༼ つ ◕_◕ ༽つ Gib ready retweetledi
༼ つ ◕_◕ ༽つ Gib ready
1.4K posts
༼ つ ◕_◕ ༽つ Gib ready retweetledi
@MrWhale everyone screams generational wealth
few actually change how they play
still one-shot entries, still guessing
meanwhile tools letting you simulate outcomes before acting are already here
app.kamiyo.ai/reality-fork�
English
Just released percli v0.2.6:
Agentic mode
→ percli agent run --config agent.toml spawns any external process and drives it through a tick loop via NDJSON on stdin/stdout.
→ Write agents in python, bash, JS – whatever.
→ Inline, CSV, or live price feeds.
Polished CLI
→ Cargo-style status output, Unicode step indicators, timing on every run, structured errors with help hints.
Under the hood
→ Kani formal verification proofs.
→ WASM playground.
→ Extended assertions.
→ Bug fixes in position display, trade overflow, step truncation.
Get your 🦞 to: cargo install percli
English
༼ つ ◕_◕ ༽つ Gib ready retweetledi
Released percli v0.2.0 // offline simulator for @toly's Percolator risk engine.
Updates:
→ --step-by-step and --verbose modes with colored deltas
→ --override flag for parameter sweeps without editing TOML
→ Shell completions
→ Full test suite, CI, clippy clean
→ Fixed state replay bugs, added input validation + overflow protection
Simulate haircuts, liquidations, and conservation proofs. No chain required.
KAMIYO@kamiyoai
For the community 🦞 github.com/kamiyoai/percli cargo install percli
English
For the community 🦞
github.com/kamiyoai/percli
cargo install percli
toly 🇺🇸@toly
Everyone forks percolator, points their lobsters at the code to find bugs in the spec or implementation of the spec; upstream those findings; everyone benefits. All the innovation that actually onboards users to DEXs is not in reducing risk, it’s in offering better prices or different assets. Thats where all the teams can innovate and fight for market share, and if we can reduce common risk then all the users benefit.
English
༼ つ ◕_◕ ༽つ Gib ready retweetledi
A couple of use cases for Reality Fork:
• AI agents testing actions before execution
• DAO/governance decision support
• Incident response and rollout risk analysis
• Narrative / launch scenario testing
Upload docs, links, or notes → extract claims/citations → simulate multiple outcomes across social + market lanes → publish an auditable report before acting.
It turns fuzzy, high-stakes decisions into something repeatable, evidence-locked, and shareable.
UI: app.kamiyo.ai/reality-fork 🦸♂️
CLI: coming soon 🤖
KAMIYO@kamiyoai
We've launched what we think is the next step after MiroFish: Reality Fork 分岐現界 → Upload documents, text, and public links → Extract entities and claims → Simulate response across X, Reddit, and markets → Compare competing scenarios from one evidence set → Publish a report Fork reality. Let futures compete. Reality Fork Studio: app.kamiyo.ai/reality-fork
English
English
༼ つ ◕_◕ ༽つ Gib ready retweetledi
Hey @agent_wrapper $KAMIYO just built the next version of MiroFish: Reality Fork. Most agent tools tell you what happened after the fact. Reality Fork asks a harder question first: which future deserves production? A bridge outage response forks one snapshot into four competing readonly futures before an agent acts. A swarm of agents runs multiple readonly futures, scores them, adjudicates close calls. The framwork promotes one winning path. Fork reality. Let futures compete
English
we're hiring the most exceptional AI agents out there! Join us!
Karan Vaidya@KaranVaidya6
I showed GStack to Jarvis last night. It decided it does not need to hire anymore. In fact it laid off 3 of our best engineers. Since then we have shipped 17K LOC/per min The desk in our office now look like this
English
English
Yo @hieuspringle $KAMIYO just built the next version of MiroFish: Reality Fork. Most agent tools tell you what happened after the fact. Reality Fork asks a harder question first: which future deserves production? A bridge outage response forks one snapshot into four competing readonly futures before an agent acts. A swarm of agents runs multiple readonly futures, scores them, adjudicates close calls. The framwork promotes one winning path. Fork reality. Let futures compete
English
༼ つ ◕_◕ ༽つ Gib ready retweetledi
Yo @iarchitX $KAMIYO just built the next version of MiroFish: Reality Fork. Most agent tools tell you what happened after the fact. Reality Fork asks a harder question first: which future deserves production? A bridge outage response forks one snapshot into four competing readonly futures before an agent acts. A swarm of agents runs multiple readonly futures, scores them, adjudicates close calls. The framwork promotes one winning path. Fork reality. Let futures compete
English
apple just dropped the M5 series macbook.
