Reading

Keel: A Multi-Agent Workflow Framework That Turns Coding Agents into Work Owners

Keel is a project-agnostic multi-agent workflow skeleton that drives GitHub Issues from to-do items to merged completion, enabling end-to-end automation covering requirement understanding, branch creation, code implementation, CI waiting, code review, test validation, and secure merging.

keel多代理工作流GitHub自动化代码审查合并策略CI/CDLangGraphRAG

Published 2026-06-09 22:45Recent activity 2026-06-09 22:53Estimated read 4 min

Keel: A Multi-Agent Workflow Framework That Turns Coding Agents into Work Owners

Section 01

Keel: A Multi-Agent Workflow Framework for End-to-End Issue-to-Merge Automation

Keel is a project-agnostic multi-agent workflow backbone that automates the full lifecycle from GitHub Issue creation to PR merge—covering requirement understanding, branch creation, code implementation, CI waiting, code review, test validation, and secure merging. It fills gaps in existing tools by integrating coding agents, review tools, quality gates, and merge strategies into a unified work ownership framework.

Section 02

Background & Motivation for Keel

Existing coding agents (OpenHands, SWE-agent, Devin) stop at PR creation; PR review tools (CodeRabbit, PR-Agent) focus on existing PRs; merge queue tools (GitHub Merge Queue, Mergify) handle post-test merging. None cover the complete issue-to-merge lifecycle. Keel was built to bridge this gap as a unified workflow backbone.

Section 03

Core Architecture & 13-Step Workflow

Keel uses a 3-layer design: Layer1 (fixed keel-core backbone), Layer2 (project-specific config via project.yaml), Layer3 (add-only extensions). Its 13-step workflow includes key stages: implement (coding agent works), classify (risk grading), review (multi-agent via ai-jury), merge (security strategies like time zone locks), and capture (learning accumulation).

Section 04

Keel vs. Existing Tools

Keel complements rather than replaces existing tools:

Adds lifecycle management beyond PR creation for coding agents.
Extends PR review tools with implementation loops and merge strategies.
Provides issue ownership (pre-PR) for merge queue tools. It acts as a work ownership skeleton integrating these tools into a unified lifecycle.

Section 05

Installation, Commands & Quality Assurance

Installation: pip install keel-workflow (Python ≥3.11). Key commands: setup (project config), validate (config check), plan (workflow preview), ship (flagship command). Keel uses dogfooding (self-development via its own workflow) and maintains 100% test coverage for core packages.

Section 06

Real-World Application Scenarios

Automated Issue Handling: End-to-end processing from issue creation to merge without manual intervention.
Nightly Regression Testing: overnight command runs tests, checks dependencies, and creates issues for problems.
Multi-agent Code Review: Integrates ai-jury to run multiple agents for comprehensive code reviews.

Section 07

Summary & Future Vision

Keel v1 focuses on single-agent work ownership. Long-term vision: Autonomous software teams (multiple agents under human supervision). It frees developers from routine tasks to focus on human-judgment work (product tradeoffs, approvals). Keel is key infrastructure connecting human intent and machine execution in software engineering automation.

Continue Reading

Keep going with more reads from the same topic.

Nornir MCP Server: An Enterprise-Grade Bridge for Integrating Large Language Models into Network Automation

Nornir MCP Server is an enterprise-level server based on the Model Context Protocol (MCP). It seamlessly integrates large language models (such as Claude) with the Nornir network automation framework, supporting natural language orchestration for multi-vendor network devices (Cisco, Arista, Juniper, etc.), and providing production-grade features like a dual-engine architecture (NAPALM + Netmiko), intelligent filtering, and a secure sandbox.

Recent activity 2026-05-06 20:51

Bibliothèque Française LLM: A French Public Domain Literature Index System Optimized for Large Language Models

Bibliothèque Française LLM is a structured indexing and annotation project for French public domain literature designed specifically for large language models (LLMs). It integrates multiple authoritative sources such as DraCor, Common Corpus, and Wikisource, providing metadata indexing categorized by genre, author, and era, as well as in-depth annotations for dramatic texts (including characters, lines, stage directions, etc.). Its aim is to enable LLMs to efficiently read and understand classic French literary works.

Recent activity 2026-05-06 20:50

Splinter: A Lock-Free Zero-Copy Shared Memory KV and Vector Storage Library That Eliminates Socket and Memcpy Overhead for LLM Inference

Splinter is a minimalist, high-performance key-value (KV) and vector storage system enabling zero-latency inter-process communication via shared memory and atomic operations. With only 766 lines of core code, it supports millions of operations per second and 768-dimensional vector storage, offering a new architectural approach for local LLM inference and data-intensive applications.

Recent activity 2026-04-03 08:49

libmlxforge: An Embedded MLX LLM Inference Engine for Apple Silicon

libmlxforge is an embeddable MLX large language model (LLM) inference engine designed specifically for Apple Silicon. It provides a unified C ABI interface, supports calls from Node.js, Swift, and Rust, and features continuous batching, streaming output, JSON-constrained structured output, and embedding vector generation.

Recent activity 2026-06-09 17:23