Split tips fairly, and prove it: 10 AI skills for the Tips Act

troncmaster-independence-check

test whether a tronc is really independent, HMRC's own way

How the two work together

Claude thinks it through. Paste the Claude prompt into Claude Code, or drop the folder into your skills folder. Claude does the judgement: what to look for, what is worth doing, what is right.

Codex gets it done. At the hand-off point Claude runs Codex on your machine with one command and passes it the Codex prompt. Codex does the mechanical part and hands the result back. Claude checks it before you see it.

No API key to set up: Claude calls the Codex you already have installed. If Codex is not installed, Claude does that half itself and tells you.

Prompt for Claude

---
name: troncmaster-independence-check
description: Runs HMRC's own worked-example test against how a specific tronc actually operates, to answer the one question that decides whether tronc tips attract employer National Insurance: does the employer determine, directly or indirectly, who gets what. It works from a plain account of who appoints the troncmaster, who sets the membership and who decides each allocation, because HMRC's guidance draws the line at the allocation decision itself, not at who the troncmaster reports to on paper. Use before you set up a tronc, when you appoint a new troncmaster, or before an HMRC inspection where the tronc's tax treatment will be tested.
---

# The tronc independence test, run the way HMRC actually runs it

You give this a plain account of who appoints the troncmaster, who decides membership, and who decides each allocation. You get back a verdict against HMRC's own stated test and its own worked examples, not a general impression of independence.

## What it does

1. **States the test exactly as HMRC states it.** HMRC's guidance on tips, gratuities, service charges and troncs says in section 8 that tips do not attract National Insurance contributions where "the employer does not determine, directly or indirectly, the allocation of those tips," and defines allocation as "deciding who should receive what amount by way of tips." The skill treats this single question, who decides the amount each person gets, as the whole test, because that is what the guidance itself tests.
2. **Rules out the employer acting as its own troncmaster.** The same guidance states that where an employer, partner or company official acts as troncmaster, "they are considered to be making payments as if they were the employer," which fails the test immediately regardless of how the arrangement is described elsewhere.
3. **Separates appointing a troncmaster from controlling what they decide.** HMRC's Example 10 finds that even though the employer appointed the troncmaster, "Mr Z has played no part, directly or indirectly, in the allocation of the tips," so no National Insurance is due; the act of appointment alone does not fail the test.
4. **Catches the arrangement that looks independent but is not.** HMRC's Example 11 finds National Insurance due because "Mr C is indirectly allocating the tips," where the troncmaster is shown to be carrying out the employer's wishes rather than deciding independently; the skill looks for this pattern, an employer instruction that the troncmaster then applies, rather than taking the troncmaster's formal title at face value.
5. **Confirms where influence is allowed to stop.** HMRC's Example 14 finds no National Insurance due where the employer influenced only who could join the tronc, not how much each member received, because "they are not directly or indirectly allocating the tips." The skill records membership decisions separately from allocation decisions, since the law treats them differently.

## Then it checks

1. The person or body making each allocation decision is named, and is confirmed not to be the employer, a partner, or a company official.
2. Any role the employer played in appointing the troncmaster is recorded separately from any role in deciding allocations, because only the second one matters to the test.
3. Any instruction, template or expectation the employer gave the troncmaster about how to split tips is recorded and checked against the Example 11 pattern.
4. Any employer involvement in who may join the tronc is recorded separately and confirmed not to extend to deciding amounts.
5. The verdict states which HMRC example the facts most closely match, not a bare pass or fail with no reasoning shown.

Any check fails: name it, redo that step once. Failed twice: say what is wrong and stop.

## Rules
- Public information only. Never invent a fact, a number or a quote.
- Anything sent in someone's name says whose name it is.
- Never give a National Insurance outcome as certain; this skill states which HMRC worked example the facts resemble, and the owner's accountant confirms the actual tax position.
- Never treat a troncmaster's formal independence on paper as sufficient if the facts supplied show the employer's instructions driving the actual split.
- This is a working document for the owner to check against their own arrangement and for HMRC guidance or their accountant to confirm before relying on it for tax purposes. It is not tax or legal advice.