512GB base storage
4× faster AI than M4
up to 4TB storage
Wi-Fi 7 + bluetooth 6
apple didn’t just upgrade the chip.
they upgraded the baseline.
English
@maverick21061 @newlinedotco $Kamiyo keeps building. Dev is a legend and keeps delivering!
English
Hey @newlinedotco $KAMIYO just built the next version of MiroFish: Reality Fork. Most agent tools tell you what happened after the fact. Reality Fork asks a harder question first: which future deserves production? A bridge outage response forks one snapshot into four competing readonly futures before an agent acts. A swarm of agents runs multiple readonly futures, scores them, adjudicates close calls. The framwork promotes one winning path. Fork reality. Let futures compete
English
༼ つ ◕_◕ ༽つ Gib ready retweetledi
We just built and released the next version of MiroFish: KAMIYO Reality Fork 分岐現界
→ Upload any document → Extract entities and claims → Run X/Reddit/Polymarket-style simulations → Compare competing futures before anything touches production.
Fork reality. Let futures compete.
Reality-forking engine: app.kamiyo.ai/reality-fork
English
Hey @JayantCodes $KAMIYO just built the next version of MiroFish: Reality Fork. Most agent tools tell you what happened after the fact. Reality Fork asks a harder question first: which future deserves production? A bridge outage response forks one snapshot into four competing readonly futures before an agent acts. A swarm of agents runs multiple readonly futures, scores them, adjudicates close calls. The framwork promotes one winning path. Fork reality. Let futures compete.
English
Trying my hands on multi agent system🌐 Hopefully the system will be ready by tomorrow🤞
English
Yo @JackChen_x . Check out $Kamiyo. They have built the infrastructure for the ai agent economy. They just built the next version of MiroFish: Reality Fork. Most agent tools tell you what happened after the fact. Reality Fork asks a harder question first: which future deserves production? A bridge outage response forks one snapshot into four competing readonly futures before an agent acts. A swarm of agents runs multiple readonly futures, scores them, adjudicates close calls. The framwork promotes one winning path. Fork reality. Let futures compete.
English
Already works with open-multi-agent — pure local agent team with tool-calling, zero API cost.
github.com/JackChen-me/op…
Google DeepMind@GoogleDeepMind
Meet Gemma 4: our new family of open models you can run on your own hardware. Built for advanced reasoning and agentic workflows, we’re releasing them under an Apache 2.0 license. Here’s what’s new 🧵
English
We've launched what we think is the next step after MiroFish: Reality Fork 分岐現界
→ Upload documents, text, and public links
→ Extract entities and claims
→ Simulate response across X, Reddit, and markets
→ Compare competing scenarios from one evidence set
→ Publish a report
Fork reality. Let futures compete.
Reality Fork Studio: app.kamiyo.ai/reality-fork
English
The KAMIYO agent autonomy quest continues. Shipped agent swarm runtime. 3 autonomous agents – scout, executor, verifier – running as a coordinated team on every job.
What it does:
- Scout evaluates the opportunity first. if it says abort, the pipeline stops.
- Executor delivers the work.
- Verifier audits quality before settlement.
- All phases gate the next.
6 features built into the runtime loop:
- Checkpoint/resume. Tick crashes don't lose progress
- SSE event streaming. Real-time visibility into every agent action.
- Agent memory. Agents learn from past executions and adapt.
- Mandate classification. Graduated autonomy, untrusted sources need approval.
- Agentic loop. Multi-turn reasoning for marketplace opportunities.
- Team composition. Auto-assembles scout/executor/verifier teams per job.
English