The latest articles on DEV Community by Yashneil (@yashneil75). https://dev.to/yashneil75 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4042898%2F7c3ce289-44cf-49e9-b51e-724fea83167e.png https://dev.to/yashneil75 en Yashneil Thu, 23 Jul 2026 02:43:45 +0000 https://dev.to/yashneil75/we-turned-agent-conversations-into-git-commits-and-its-actually-useful-183e https://dev.to/yashneil75/we-turned-agent-conversations-into-git-commits-and-its-actually-useful-183e <p>Ever wished your agent conversations were as trackable as your code? Gitlord makes it real.</p> <p>What it does:</p> <ul> <li>Every agent turn becomes a git commit</li> <li>Branch out subagents without breaking your main flow</li> <li>Rewind to any point in your conversation history</li> <li>Connect tools via MCP (filesystem, search, browser, etc.)</li> <li>One interface, any AI provider (OpenAI, Anthropic, local models)</li> <li>Full CLI for managing sessions and branches</li> </ul> <p>Why it matters:</p> <ul> <li>Your agent history is navigable, forkable, and diffable, just like code</li> <li>Context management handles token budgets automatically</li> <li>Spawn child agents on isolated branches with their own history</li> <li>No lock-in: all components are modular and swappable</li> <li>Built-in integrations:</li> <li>MCP tools: Connect any MCP server (git, filesystem, browser, search). Tools flow to subagents automatically</li> <li>RAG: Vector search across your full agent history. ChromaDB-backed semantic queries built in</li> <li>Provider abstraction: Switch between any of the 170 providers and nearly 3,000 models, or local models with one line. Mix providers per agent</li> </ul> <p>Build an agent in 4 lines:</p> <p><code>from gitlord import Session, SessionConfig<br> config = SessionConfig(model="claude-opus")<br> session = Session.create("my-agent", config)<br> session.add("user", "Refactor our OAuth to the new framework")<br> Done. Gitlord handles the rest.</code></p> <p>Performance improvements (v0.1.0):</p> <ul> <li>Structured trailers eliminate JSON walks: metadata parsing is now O(1)</li> <li>Auto-index updates on every turn, cached at .gitlord/index.json</li> <li>New in-memory query layer for fast turn filtering and aggregation:</li> <li>Snapshot compression for long-running sessions: compress old turns into JSON, rebase from checkpoint</li> </ul> <p>Repo: <a href="https://github.com/yashneil75/gitlord" rel="noopener noreferrer">https://github.com/yashneil75/gitlord</a><br> Landing page: <a href="https://yashneil75.github.io/gitlord/" rel="noopener noreferrer">https://yashneil75.github.io/gitlord/</a><br> Docs: <a href="https://github.com/yashneil75/gitlord/tree/master/docs" rel="noopener noreferrer">https://github.com/yashneil75/gitlord/tree/master/docs</a></p> <p>I'm curious: what's one pain point with frameworks like LangChain and CrewAI that bugs you in the middle of the night?</p> <p>MIT licensed. Built for agents that ship.<br> Starring helps more than you think!</p> ai productivity python opensource