{
"markdown": "# model-ledger\n\n**git for models** โ know what models you have deployed, where they run, what they depend on, and what changed.\n\n[](https://github.com/block/model-ledger/actions/workflows/ci.yml)\n[](LICENSE)\n[](https://python.org)\n[](https://pypi.org/project/model-ledger/)\n[](https://pypistats.org/packages/model-ledger)\n[](https://block.github.io/model-ledger/)\n\n๐ **[Documentation](https://block.github.io/model-ledger/)** ·\n[Quickstart](https://block.github.io/model-ledger/quickstart/) ·\n[Concepts](https://block.github.io/model-ledger/concepts/) ·\n[Governance](https://block.github.io/model-ledger/governance/)\n\n---\n\nmodel-ledger is a model inventory for any organization with deployed models. It\n**discovers** models, heuristic rules, and ETL across your platforms, maps the\n**dependency graph** automatically, and records **every change as an immutable event**.\nUnlike registries tied to a single platform (MLflow, SageMaker, W&B), it spans all of\nthem โ as one connected graph โ and it's built to be driven by AI agents through a\nnative MCP server.\n\nBenchmarked at production scale: full inventory reconstruction over a ledger of 28.8k\nmodels and 212k events runs in under a second ([CHANGELOG, v0.7.4](CHANGELOG.md)).\n\n## Install\n\n```bash\npip install model-ledger\n```\n\n## The graph builds itself\n\nEvery model is a `DataNode` with typed input and output ports. When an output port name\nmatches an input port name, `connect()` creates the dependency edge โ no hand-wiring.\n\n```python\nfrom model_ledger import Ledger, DataNode\n\nledger = Ledger()\n\nledger.add([\n DataNode(\"segmentation\", platform=\"etl\", outputs=[\"customer_segments\"]),\n DataNode(\"fraud_scorer\", platform=\"ml\", inputs=[\"customer_segments\"], outputs=[\"risk_scores\"]),\n DataNode(\"fraud_alerts\", platform=\"alerting\", inputs=[\"risk_scores\"]),\n])\nledger.connect()\n\nledger.trace(\"fraud_alerts\")\n# ['segmentation', 'fraud_scorer', 'fraud_alerts']\n```\n\nEvery mutation is recorded as an immutable **Snapshot** โ an append-only event log that\ngives you full history and point-in-time reconstruction, because nothing is overwritten.\n\n## Talk to your inventory\n\nThe MCP server is a first-class surface โ point Claude (or any MCP agent) at it:\n\n```bash\npip install \"model-ledger[mcp]\"\nclaude mcp add model-ledger -- model-ledger mcp --demo\n```\n\n> **You:** if we deprecate `customer_features`, what breaks?\n>\n> **Claude:** 3 models consume it directly, 2 more transitively.\n\n## Documentation\n\nEverything lives at **[block.github.io/model-ledger](https://block.github.io/model-ledger/)** โ\nand it can't drift, because the API reference is generated from source and every example\nruns in CI:\n\n- **[Quickstart](https://block.github.io/model-ledger/quickstart/)** โ install to your first dependency trace in 60 seconds\n- **[Concepts](https://block.github.io/model-ledger/concepts/)** โ DataNode, Snapshot, and Composite, in three ideas\n- **[Agents (MCP)](https://block.github.io/model-ledger/guides/agents/)** โ the eight-tool agent surface, with a worked transcript\n- **[Connectors](https://block.github.io/model-ledger/guides/connectors/)** โ discover from SQL, REST, GitHub, or your own platform\n- **[Backends](https://block.github.io/model-ledger/guides/backends/)** โ in-memory, SQLite, JSON, Snowflake, or remote HTTP\n- **[Governance](https://block.github.io/model-ledger/governance/)** โ how the primitives map to SR 11โ7/SR 26โ2, the EU AI Act, and NIST AI RMF\n- **[API reference](https://block.github.io/model-ledger/reference/)** โ generated from the source\n\n## Architecture\n\n```mermaid\nflowchart LR\n subgraph Sources\n C1[SQL / REST / GitHub / Prefect
connectors]\n end\n subgraph Core\n L[Ledger
append-only event log,
point-in-time reconstruction]\n G[Dependency graph]\n V[Compliance profiles
SR 11-7/SR 26-2 ยท EU AI Act ยท NIST AI RMF]\n end\n subgraph Surfaces\n S1[Python SDK]\n S2[CLI]\n S3[REST API]\n S4[MCP server ยท 8 tools]\n end\n B1[(in-memory ยท SQLite ยท JSON ยท
Snowflake ยท remote HTTP)]\n C1 --> L\n L --> G\n L --> V\n L --- B1\n S1 --> L\n S2 --> L\n S3 --> L\n S4 --> L\n```\n\n## For organizations\n\nThe OSS core handles discovery, graph building, change tracking, storage, the agent\nprotocol, and compliance validation โ the SR 11โ7/SR 26โ2, EU AI Act Annex IV, and\nNIST AI RMF profiles ship in `model_ledger.validate`. Your internal package provides\nonly the thin layer on top: connector configs, custom connectors for internal\nplatforms, and credentials. Thin config, not reimplemented logic.\n\n## Contributing\n\nSee [CONTRIBUTING.md](https://github.com/block/model-ledger/blob/main/CONTRIBUTING.md).\nAll commits require DCO sign-off.\n\n## Security\n\nSee [SECURITY.md](SECURITY.md) for how to report vulnerabilities privately.\n\n## License\n\nApache-2.0. See [LICENSE](https://github.com/block/model-ledger/blob/main/LICENSE).\n\nCreated and maintained by [Vignesh Narayanaswamy](https://github.com/vigneshnarayanaswamy) at Block.\n\n\n",
"bytes": 5493,
"sha": "23dd47e60169377fef6bc04a9cca7032b79b59a16197030e23156061a5f2cf15",
"repo_slug": "block/model-ledger",
"fonte": "repo",
"truncated": false,
"api": "https://agentalog.com/api/listings/mcp_io_github_block_model_ledger_72bc022d/readme"
}