Saltar al contenido
Anuar Harb
ENES

Trabajo

Sistemas de agentes, herramientas open source y trabajo de producto de Anuar Harb, con arquitectura y resultados verificados.

2 proyectos

  • Quillami Code

    An open-source coding agent for the terminal, built from scratch on the raw Anthropic API with no orchestration framework.

    Rol
    Sole author. Design, implementation, tests, evals and release.
    Año
    2026
    Stack
    TypeScript, Node.js 22, Anthropic SDK, MiniMax (Anthropic-compatible), node:test, GitHub Actions
    • Agentes
    • Open source
    Quillami Code architecture The CLI sends each turn into an agent loop. The loop compacts history over 20k tokens, streams the model response, asks permission for risky tools, snapshots files for undo, runs one of eight tools and feeds the result back. Model routing picks Anthropic or MiniMax. Memory files, a usage ledger, and tests, evals and CI sit around the loop. Developer terminal prompt CLI src/index.ts Model routing Anthropic, MiniMax turn stream API Agent loop src/agent/loop.ts, up to 20 tool steps per turn compact over 20k tokens stream model response permission write, edit, bash checkpoint snapshot for /undo run tool 8 tools tool_result Memory QUILLAMI.md, AGENTS.md Usage ledger tokens and USD per turn Tests, evals, CI node:test, 4 evals, Actions

  • Production agents at Inmercity Labs

    Four LLM agent systems in production for clients over WhatsApp, connected to their systems with RAG, tool calling and MCP.

    Rol
    Co-founder, AI Engineer and Technical Lead
    Año
    2026
    Stack
    WhatsApp Cloud API, LLM tool calling, RAG, MCP, Programmatic evals
    • Agentes
    Inmercity Labs agent architecture A user writes on WhatsApp. The WhatsApp Cloud API forwards the message to the agent runtime. The agent loop uses retrieval over client documents, tools for live systems like inventory and calendars, and MCP servers for other client systems. A programmatic eval suite gates every release before deploy. User WhatsApp WhatsApp Cloud API Agent runtime connected to client systems Agent loop LLM with tool calling RAG client docs Tools live data MCP systems for example live inventory and advisors' calendars Programmatic evals every release is gated on the eval suite before it reaches users