everyone else does ↓

Sanabi

214,091 görüntüleme • 8 gün önce