For AI agents: Markdown version of this page · llms.txt
Setup guide

Spend Request Agent for Claude - approvals with the policy check done first

The free Spend Request Agent runs on Claude as three plain-text files: set it up once in a Claude Project, approve a sample-data test, then trigger each run with the usage prompt and this run's inputs.

Get the Spend Request Agent - freeClaude overviewWhat’s new on Claude

What the Spend Request Agent does

If Claude is where your documents live, this agent slots into that workflow. It normalizes incoming spend requests and runs your policy and evidence checks: request, amount, policy check, evidence, approver and status. Approvals become one review, not a reconstruction.

Unrecorded approvals are next year's audit problem. This assistant structures each request as it arrives, checks it against your written policy, lists missing evidence, and drafts the response - with the decision trail saved per request. Hallway asks become a queue you can actually clear.

Where the Included files go in Claude

Create a Project for the job and add skill.md as Project knowledge with the setup prompt as instructions. Run the setup conversation, check the sample-data output, then start each run in that Project so the rules stay loaded. Artifacts Claude produces can be saved out to your Drive.

What a run looks like on Claude

A run looks like this: you open the Project, paste the usage prompt with this run's inputs, and Claude works through spend-approvals with the Project's knowledge loaded. The output - it normalizes incoming spend requests and runs your policy and evidence checks: request, amount, policy check, evidence, approver and status - comes back as an artifact you can save out to Drive. You approve, adjust or discard; nothing is final until you say so.

Setup, step by step

  1. Create a Project in Claude and name it after the job (for example: spend approvals).
  2. Add skill.md as Project knowledge and paste setup-prompt.txt into the Project's custom instructions.
  3. Run the setup conversation; the agent runs spend-approvals on fictional sample data first - approve that test before real work.
  4. Start each real run inside that Project and paste usage-prompt.txt with this run's inputs. Save artifacts Claude produces out to your Drive.

Honest limits on Claude

Without a Project, Claude chat context resets between sessions and you re-paste the rules. Keep the skill file in Project knowledge so the workflow stays put.

Questions people ask

Does the Spend Request Agent work on Claude?

Yes - the agent is three plain-text files, and on Claude they live in a Claude Project. The four setup steps below are the whole job: Create a Project in Claude and name it after the job (for example: spend approvals). Then add skill.md as project knowledge and paste setup-prompt.txt into the project's custom instructions.

What does the Spend Request Agent produce on Claude?

It normalizes incoming spend requests and runs your policy and evidence checks: request, amount, policy check, evidence, approver and status. Claude hands it back as an artifact you save to your Drive. Every run waits for your approval before anything is sent, spent, published or scheduled.

Is this affiliated with Claude?

No. AI Agent Skills is an independent store. We are not affiliated with, endorsed by, or sponsored by Claude or its makers. Platform names belong to their owners; capabilities change, so check Claude's current documentation.

Run this agent on another assistant

Microsoft Copilot Hermes Instinct OpenClaw Muse Perplexity ChatGPT ChatGPT Dots DeepSeek Town Grok Gemini

More finance teams agents on Claude

Cash Flow Forecast Agent Budget Variance Agent Management Reporting Agent

AI Agent Skills is an independent store, not affiliated with Claude or its makers. Platform names belong to their owners.