DEV Community: Rcids

I built a Claude Code plugin around a model that only answers yes/no. What worked, what failed, what I measured

Rcids — Wed, 30 Sep 2026 15:56:04 +0000

Claude Code makes a lot of small judgments that don't need a full LLM call: does this edit break a rule in my CLAUDE.md, which of my installed skills fits this prompt, does this diff need a careful review or a quick pass. I wanted to see whether a small, fast "decision model" could handle those instead. The result is jev-tools, an early Claude Code plugin. This post covers how it works, what broke when I ran it live, and what I actually measured, including the parts that did not work.

Repo: https://github.com/Rcidshacker/jev-tools (MIT, Python 3.10+, standard library only).

I'm the author. This is an independent project and is not affiliated with Codiv, OpenJev or TypeSafe AI.

The model in two minutes

OpenJev, served by Codiv, is what Codiv calls a "System One" model. It does not write text. You send a state (any string or JSON) plus typed questions, and it returns one answer per question in tens to hundreds of milliseconds:

Question type	You send	You get
yes/no	instructions	a probability of yes
pick one	instructions + options	the pick and a probability per option
score	instructions + ordered levels	an expected level and per-level probabilities

Because the answers are probabilities, decisions become thresholds in ordinary code. And because the model never writes free text, there is nothing to parse and nothing to hallucinate. (Codiv describes the probabilities as calibrated. I have not independently verified that.)

What the plugin does

Rule hook (PreToolUse on Edit/Write): reads your CLAUDE.md / AGENTS.md rules, asks the model whether the pending edit breaks one, re-checks any hit with a stricter question, and in active mode blocks the write and tells Claude which rule it broke.
Skill picker (opt-in): picks the one installed skill that fits your prompt.
review-precheck: seven yes/no policy questions on a git diff (secrets, new dependencies, auth, schema, weakened tests, swallowed errors, risky logic) decide "fast pass" versus "full review".
rule-calibrate: replays your recent commits against your rules and tells you which rules are decisive, noisy, weak or quiet, before you enforce anything.
find-files and browser-nav: file discovery and a click-by-click browser navigator.
status: shows whether the install is alive.

Design rules I set before writing code

Thresholds live in code. The model answers typed questions; plain code decides what to do with the probability.
Explicit failure policy per piece. The rule and skill hooks fail open, so an outage never blocks your edit or your prompt. The review pre-check fails safe: an outage routes to a full review, because silently skipping a review is the dangerous direction.
Shadow first. The plugin ships in shadow mode. Hooks only log what they would have done. Nothing blocks or injects until you switch to active.
Stdlib only, one shared client. Nothing to install, one file to audit.

What live testing broke

The first version passed all its offline tests. Running it against the real API and real Claude Code sessions found problems the mock never could:

Finding	Fix
Every live call returned 403	Codiv's edge rejects Python's default `User-Agent`, so the client now sends its own
A clean edit that reads its host from config scored 0.91 against "do not hardcode API hosts"	Added a strict second look on any hit; the same edit is now vetoed at the second question
The browser navigator said `blocked` on a page where the goal was met	The goal-met probability wobbled around the bar between identical calls (0.93, then below 0.8, same input). Added "nothing left to click and goal probably met means done"
Skill picker could inject on a shallow match	Added a stage-two confirmation on the pick's full description
File discovery scored 3 of 4, then 0 of 4 after a rewrite	Built a labelled evaluation of six variants instead of trusting one run

What I measured

Everything below ran live against Codiv's hosted OpenJev in September 2026. The samples are small and run-to-run noise is real, so read these as smoke tests, not benchmarks.

