Skip to content
View morgan-coded's full-sized avatar

Highlights

  • Pro

Block or report morgan-coded

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
morgan-coded/README.md

Morgan Carr

AI/LLM & TypeScript engineer — I build LLM integrations, MCP tooling, and static-analysis / developer tools, and I ship focused, tested fixes upstream in the libraries I use.

Night-time coder. My work lands as small, well-tested pull requests in real projects — 40+ merged into external open-source repos, including the official OpenAI Node SDK, promptfoo, txtai, kysely, react-jsonschema-form, ast-grep, protobuf.js, and meriyah.

Focus areas

  • AI / LLM — OpenAI/Anthropic SDKs, structured outputs, streaming, tool/function calling, RAG, and MCP (Model Context Protocol) servers
  • TypeScript / Node — SDK & API work, type-level correctness, parsers/ASTs
  • Static analysis & tooling — ESLint plugins, linting, package analysis, and the edge-case fixes that get merged into major libraries
  • Backend reliability — REST APIs, Postgres/Prisma, webhooks, idempotency

Projects

  • ast-lens-mcp — MCP server for AST-based code intelligence on TypeScript/JavaScript codebases (npm).
  • mcp-source-lint — static checks for MCP server source that catch stdio corruption and missing handler input schemas.
  • agent-bisect — git-bisect for agent runs: deterministically localizes the first breaking step in an agent transcript, and abstains when it can't.
  • chrome-review — TypeScript static analyzer for Chrome MV3 extensions.

Recent upstream contributions

…plus merged fixes across the ESLint plugin ecosystem (eslint-plugin-n, -regexp, -yml, -playwright, -import-x, eslint-stylistic, and more).

Tech

TypeScript · Node.js · Python · React/Remix · Postgres/Prisma · Vitest · Playwright · MCP · OpenAI/Anthropic SDKs

Pinned Loading

  1. ast-lens-mcp ast-lens-mcp Public

    MCP server for AST-based code intelligence on TypeScript/JavaScript codebases

    TypeScript 1

  2. agent-bisect agent-bisect Public

    git-bisect for agent runs: deterministically localizes the first breaking step in an agent transcript, and abstains when it can't see one. Precision-first, honest measurement.

    Python