Skip to content
Get packs

Guide · Agent skill

How to use the Procurement Optimizer agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The procurement-optimizer skill is an agent skill for Claude that turns a set of spending records into a category review, a purchasing-cycle bottleneck report and a supplier consolidation plan. It is aimed at finance and BizOps leads running an annual SaaS audit or category spend review. After this guide you will know how to install the skill, how to phrase requests that trigger it, what input files it needs and what limits to respect.

Procurement Optimizeralirezarezvani/claude-skills · MIT

Quick start

  1. 1Download and unzip the skill. Download the ZIP from the skill page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. Alternatively, upload the ZIP under Settings → Capabilities → Skills in the Claude apps.
  2. 2Confirm Python 3 is available. The scripts need Python 3 and use only the standard library, so no extra packages are required.
  3. 3Prepare your input files. Fill in assets/spend_intake_template.md to build spend, purchase-order and supplier JSON files for your business.
  4. 4Ask for the review in plain language. Use a request such as a spend audit or supplier consolidation request, and attach your files.
  5. 5Run the categorizer. Run scripts/spend_categorizer.py with your spend file and chosen profile to get categories and a Pareto breakdown.
  6. 6Run the cycle analyzer and consolidation planner. Run scripts/purchasing_cycle_analyzer.py on purchase orders and scripts/supplier_consolidation.py on suppliers to get bottlenecks and risk-flagged consolidation plans.
  7. 7Review the digest. Check top growth categories, bottlenecks, consolidation opportunities and any tier-1 items marked DO NOT CONSOLIDATE before acting.

What is an agent skill and what does this one do?

An agent skill is a folder of instructions, reference material and optional scripts that an AI agent loads when a request matches the skill's purpose. This skill is described as working like a procurement or BizOps lead reviewing company spending.

Once installed, the agent sorts line items into a UNSPSC-style category map with a Pareto breakdown showing which categories drive most cost. It measures how long purchase orders take per category, flags bottlenecks, and drafts a supplier consolidation plan for duplicate tools.

  • Spend categorization into a UNSPSC-aligned Class, Family and Segment map (a built-in map of about 30 categories, not the full UNSPSC database)
  • Pareto analysis of which categories drive most spend, plus top year-over-year growth categories when prior-year data is provided
  • Purchasing-cycle analysis with bottleneck flags, based on Goldratt's Theory of Constraints
  • Risk-balanced supplier consolidation for duplicate-function clusters
  • The bundled Python scripts do the arithmetic, so the numbers are not left to the model to guess

How do you install the procurement-optimizer skill?

You can install the skill in one of two ways. Download the ZIP from the skill page on this site and unzip it into a skills folder. Use ~/.claude/skills/ for personal use across projects, or a project's .claude/skills/ folder to limit it to one project.

Alternatively, upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

  • Option A: unzip into ~/.claude/skills/ for a personal install
  • Option B: unzip into the .claude/skills/ folder inside a project
  • Option C: upload the ZIP in the Claude apps under Settings → Capabilities → Skills
~/.claude/skills/

What do you need before you run it?

The skill needs Python 3. Its scripts use only the Python standard library, so there are no packages to install. It does not use the network.

You also need to prepare input files following the templates in assets/spend_intake_template.md. The skill expects a spend JSON file, a purchase-order JSON file and a supplier JSON file. The intake template suggests gathering the top 100 to 200 line items by spend from your accounting, SaaS-management, corporate-card or expense exports.

  • Python 3
  • Spend JSON, purchase-order JSON and supplier JSON in the formats from the templates
  • Line-item fields include supplier, description, category_hint, annual_spend, frequency and currency
  • Prior-year spend is optional; without it the year-over-year analysis is skipped

How do you trigger the skill with natural requests?

You do not need to call the skill by name. Describe the job in plain language using the kinds of phrases the skill is written to respond to. The skill's trigger phrases include spend audit, SaaS audit, spend categorization, supplier rationalization, supplier consolidation, category strategy, duplicate SaaS and renewal cluster.

Be specific about the goal and attach your input files. Mention the profile if you know which one fits your business.

  • "Run a spend audit on this export and show me the Pareto by category."
  • "Find duplicate SaaS tools in this supplier list and draft a consolidation plan."
  • "Which purchasing categories are bottlenecks based on these purchase orders?"
  • "Categorize this spend using the scaleup profile."

What does a typical run look like?

