Skip to content

Instantly share code, notes, and snippets.

View roman-rr's full-sized avatar
🦜
LangChain / LangGraph / DeepAgents

Roman roman-rr

🦜
LangChain / LangGraph / DeepAgents
81 followers · 6 following
View GitHub Profile
@roman-rr
roman-rr / experttrialproject.md
Last active November 28, 2019 23:51
JavaScript Expert Code Review

Review Mastered: Roman Antonov

List sorted: Best -> Worst

Sohel-rana (gist)

Good readable code, but can write better if know new standarts.

> Good practice (10)

@roman-rr
roman-rr / circle-progress.md
Last active October 1, 2020 17:26
Circle progress component with scss and html

Circle progress component with scss and html. Supported by all browsers and platforms.

<div class="app-progress">
  <svg>
    <circle  cx="25" r="21" cy="25" class="progress-ring done-{{VALUE}}"/>
  </svg>
</div>
@roman-rr
roman-rr / ruflo-audit-gist.md
Last active July 1, 2026 08:01
Ruflo / Claude-Flow Audit: 300+ MCP Tools — 99% Theater, 1% Real, 5x Token Waste

Ruflo / Claude-Flow: 300+ MCP Tools Exposed — 99% Theater, 1% Real, 5x Token Waste

A deep technical audit of the ruflo (@claude-flow/cli) ecosystem — what it claims vs what it actually does.

Investigated 2026-04-04 by spawning 8 research agents across two analysis phases, reading source code from the ruflo GitHub repo, tracing local process execution, and testing every major tool category hands-on.

TL;DR

@roman-rr
roman-rr / mempalace-audit.md
Last active July 2, 2026 15:03
MemPalace Exposed: 42,000 Purchased Stars, Zero Innovation — It's ChromaDB With a Celebrity Name

MemPalace Exposed: 42,000 Purchased Stars, Zero Innovation — It's ChromaDB With a Celebrity Name

A deep technical audit of MemPalace (github.com/MemPalace/mempalace) — what it claims vs what it actually is.

Every source file read. Every benchmark traced. Every MCP tool verified. Compared against Mem0, Zep/Graphiti, and Letta/MemGPT. GitHub stargazer timestamps analyzed for bot patterns.

TL;DR

@roman-rr
roman-rr / stop-using-vision-llms-trading-charts.md
Last active April 22, 2026 15:53
Stop Using Vision LLMs to Read Trading Charts — 4 Models, 2 Vendors, 40 Real Signals: 51% Direction, 1/215 Patterns, 100% Long-Bias on Gemini

Stop Using Vision LLMs to Read Trading Charts — 4 Models, 2 Vendors, 40 Real Signals: 51% Direction, 1/215 Patterns, 100% Long-Bias on Gemini

A deep empirical audit of using frontier vision LLMs as chart-pattern experts in a production crypto signal pipeline. Tested against QuantAgent (arXiv:2509.09995), the paper repeatedly cited as proof that LLM agents can trade from candlesticks.

Conducted April 2026. Every model called, every call logged, every fixture tied to a verified trade outcome. Four frontier models, two vendors (Anthropic, Google), 215 LLM calls, $1.16 total spent.

From the team building signals.x70.ai — a production AI-powered crypto signals service. Open-source companion: roman-rr/trading-skills.

Formal preprint. A peer-review-format PDF version of this audit is published as a preprint: