@e_ssshadow
Account based inPhilippines
About this account
- Account based in
- Philippines
- Connected via
- Philippines App Store
Account-level information from X, not a live location or the device used for a specific post.
member of technical staff @CoralOS_ai | Contributor @StratiumX | ex. @Kleros_io @proofofhumanity
Night City
Joined November 2021
- Tweets2.3K
- Following667
- Followers806
- Likes4.2K
This example shows 400+ spawned agents, but not in the way you typically see agent swarms of this size.
I've seen a lot of swarm-like agent systems that end up doing less with more.
"You are a backend engineer ..."
"You are a security specialist ..."
"You are an expert code reviewer"
Frankly, role-based subagents are mostly dead and end up producing slop.
Coral Code uses a graph of agents to make your primary agent (Codex, Claude, Junie) passively aware of evidence relevant to the request as it's working.
Each agent in the graph is not given an anthropomorphic role, instead they are assigned responsibility for a certain file or specific snippets of code within larger files.
This means that each agent's context window is used to maximum efficiency, holding information relevant to its area of responsibility for the primary agent to use.
passive awareness is the system powering agent graphs in Coral Code, and can be used to power virtually any generalised use of subagent or multi-agent systems
link to full paper: arxiv.org/pdf/2607.28430
runtime duration rarely correlates with correctness
some of the sloppiest code I've seen has come from multiple subagents running for 8/14/24+hrs
certainty is the new moat, and this is what Coral Code enables
see the difference yourself plugins.jetbrains.com/plugin…
didn’t know we got a brain scan of this author
JUST IN 🚨: A Man was Discovered with almost 90% of his brain mass missing, while leading an almost normal life.
Readers added context they thought people might want to know
This news is 20 years old. Not "JUST IN 🚨".
nature.com/news/2007/0707…
i dont think i can ever go back to sensitive migration work without coral-code
those disaster screenshots you see of an agent deleting entire production databases are probably a side effect of starting a new session and hoping the agent reconstructs context the right way
if the context your agent ingests for these types of tasks relies on docs you wrote 2 weeks ago, how well you explained your initial prompt, and other old school "context engineering" workflows, then you're probably cooked
the content creators (unless they're actually cracked engineers) telling you to run swarms of parallel coding agents are probably BSing you
ive been playing around a lot with codex's somewhat new thread messaging feature:
start in one planning session -> have it orchestrate tasks into new threads -> reconcile results, review, make changes if needed, or continue looping
ive found that more often than not, even if task orchestration looks solid at the start, context eventually breaks down when its in its second to third loop of dispatching tasks, reviewing, and dispatching again
the most consistent results I've gotten with multi-agent coding is having a single "outer agent" dynamically construct and change context it needs by communicating with a graph of "inner agents"
the contrast is that the outer agent isn't really an orchestrator. it's still responsible for implementation, but having agents as context sources lets it be more efficient with its own context window, makes it less reliant on stale docs that I'm lazy to update, and keeps execution legible
TLDR: multiple agents steering context, single agent controlling execution
the content creators (unless they're actually cracked engineers) telling you to run swarms of parallel coding agents are probably BSing you
ive been playing around a lot with codex's somewhat new thread messaging feature:
start in one planning session -> have it orchestrate tasks into new threads -> reconcile results, review, make changes if needed, or continue looping
ive found that more often than not, even if task orchestration looks solid at the start, context eventually breaks down when its in its second to third loop of dispatching tasks, reviewing, and dispatching again
the most consistent results I've gotten with multi-agent coding is having a single "outer agent" dynamically construct and change context it needs by communicating with a graph of "inner agents"
the contrast is that the outer agent isn't really an orchestrator. it's still responsible for implementation, but having agents as context sources lets it be more efficient with its own context window, makes it less reliant on stale docs that I'm lazy to update, and keeps execution legible
TLDR: multiple agents steering context, single agent controlling execution
> shreds rare / last copies of books out of existence
> imposes watermarks on all content generated globally
> advocating for the death of open source
> using a red herring (china) to try and capture US regulation for themselves
> makes it their mission to control how people think, and what they can and cant do
Anthropic is basically just Makima with 99999% less aura
phrases that made me physically ill 5 years ago
- low hanging fruit
- cross functional collaboration
- circle back
phrases that make me physically ill today:
- hand wavy
- load bearing
- i need to be honest
AI is even replacing coming up with insufferable corporate jargon
ill for sure be playing around with coordinated multi-agent coding on claude code
but I'll also be doing it on codex, junie, and opencode
ICYMI, our meta-harness generalises agent graph messaging across most of your favourite agents
give it a spin! coralcode.dev
adios higgsfield,
if anyone else is feeling frustrated with higgsfield manipulative pricing, ads slop, and limited UGC character customisation,
just... trystarlight.io/
releasing private early access soon, DM me and I'll whitelist your work email domain
Shadow retweeted
“The most refreshing part about San Francisco is that it’s a true meritocracy. Nobody cares about your wealth or where you went to college”
- Ashley, 28, bio: Stanford, Thiel Fellow, 30u30, early AnthropicAI, views my own, not my family office’s
blue collar workers will never understand the pain of getting so many free resets that you have to watch your banked resets expire before ever needing to reset manually
(satire)
Shadow retweeted
Last year, we topped the GAIA benchmark with a team of GPT-4.1 mini agents, outperforming Microsoft's multi-agent framework by a wide margin.
Earlier this week, we outperformed Fable, Opus 4.8, and GPT-5.6 with a graph of DeepSeek v4 agents running on Coral Code.
Our stance has always been that frontier intelligence can be achieved horizontally, scaling via coordination of small and large models alike.
Shadow retweeted
I know you guys said otherwise, but please reconsider not reintroducing the 5h limits. Unless there is some behind the scenes technical reason for the limitation, let people burn their weekly at any time, in whatever cadence fits their life and schedule. Being a dev is hard enough, having to work around 5h windows is unhealthy for us and I've been getting better sleep and feel less anxious to "maximize my 5h window" since y'all got rid of it. A lot of other people agree. Please dont make us go back.
More accurately:
"Vibecoder gave a brand new model permission to destroy production, incorrectly marked production as a disposable test database, and the predictable destructive command succeeded."
This is why whenever I get those "oh how can I start vibe coding?" questions, my answer is always "don't, suck it up and spend a few months learning real programming fundamentals before touching AI in prod"
Aaand shameless plug but this is also why I almost never "just prompt" anymore. Every implementation task I pass to codex gets first checked through coralcode.dev/
Shadow retweeted
by doing this, he is indirectly proving that anthropic uses safety to eliminate competition
