procurement-graph

A dependency graph of strategic sourcing artifacts and analyses, exposed as an MCP server.

This is an open specification I publish through Canonical Agency. It generalizes patterns from procurement and spend-intelligence engagements at two enterprises under NDA.

What it models

The strategic sourcing process is usually drawn as a linear set of phases. I find that framing brittle. Phases describe when work happens; they do not describe what depends on what. The same spend cube feeds the diagnostic, the category strategy, and the ongoing performance dashboard. The same supplier scorecard depends on baseline metrics captured during onboarding. Treating these as a graph (nodes plus edges) makes it possible to ask the questions that actually matter on an engagement:

• What deliverables are blocked by missing inputs?
• If raw spend changes, what gets invalidated?
• Given my entity-profile data, which analyses can I run today?
• What is the build order to produce a category strategy?

Phases (0 through 6) remain in the spec as a navigational frame. They are not the structural spine.

Phase model

Seven phases, zero through six, anchoring the program from opening diagnostic through ongoing performance management.

Phase 0  Opportunity Assessment & Business Case
Phase 1  Spend Analysis & Category Segmentation
Phase 2  Supply Market Research
Phase 3  Strategy Development
Phase 4  Sourcing Execution (RFx + Negotiation)
Phase 5  Implementation & Supplier Transition
Phase 6  Performance Management & Benefits Tracking

Layer model

Every node lives in one of five layers. Layers describe what kind of work a node is, independent of which phase it belongs to.

Layer 1  Truth and facts          (raw data, supplier master, contracts)
Layer 2  Analytical assets        (spend cube, should-cost, risk scores)
Layer 3  Decision artifacts       (category strategy, negotiation strategy)
Layer 4  Execution artifacts      (RFx packages, contracts, transition plans)
Layer 5  Proof and control        (benefits tracking, scorecards, compliance)

Tools (v0.4)

Navigation and catalog:

Tool	Purpose
list_phases()	Seven phases with one-line purpose
get_phase(number)	Full phase spec (purpose, included activities, layer focus)
list_analyses(primary_phase?, layer?, feasibility?)	Layer 2 analyses, filterable by feasibility with PO-line data
get_analysis(slug)	Full analysis spec (firms, inputs, method, outputs, watch-outs)
list_deliverables(primary_phase?, layer?)	Layer 3-5 deliverables
get_deliverable(slug)	Full deliverable spec (purpose, audience, components, dependencies)
list_horizontal_artifacts()	Cross-phase artifacts (governance, RAID, benefits tracker)
get_engagement_context()	Shared fictional engagement (TechnoFlex PSA resin) that anchors all examples
list_artifact_examples()	Filled-in artifact examples, each tied to a parent node slug
get_artifact_example(slug)	Fetch a filled-in TechnoFlex example for a given parent node slug

Graph traversal:

Tool	Purpose
get_dependencies(slug)	Direct upstream nodes for a given node
get_dependents(slug)	Direct downstream nodes for a given node
what_breaks_if(slug)	Transitive closure of downstream nodes (impact analysis)
build_order(target_slug)	Topological sort of every upstream node plus the target
feasible_now()	Nodes flagged HIGH feasibility against PO-line + entity-profile data
data_gap_analysis()	Nodes blocked by MISSING inputs, ranked by transitive impact

Install (local / editable)

cd /Users/fbaig/Projects/procurement-graph
uv sync
uv run procurement-graph

Validate the content corpus

procurement-graph validate runs the integrity checks the test suite enforces (edge consistency, acyclicity, example linkage, phase continuity) as a standalone CLI. Useful for content editors to verify their work before committing.

uv run procurement-graph validate

Claude Desktop config

{
  "mcpServers": {
    "procurement-graph": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/fbaig/Projects/procurement-graph",
        "run",
        "procurement-graph"
      ]
    }
  }
}

Repository layout

content/
├── phases/         seven phase descriptions
├── analyses/       Layer 2 analyses from the procurement analytics catalog
├── deliverables/   Layer 3-5 deliverables from the phase-by-phase taxonomy
├── horizontal/     cross-phase artifacts (governance, RAID, benefits tracker)
├── engagement/     shared fictional engagement (TechnoFlex PSA resin)
└── examples/       15 filled-in artifact examples anchored to the TechnoFlex engagement

src/procurement_graph/
├── __main__.py     entrypoint dispatching to MCP server or validate CLI
├── server.py       FastMCP app and tool registrations (16 tools)
├── loaders.py      cached file readers
├── parsing.py      YAML frontmatter and markdown body parsing
├── models.py       pydantic node, phase, and example models
├── graph.py        dependency-graph traversal (BFS, topo sort, feasibility)
└── validate.py     standalone content-integrity checks

The methodology content lives in content/ as markdown with YAML frontmatter, so the source of truth is editable without touching the server code.

License

Apache-2.0.