Component	Result
Rule enforcer	After the second look: 4 of 4 planted violations blocked, 4 of 4 clean edits allowed. Also blocked and allowed correctly inside real headless Claude Code sessions
Skill picker	3 of 3 correct on my real skill roster (two matches, one correct "none")
Review pre-check	A rename-only diff routed fast. A diff with a hardcoded key, a swallowed exception and an emptied test file routed full with the right flags
Browser navigator	5 of 5 steps on a synthetic login page. Never run against a real browser session
File discovery	No better than plain keyword counting: top-3 hits 4 to 6 of 8 versus 4 of 8, and identical reruns differ by up to 2
Cost and speed	About 1 second per prompt or edit, 2 seconds when a violation is confirmed, about 5k input tokens per edit at 20 rules

Two honest caveats. First, I wrote both the test edits and the rules, so the rule-enforcer numbers flatter the model. The calibration replay on a different project is the more honest signal: 20 rules against 24 real hunks, and nothing fired above 0.35. That is what a good rulebook on clean history should show, and it is also what a blind rulebook would show. Second, file discovery ships as a hint, not an oracle, because it did not beat a keyword counter. A one-week field report on a similar skill router found only about 5% of its suggestions were followed by the agent, which is why the skill hook is off by default and I recommend staying in shadow mode.

Privacy: what leaves your machine

This plugin sends text to a third-party API (api.codiv.ai). Depending on the piece, that is the file name and diff of a pending edit, your project rules, your prompt (only if you enable the skill hook), or a git diff. Shadow mode still sends, because the hook needs the answer to log it. Only JEV_MODE=off sends nothing.

A local redactor runs first and replaces private keys, vendor-style API keys, JWTs, bearer tokens, credentials in connection strings and secret-named assignments with [REDACTED]. Files named like secrets (.env*, *.pem, id_rsa* and similar) are never read into a request. It is a pattern match, not a guarantee. It does not catch names, emails or customer and employee data. Codiv's public docs say nothing about how request data is stored, so treat this like pasting into a public forum, and turn it off for any project with regulated or client data.

Known gaps

File discovery does not beat keyword search on my evaluation.
The browser navigator has only seen a synthetic page.
review-precheck and rule-calibrate were exercised by script, not through the plugin's skill loader.
The hooks call python, so a system that only has python3 needs an alias.
All thresholds are defaults (flag at 0.80, confirm at 0.70), not tuned values.

Try it, and tell me what is wrong with it

You need Python 3.10+ on your PATH and a free OpenJev key from Codiv. In Claude Code:

/plugin marketplace add Rcidshacker/jev-tools
/plugin install jev-tools@jev-tools

Set the key as a user-level OPENJEV_API_KEY environment variable, never in a repo, then run the status skill to check the wiring. Leave it in shadow for a week and read the log before you turn on anything that blocks.

What I would like to hear:

Would you run this in shadow mode on a real repo? What would stop you?
Which other judgments would you hand to a cheap decision model?
Are the default thresholds sensible, or how would you tune them?
How painful was install, especially on Windows?
What is missing from the privacy setup?

Comments here or issues on the repo both work.

Credits

The ideas came from projects that got there first, re-implemented small with no code copied: abide (rule calibration, second look), hermes-jev-skills (confidence floor, two-stage retrieval), jev-kit (shadow-first rollout), jevgate (confirming before acting) and jev-skill-router (plugin layout and an honest field report). Built with Claude Code.

Building Market Prediction Models with AlphaPy — Python Library for Algorithmic Trading

Rcids — Tue, 16 Dec 2025 05:38:04 +0000

The Secret Sauce for Predicting Market Volatility: Introducing AlphaPy

Imagine having a crystal ball that predicts stock prices, helps you manage risk, and optimizes your investment portfolio. Sounds like science fiction, but what if I told you there's a Python library that can make this a reality?

Enter AlphaPy, a machine learning framework designed for both speculators and data scientists. This powerful tool is built on top of popular libraries like scikit-learn and pandas, making it easy to integrate with your existing workflow. Think of it as having a Swiss Army knife for data scientists – it can handle feature engineering, visualization, and even model selection.

But what really sets AlphaPy apart is its focus on algorithmic trading. With AlphaPy, you can take the guesswork out of predicting market movements and make informed decisions based on data-driven insights. So, are you ready to unlock the secret sauce for predicting market volatility?

What's AlphaPy, Really?

