agentleFS
Sign inStart free
norman-financenorman-finance/norman-mcp-server/skills/run-workflow/SKILL.md

run-workflow

Skill65 starsChanged 11 days agoMITView on GitHub

Run a Norman workflow (month-end close, monthly reconciliation, VAT-return readiness, invoice to payment, client document requests) on Norman's servers, answer the questions it asks, and schedule it to run by itself. Use when the user asks Norman to close a month, reconcile the bank, get ready for the VAT return, or "do it every month".

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Call list_workflows. Match $ARGUMENTS or the user's words to a workflow key:

    line 18
  2. Distinctive:

    If a run of that workflow is already active, continue with it (Step 3) instead of starting another.

    line 22
  3. Very specific:

    Call start_workflow with the key.

    line 26
  4. Very specific:

    Call get_workflow_run when the user asks how it is going.

    line 30
  5. Very specific:

    When state is done or done_with_warnings, summarize each step's summary in plain words.

    line 39
  6. Distinctive:

    Start, answer, schedule and stop only when the user asked.

    line 44

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Norman runs its workflows on its own servers. You start one, relay the user's answers, and report the result. You do not do the steps yourself.

Step 1: Pick the workflow

  • Call list_workflows. Match $ARGUMENTS or the user's words to a workflow key:
    • "close the month" → month-end-close
    • "reconcile" → monthly-reconciliation
    • "VAT return", the German "UStVA", the Polish "JPK_V7" → vat-readiness
  • If a run of that workflow is already active, continue with it (Step 3) instead of starting another.
  • Tell the user which steps it will take (the steps of the workflow) and for which period (defaultScope, or the month they named).

Step 2: Start it

  • Call start_workflow with the key. Pass period_start/period_end only if the user named a different period.
  • Say that Norman is now working on it and will stop only if it needs them.

Step 3: Follow the run

  • Call get_workflow_run when the user asks how it is going. Do not poll in a loop.
  • Act on blockedReason:
    • user_input: show the question from blockedDetail. Send the user's own answer with answer_workflow_question. Never answer on their behalf.
    • manual_step: explain the step's instruction. Once the user did it, or wants to skip it, call complete_workflow_step.
    • source_required: ask them to connect the bank or source in Norman, then continue_workflow.
    • step_failed: say what stopped it (from blockedDetail). Offer continue_workflow to try again.
    • ai_limit or chat_limit_reached: this month's AI messages are used up. The workflow can continue when the messages renew next month.

Step 4: Report and offer the schedule

  • When state is done or done_with_warnings, summarize each step's summary in plain words.
  • If the workflow has no schedule, offer to run it by itself: month-end close on day 3, reconciliation on day 5, VAT readiness 5 days before the deadline (before_vat_deadline follows the German UStVA only: for a Polish company schedule it on a day of the month before the JPK_V7 is due on the 25th). Only after the user agrees, call schedule_workflow.
  • A refusal with code automation_paid_feature means automatic runs need a paid plan. Say so; running by hand stays free.

Important:

  • Start, answer, schedule and stop only when the user asked.
  • Invoice to payment and client document requests send to clients only when the user asked for sending. Confirm before you relay a "yes, send it".
  • For "what did Norman do this week?", call get_agent_week and quote its numbers.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free