The skill's workflow has five steps. First, you fill in the intake template. Second, the spend categorizer maps line items to categories and produces a Pareto. Third, the purchasing-cycle analyzer computes cycle times for each category. Fourth, the supplier consolidation planner finds duplicate clusters and flags risk. Fifth, the results are combined into a single review digest covering top growth categories, bottlenecks, consolidation opportunities and renewal clusters.

You can run the scripts directly as well. Each accepts --input, --output, --sample and --help, and the categorizer and consolidation planner accept --profile with tech-startup, scaleup, enterprise, services or manufacturing.

  • Categorizer: scripts/spend_categorizer.py --input spend.json --profile <profile> --output categorized.md
  • Cycle analyzer: scripts/purchasing_cycle_analyzer.py --input pos.json --output cycle.md
  • Consolidation planner: scripts/supplier_consolidation.py --input suppliers.json --profile <profile> --output consolidation_plan.md
cd business-operations/skills/procurement-optimizer && python3 scripts/spend_categorizer.py --sample

How does the consolidation planner protect critical categories?

The planner will not recommend collapsing a tier-1 category to a single supplier unless your input documents a break-glass plan. Where that plan is missing, the output says: DO NOT CONSOLIDATE, tier-1 cluster, no break-glass on record, and asks you to add a 72-hour contingency plan first.

Savings are estimated as current cluster spend minus the winner's spend minus migration cost, where migration cost is the sum of switching-cost estimates for the losing tools. The planner also flags categories where three or more contracts renew in the same calendar month, since that clustering weakens negotiating leverage.

  • Tier-1 single-source recommendations are blocked without a documented break-glass plan
  • Savings = current cluster spend − winner spend − migration cost
  • Renewal clusters are flagged when three or more contracts renew in the same calendar month

Safety notes: scripts and network access

The skill runs Python scripts on your machine, so review the scripts in the scripts/ folder before running them on sensitive financial data. The skill is documented as not using the network, so your spend files are not sent anywhere by the skill itself.

Treat the input files as confidential. They can include supplier names, contract amounts and approver information. Keep copies in a location your organization allows for financial records.

  • The skill runs scripts: yes
  • The skill uses network: no
  • Inspect scripts before running them on real financial exports
  • Keep input files confidential

Tips for getting useful results

The category map is tuned for tech-startup spend. If your spending is heavy in specialised areas, results may misclassify some items, so check the categorized output against your own knowledge. The Pareto holds for most businesses, so the top 100 to 200 line items usually answer most questions.

Use the sibling vendor-management skill for scoring a vendor you already plan to keep. Use this skill for deciding what to buy and from whom.

  • Check specialised spend categories manually, since the map is not the full UNSPSC database
  • Start with the top line items by spend
  • Provide prior-year data if you want growth analysis
  • Do not rely on the consolidation output for tier-1 categories without a real contingency plan

What people use it for

  • Annual SaaS audit to find which software categories drove spend growth
  • Identifying duplicate monitoring, expense or email-marketing tools and building a consolidation case
  • Finding purchasing categories where approvals slow every order down
  • Merging two procurement taxonomies and supplier lists after an acquisition
  • Spotting renewal dates clustered in one month before negotiating

Frequently asked questions

Is the procurement-optimizer skill free?

The skill is published under the MIT licence, which is stated on this page. Check the licence file in the download for the full terms before using it commercially.

What do I need to run the procurement-optimizer skill?

You need Python 3, since the scripts use only the standard library. You also need input files for spend, purchase orders and suppliers that follow the templates in the skill's assets folder.

How do I install the procurement-optimizer skill in Claude?

Download the ZIP from the skill page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

Does the skill send my spend data to the internet?

The skill is documented as not using the network. It does run Python scripts locally, so review the scripts before running them on sensitive financial data.

Can the skill replace a full UNSPSC database?

No. The categorizer uses a built-in map of about 30 categories tuned for tech-startup spend, not the full UNSPSC database, so specialised spend may be misclassified.

How is this different from the vendor-management skill?

This skill focuses on what to buy and from whom, covering categories and consolidation. The vendor-management skill is for scoring an individual vendor you already plan to keep. Contract drafting and financial close are routed to other skills.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you are a finance or BizOps lead running an annual SaaS audit, a category spend review, or a consolidation case for duplicate tools. Pick something else if you need vendor scoring, contract drafting or P&L analysis, or if your spend is heavily specialised and needs the full UNSPSC database.

Download the skill