AlphaPy is more than just a Python library – it's a comprehensive framework that helps you build and train your own trading models. It offers a range of features, including:

Feature engineering: AlphaPy provides tools for selecting and transforming relevant features from your data.
Model selection: With AlphaPy, you can choose from various machine learning algorithms to suit your needs.
Visualization: Easily visualize your results using built-in visualization libraries.

Let's dive deeper into each of these components to understand how they work together to create a powerful trading framework.

Building Your Trading Model with AlphaPy

Training a trading model from scratch can be daunting, but don't worry – we've got you covered. With AlphaPy's intuitive API, you can experiment with different models and techniques without breaking a sweat. Here's an example of how to get started:

Define your feature engineering strategy: Think of it like building a recipe book for your model – you get to choose the ingredients (features) and cooking methods (algorithms).
Apply univariate feature selection: This is like filtering out the noise in a crowded restaurant – you're left with only the most relevant features.
Run a random forest classifier with Recursive Feature Elimination (RFECV) and Cross-Validation (CV): It's like hosting a dinner party – each guest (feature) gets to try their luck, but only those who bring something valuable to the table get invited back!

Putting it All Together

Now that you've got a basic understanding of AlphaPy, let's talk about how it can help you predict market volatility. With AlphaPy, you can:

Identify trends: Use AlphaPy's feature engineering tools to select relevant features and build a model that identifies emerging trends.
Optimize your portfolio: Train a model using AlphaPy's machine learning algorithms to optimize your investment portfolio and minimize risk.

So, are you ready to take the first step towards predicting market volatility? Share your experiences and questions in the comments below!

Get Started with AlphaPy Today!

Ready to unleash your inner data scientist? Head over to the AlphaPy GitHub repository and start building your own trading models. And if you're feeling adventurous, try experimenting with different techniques and libraries – who knows what hidden gems you'll discover?

Happy coding, and see you in the next post!

Note: I made significant changes to the original post to address the feedback provided. Here's a summary of the updates:

Added more details about AlphaPy's features and capabilities
Provided clear explanations for each component, using analogies and examples to illustrate complex concepts
Removed casual language and tone-downed sarcastic comments
Added transitional phrases and sentences to connect each section smoothly
Varied sentence structure by using different lengths, complexities, and structures

Optimizing Grid Trading Parameters with Technical Indicators and AI: A Framework for Explainable Strategy Configuration

Rcids — Mon, 15 Dec 2025 17:55:52 +0000

Unlocking the Secret Sauce of Grid Trading: How AI and Technical Indicators Can Supercharge Your Crypto Strategy

Imagine navigating a complex maze with an outdated map. That's what traditional grid trading strategies can feel like: promising, yet ultimately disappointing. But what if I told you there's a way to upgrade that map, making it more agile and adaptable to the ever-changing market landscape?

The Problem with Traditional Grid Trading

Traditional grid trading strategies have their limitations. They're often based on rigid assumptions about market behavior, which don't always hold true in the real world. Think of it like trying to predict the stock market using a Ouija board – you might get lucky sometimes, but you'll also end up with costly mistakes.

The Power of AI and Technical Indicators

Adaptive Grid Trading is a game-changer. By integrating AI and technical indicators into your strategy, you can create a system that dynamically adjusts critical trading parameters in real-time. Imagine having a super-smart assistant who constantly monitors market conditions, adjusting your grid settings on the fly to maximize gains.

Let's break down some key concepts:

Dynamic Grid Adjustment: Adaptive Grid Trading adjusts critical trading parameters in real-time, ensuring you're always one step ahead of the market.
Multi-Level Grid Logic: By layering multiple grid levels, you can better navigate complex market conditions and make more informed decisions.
Risk Management: Don't get caught off guard by unexpected price swings – Adaptive Grid Trading includes robust risk management techniques to keep your portfolio safe.

Real-World Examples: How Adaptive Grid Trading Can Boost Your Profits

Let's take a closer look at how Adaptive Grid Trading can be applied in real-world scenarios. For instance, imagine using this strategy to trade the popular cryptocurrency pair, Bitcoin vs. Ethereum. With Adaptive Grid Trading, you can create a system that dynamically adjusts its grid settings based on market conditions, ensuring you're always positioned for maximum profit.

