🤖 AI & Autonomous Agents Tools
Interactive prompt engineering sandboxes, agent pattern simulators, RAG calculators, and large language model token budgeters. 100% private in-browser AI experimentation.
🧠 Advanced RAG Playground — BM25 + Embeddings + Re-ranking
Launcher for the standalone AI: hybrid retrieval over documents you paste in — BM25, hashed dense vectors, HyDE, reciprocal rank fusion and MMR — with per-pass scores and a trace.
🤖 AI Agent Orchestrator — Multi-Agent ReAct & Tool-Use
Launcher for the standalone AI: a planner, retriever, reasoner, actor and verifier pipeline that really executes local tools, then reviews the answer constitutionally. In-browser, no key.
🔄 AI Agent ReAct Loop Studio
Launcher for the standalone AI: the ReAct loop — thought, action, observation — with real tool calls and a stop condition you can watch. Also runs inside the agent pipeline. In-browser.
🧠 AI Chain-of-Thought Parser
Launcher for the standalone AI: chain-of-thought reasoning as numbered steps with a visible trace, plus chain-of-verification and self-consistency on the same question. In-browser, no key.
📊 AI Context Window Budget Lab
Launcher for the standalone AI: budget a context window across system prompt, retrieved chunks, history and reserved output, with warnings and a suggested chunk count. In-browser, instant.
⚡ AI Few-Shot Prompt Builder
Launcher for the standalone AI: build few-shot prompts in XML, JSON, markdown or chat-message framing, with a token estimate and guidance on how many exemplars you need. In-browser.
🤖 AI Function Calling Validator
Launcher for the standalone AI: validate function/tool declarations in OpenAI, Anthropic and Gemini shapes, get errors and warnings, and copy a normalised version for each provider. In-browser.
🕸️ Graph-of-Thought Lab — Aggregation, Transformation, GoT Reasoning (Besta 2023/2024)
Launcher for the standalone AI: thoughts as a graph — merge, refine and loop back over candidate ideas with per-node scoring, instead of one linear chain. Runs in your browser.
🔍 AI Hallucination & Fact Check
Launcher for the standalone AI: extract claims from an answer, check each against your indexed documents, and read coverage, verdict counts and token entropy as hallucination risk. In-browser.
🔌 AI MCP Tool Protocol Inspector
Launcher for the standalone AI: a mock MCP server in the page — initialize, ping, tools/list, tools/call, resources/list, prompts/list — with request checks and compliance report. In-browser.
🧠 Browser Vector Memory — IndexedDB + Semantic Search
Launcher for the standalone AI: vector memory in IndexedDB scored like Generative Agents — relevance, importance and recency — with tiers, reflections, consolidation and import/export.
💰 AI Model Cost & Speed Matrix
Launcher for the standalone AI: compare models on cost per request, per day and per period with prompt-caching discounts, tokens/sec and modelled latency. Your numbers, nothing fetched.
🛡️ AI Prompt Injection Defense
Launcher for the standalone AI: scan text against twelve injection patterns with severities and a risk score, get a sanitised copy, and read the defences that actually hold. In-browser.
🎭 AI Realistic Face Avatar — Procedural Canvas, FACS, Eye-Tracking, Lip-Sync
Byte 3.0's realistic face live — 100% procedural Canvas 2D, no images, no external models. Skin subsurface scattering, iris texture with flecks, micro-saccades, FACS action units (AU1, AU2, AU4, AU...
🧠 AI Reasoning Trace Visualizer — Chain-of-Thought & Self-Critique
Launcher for the standalone AI: every reasoning run emits a step-by-step trace plus a tree, graph, DAG, vote or iteration visualisation. Export any trace as JSON. In-browser, no account.
🌟 AI Super Realistic Face v4 — FACS Continuous Blending, Micro-Expressions 1/25s, 12 Visemes, GoT Eye-Tracking
Byte 4.0 super face v4 — upgrade from v3 discrete emotions to continuous FACS blending (happy 0.12 + thinking 0.34), micro-expressions 80ms flashes 1/25s (Ekman 2003), 12 viseme phoneme groups, eye...
🧠 AI Thinking Machine Lab — ToT, Reflexion, HyDE, RRF, Generative Memory (Research Stack 2022-2026)
Launcher for the standalone AI: 18 implemented reasoning methods with an automatic router that picks the method that fits your question, and an honest label on every answer. In-browser.
🔧 AI Tool-Use Simulator — Function Calling & Byte 2.0 Integration
Launcher for the standalone AI: 13 local tools with JSON schemas — calculator, units, dates, text stats, encoders, JSON, regex, hashing, retrieval and memory. Run or parse any call. In-browser.
📐 AI Vector Similarity Matrix
Launcher for the standalone AI: compare two texts by cosine, Euclidean and dot product over 64-dim hashed embeddings, with token overlap, Jaccard and per-dimension sparklines. In-browser.