MCP App StoreMCP App StoreThinkwell is a TypeScript framework that blends deterministic code with LLM reasoning. It gives developers a fluent Plan API (agent.think → plan → run) and a @JSONSchema pattern to declare typed outputs. The skill documents lifecycle, prompt composition helpers, tool registration, streaming thought events, and CLI/script usage so agents can produce structured JSON results and integrate tools safely.
Use Thinkwell when you need typed, reliable LLM outputs inside TypeScript projects: generating structured summaries, extracting data via schemas, wiring tools with typed inputs/outputs, and building repeatable agent scripts or CI tasks. It's suited for developer workflows that require programmatic control over prompts, streaming events, and tool callbacks.
Works with agents and runtimes referenced in the docs (Claude, Codex, Gemini, OpenCode) and with TypeScript-based agent runtimes that can import the Thinkwell library.
Thinkwell is a TypeScript framework for composing LLM prompts with a fluent Plan API, structured JSON output via @JSONSchema, tools, skills, and streaming. The SKILL.md is comprehensive documentation with clear code examples and progressive disclosure. No bundled scripts were present, so no runtime testing was performed.
Pure documentation skill, no executable scripts. Clean SKILL.md with no security concerns. Well-structured with references to external docs for API details, schema guide, and examples. Could benefit from adding metadata.requires section for dependencies (node, npm, thinkwell CLI).
Retriever (Membrane CLI)
Integrate Retriever search and discovery via the Membrane CLI — manage connections and run queries across Retriever data sources.
Yielding Bear — Unified LLM Routing
A unified LLM routing API that automatically routes requests to the cheapest capable model across providers to reduce inference cost by ~60-80%.
Markdrop
Convert PDFs to structured Markdown or interactive HTML and generate AI-powered descriptions for images and tables using multiple LLM providers.
Seedance Prompt — Short
Build, validate, and compress short-form Seedance 2.0 prompts (30–100 words) using a layered compression engine and @Tag delegation for quad-modal (T2V/I2V/V2V/
Node.js Best Practices
Guidelines and decision-making for Node.js architecture, runtime, async patterns, security, validation, and testing to inform framework and system choices.
Bash Pro
Defensive, production-grade Bash scripting patterns and CI/CD best practices: strict mode, safe argument parsing, testing with Bats, and tooling (ShellCheck/shf
SSE Stream Resilience
Provide robust server-sent-events (SSE) stream management with Redis-backed registry, heartbeat monitoring, completion persistence, and background guardian clea
Prompt Engineering Patterns
A practical collection of prompt engineering techniques and patterns for building reliable, controllable LLM prompts: CoT, few-shot, structured outputs, templat