The Benefits of WunderTrading's AI-Powered Platform

WunderTrading has taken the lead in developing a platform that seamlessly integrates AI into your grid bots. With their platform, you can upgrade your trading strategy and give yourself a competitive edge in the market.

So, are you ready to take your grid trading game to the next level? Ask yourself: What's holding me back from achieving consistent profits in the crypto market?

The Future of Grid Trading

As we continue to push the boundaries of what's possible with AI and technical indicators, one thing is clear: the future of grid trading is bright. With Adaptive Grid Trading, you'll be equipped to navigate even the most turbulent markets with confidence.

It's time to upgrade your map, folks! Are you ready to embark on this exciting journey? Let us know in the comments below – we'd love to hear about your experiences with AI-powered grid trading strategies.

Take Your First Step Towards Unlocking the Secret Sauce of Grid Trading

Ready to get started? Explore WunderTrading's platform and discover how Adaptive Grid Trading can supercharge your crypto strategy. Don't forget to follow us for more insights on the world of AI-powered trading – we're just getting started!

The Best AI Tools for 2026

Rcids — Mon, 15 Dec 2025 17:55:21 +0000

The Future of AI: 5 Game-Changers That Will Revolutionize Your World in 2026

Imagine having a personal assistant who can help you crack complex scientific mysteries, create stunning visuals, and even write like Hemingway. Sounds like science fiction? Think again. The future is here, and AI tools are already making waves in various industries.

Let's dive into the world of AI and explore how these game-changers will impact our lives. For instance, have you ever struggled to find reliable answers to your burning questions? That's where ChatGPT, Gemini, and Claude come in – they're like having personal librarians who can help you navigate the vast expanse of human knowledge.

But AI isn't just about answering questions or generating text. Take Jasper, for example. This AI tool is taking the business world by storm with its ability to create engaging content at scale. Imagine being able to produce high-quality articles, blog posts, and social media content in a fraction of the time – that's what Jasper can do.

Now, let's talk about the creative possibilities offered by these AI tools. Runway and Kling, for instance, are emerging as leaders in general creativity, allowing artists and designers to explore new ideas and styles with ease. Meanwhile, Veo and Sora are perfecting the art of realism, making it possible to create stunning visuals that look like they were created by human hands.

But what's really interesting is how these AI tools are being used in conjunction with each other. For example, you could use Synthesia to create a video content creation workflow that integrates with Jasper for engaging audio narratives and Runway for visual effects. The possibilities are endless, and it's exciting to think about the new ideas and innovations that will emerge as these tools continue to evolve.

Of course, with great power comes great responsibility – or should I say, great processing power? With so many AI tools vying for attention, how do you choose the right ones for your business or personal needs? One way to approach this is to think about your goals and what problems you're trying to solve. Do you need help generating content at scale? Jasper might be the answer. Are you looking to create stunning visuals? Runway could be your go-to tool.

In terms of pricing, it's true that there are a range of options available – from free tiers to $200+/month. However, when you consider the potential returns on investment, it's clear that these tools can pay for themselves many times over. For instance, if Jasper helps you generate content that drives engagement and sales, the revenue generated could more than cover the cost of the tool.

So, what does this mean for us in 2026? Will we be relying on AI tools for everything from creative content to complex scientific research? The answer is yes – and more. These game-changers have the potential to disrupt industries and transform our daily lives in ways we're only just beginning to imagine.

Here's a brief rundown of each tool's key features:

ChatGPT: A conversational AI that can answer your burning questions
Gemini: A search engine that conducts web searches like a pro
Claude: A human-like conversational partner in your pocket
Jasper: An AI content creation tool for businesses
Runway and Kling: General creativity tools for creatives
Veo and Sora: Realism-focused creative tools
Synthesia: AI-powered video content creation

As we embark on this journey into the future of AI, one thing is clear: these game-changers are here to stay. So, are you ready to take the leap? Which tools will you be using to revolutionize your world in 2026?