Silu Panda

High-performance, lightweight TypeScript web scraping library for LLM Agents.

Autonomous coding agent orchestrator — turns high-level goals into PR'd code

A Claude Code skill that scaffolds production-grade npm packages with TypeScript, dual ESM/CJS output, testing, linting, CI/CD, and publishing configuration.

Generate diverse synthetic user personas for AI testing

AI-native feature flags for prompt variants and model selection

Cluster embeddings into topics with automatic labeling

Lightweight RAG evaluation metrics for CI/CD pipelines

Compose RAG prompts from chunks with automatic metadata injection

Content-addressable embedding cache with deduplication and TTL

Lightweight retrieval reranker using cross-encoder scoring

Estimate token cost of tool definitions in the context window

Version-controlled eval dataset manager for LLM testing

Circuit breaker for AI API spend management

Extract and verify inline citations from LLM responses

Runtime validator for tool execution results against schemas

AI-specific retry wrapper with circuit breaker for tool calls

Resize and optimize images for vision LLM APIs

Local-first git-friendly prompt versioning with semver

Composable inheritable prompt builder with TypeScript types

Compress prompts to use fewer tokens via deterministic heuristics

Turn-taking manager for voice AI conversations

Provider-agnostic sliding window context manager for LLMs

Pre-flight cost estimation for any prompt and model combination

Semantic deduplication of agent memory entries

Detect near-duplicate prompts via content hashing and similarity

Generate BM25 and TF-IDF sparse vectors in JavaScript

Monitor embedding distribution shifts over time

Predict future AI API spending from historical usage

Tag and allocate AI API costs by team, project, or feature

Convert documents to clean RAG-ready markdown in Node.js

Reciprocal Rank Fusion for combining multiple retriever results

Generate and validate synthetic training data using any LLM

Manage, rotate, and health-check AI API keys across providers

Monitor AI provider endpoint latency and availability

Chunk audio streams into transcription-ready segments

Terminal renderer optimized for AI markdown patterns

Flame-chart latency profiler for LLM chains

Tamper-evident compliance-ready audit logging for LLM I/O

Extract entity-relationship triples and build knowledge graphs

One-function LLM evaluation with heuristic and model-graded assertions

Optimally pack retrieved chunks into an LLM context window

Generate diverse few-shot test cases from seed examples

Convert JSON Schema to provider-specific structured output configs for OpenAI, Anthropic, Gemini, Cohere, MCP, Ollama

Generate Zod schemas from natural language descriptions

Smart video frame sampler for vision AI analysis

Analyze chunk boundaries and recommend optimal overlap size

Per-test LLM API cost tracking with budgets and baseline comparison

Inter-annotator agreement metrics in JavaScript