Version history for every DjangoPlay app — pick one to see what changed and when. Looking for what’s coming next? Explore the latest roadmap and upcoming releases?
v1.1.3 keeps the existing gitaiflow AOT CLI as the core execution engine and adds a thin MCP layer around it. You can now use the same Git change intelligence from Claude Desktop or other MCP clients, run the MCP server locally or in Docker, select repositories from a multi-repository workspace, and install signed release artifacts with checksum and signature verification.
The AOT CLI remains the core gitaiflow implementation; MCP is an adapter that delegates Git, diff, AI, caching, usage, changelog, release-note, and artifact work to the existing executable.
Claude Desktop users can install the packaged MCPB extension directly. The extension is an additional integration path alongside the existing AOT binary, not a replacement for it.
Developers can run MCP over stdio for local clients or streamable HTTP for service-style deployments, with bearer-token authentication and request rate limiting.
A configured parent workspace can contain multiple independent Git repositories, with MCP selecting and validating the requested repository inside that workspace.
Docker deployment mounts the parent repository workspace and the matching native AOT binary, keeping the MCP layer isolated without creating a second Git/AI execution engine.
AOT installers now enforce SHA-256 checksum verification and can verify Ed25519 release signatures before completing installation.
The release pipeline separates AOT and MCP artifacts while keeping their version relationship explicit for v1.1.3.