Tool · Workflows & capabilities / Data & provenance

Capability Graph Importer

Turn capability packages and Foundry blueprints into a graph plan you can inspect before importing.

Repository: CinvanaAI/capability-graph-importer

Explore the repository ↗Try or inspect it ↓

Recorded synthetic example. Source with a complete offline example.

See the idea in action.

Two packages, ten nodes, eight edges, and the exact source filenames and SHA-256 hashes in the final plan.

Example input

Normalize Text as JSON and Format Name as a historical-style text blueprint.

Captured result

{
  "schema_version": "capability-graph.plan.v1",
  "source_label": "Synthetic Packages",
  "nodes": [
    {
      "id": "package:Format Name",
      "label": "Package",
      "properties": {
        "name": "Format Name",
        "source": "Synthetic Packages"
      }
    },
    {
      "id": "logic:Format Name",
      "label": "Logic",
      "properties": {
        "package_name": "Format Name",
        "source": "def format_name(value):\n    return value.strip().title()"
      }
    },
    {
      "id": "argument:Format Name:value",
      "label": "Argument",
      "properties": {
        "name": "value",
        "record": {
          "name": "value"
        }
      }
    },
    {
      "id": "return:Format Name:formatted",
      "label": "Return",
      "properties": {
        "name": "formatted",
        "record": {
          "name": "formatted"
        }
      }
    },
    {
      "id": "recognition:Format Name:synthetic-example",
      "label": "Recognition",
      "properties": {
        "name": "synthetic-example",
        "record": {
          "name": "synthetic-example"
        }
      }
    },
    {
      "id": "package:Normalize Text",
      "label": "Package",
      "properties": {
        "name": "Normalize Text",
        "source": "Synthetic Packages"
      }
    },
    {
      "id": "logic:Normalize Text",
      "label": "Logic",
      "properties": {
        "package_name": "Normalize Text",
        "source": "def normalize(value):\n    return value.strip().lower()"
      }
    },
    {
      "id": "argument:Normalize Text:value",
      "label": "Argument",
      "properties": {
        "name": "value",
        "record": {
          "name": "value"
        }
      }
    },
    {
      "id": "return:Normalize Text:normalized",
      "label": "Return",
      "properties": {
        "name": "normalized",
        "record": {
          "name": "normalized"
        }
      }
    },
    {
      "id": "recognition:Normalize Text:synthetic-example",
      "label": "Recognition",
      "properties": {
        "name": "synthetic-example",
        "record": {
          "name": "synthetic-example"
        }
      }
    }
  ],
  "edges": [
    {
      "from": "package:Format Name",
      "type": "HAS_LOGIC",
      "to": "logic:Format Name"
    },
    {
      "from": "package:Format Name",
      "type": "HAS_ARGUMENT",
      "to": "argument:Format Name:value"
    },
    {
      "from": "package:Format Name",
      "type": "HAS_RETURN",
      "to": "return:Format Name:formatted"
    },
    {
      "from": "package:Format Name",
      "type": "USES_RECOGNITION",
      "to": "recognition:Format Name:synthetic-example"
    },
    {
      "from": "package:Normalize Text",
      "type": "HAS_LOGIC",
      "to": "logic:Normalize Text"
    },
    {
      "from": "package:Normalize Text",
      "type": "HAS_ARGUMENT",
      "to": "argument:Normalize Text:value"
    },
    {
      "from": "package:Normalize Text",
      "type": "HAS_RETURN",
      "to": "return:Normalize Text:normalized"
    },
    {
      "from": "package:Normalize Text",
      "type": "USES_RECOGNITION",
      "to": "recognition:Normalize Text:synthetic-example"
    }
  ],
  "provenance": {
    "package:Format Name": [
      {
        "source_type": "blueprint",
        "source_file": "format.blueprint.txt",
        "source_sha256": "beb4cf17f0fcd9b2f00dab8d9be04423213ac98856a2fc5ccce462e908780ac7"
      }
    ],
    "package:Normalize Text": [
      {
        "source_type": "json",
        "source_file": "normalize.package.json",
        "source_sha256": "0f408e7fa02525a17f10703fdb5f955e7c914a56145e0e003989bbf1f62f327a"
      }
    ]
  },
  "conflicts": []
}

Try the example.

From the repository root, follow the dependency requirements in the README. This example uses supplied synthetic material.

python -m pip install -e .
python -m examples.walkthrough

Two packages, ten nodes, eight edges, and the exact source filenames and SHA-256 hashes in the final plan.

Complete setup and instructions ↗

The interesting part.

Conflicting duplicate definitions require an explicit preference and can retain the conflicting sources.

Public continuation

Where it came from.

Graph-import script in the historical Python Agent Foundry workbench.

Rebuilds discovery, merging and graph projection as a deterministic graph plan with source evidence.

It ends at a graph plan; the historical direct database writer is not the public interface.

Source ↗

Follow the family: Builder, Factory and the Python Foundry

Extracted fromPython Agent Foundry Workbench
See the integrated setting from which this independently useful mechanism was separated. Rebuilt from the historical script; the public interface now emits a plan.
Source ↗