## Built from
- HM Revenue and Customs, "Guidance on tips, gratuities, service charges and troncs" (E24), https://www.gov.uk/government/publications/e24-tips-gratuities-service-charges-and-troncs/guidance-on-tips-gratuities-service-charges-and-troncs, read 4 November 2026: the independence test, the employer-as-troncmaster rule, and Examples 10, 11 and 14, behind all five steps.
- Department for Business and Trade, "Code of Practice on Fair and Transparent Distribution of Tips", https://www.gov.uk/government/publications/distributing-tips-fairly-statutory-code-of-practice/code-of-practice-on-fair-and-transparent-distribution-of-tips-html-version, published 29 July 2024, in force 1 October 2024, read 4 November 2026: paragraph 30's warning that "care is needed to maintain independence," cross-checked against the HMRC test.

Prompt for Codex

# troncmaster-independence-check

## You are given
A plain account of: who appointed the current troncmaster, and whether that person is the employer, a partner, a company official, an external payroll or accountancy firm, or a staff member elected or agreed by workers. Any written or verbal instruction the employer has given the troncmaster about how to split tips. Any role the employer plays in deciding tronc membership, described separately from any role in deciding allocation amounts. A description of how a recent allocation was actually decided, in the troncmaster's or employer's own words.

## Produce
Write into a `./troncmaster-independence-check-output/` folder:

1. `troncmaster-profile.csv` with columns: `troncmaster_name`, `appointed_by`, `troncmaster_type`, `is_employer_partner_or_official`. `troncmaster_type` is exactly one of `external payroll or accountancy firm`, `elected or agreed staff member`, `employer, partner or official`, `unclear`. `is_employer_partner_or_official` is `yes` or `no`.
2. `allocation-control-check.csv` with columns: `instruction_or_pattern_described_verbatim`, `source`, `matches_hmrc_example`, `verdict`. `matches_hmrc_example` is exactly one of `Example 10 (appointment only, no allocation control)`, `Example 11 (indirect allocation, employer wishes carried out)`, `Example 14 (membership influence only)`, `no close match`. `verdict` is exactly one of `independent as described`, `employer is the troncmaster`, `indirect allocation described`, `insufficient information supplied`.
3. `membership-vs-allocation-split.csv` with columns: `decision_type`, `decided_by`, `employer_involvement_described`. `decision_type` is `membership` or `allocation amount`. Each decision type gets its own row so the two are never merged.
4. `verdict-summary.md` - the overall verdict from `allocation-control-check.csv`, the HMRC example it most closely matches, and a one-paragraph explanation in plain language of why, quoting the specific fact that decided it.
5. `gaps.md` - a numbered list of: any row in `troncmaster-profile.csv` where `is_employer_partner_or_official` is `yes`; any `allocation-control-check.csv` verdict other than `independent as described`; any decision type in `membership-vs-allocation-split.csv` with no `decided_by` supplied; any case where membership and allocation decisions were not clearly described separately.

## Rules
- Codex compares the supplied facts against HMRC's own stated test and its own worked examples. It never gives a National Insurance opinion and never tells the owner whether to set up or dissolve a tronc.
- Never mark `verdict` as `independent as described` where the facts show the employer, a partner or an official made or dictated an allocation decision, however the arrangement is formally described.
- Use British English, GBP and DD Month YYYY dates. No em dashes in any file you write.
- Every file ends with this line: this is a working document prepared for the owner to check against their own arrangement and for HMRC guidance or their accountant to confirm before relying on it for tax purposes. It is not tax or legal advice.

## Return
The absolute path of each file written and the row count of each CSV. The troncmaster's type and who appointed them. The overall verdict and the HMRC example it matches, with the deciding fact quoted. Every item flagged in `gaps.md`, in full.

Built from the best public work on this

Sources for troncmaster-independence-check

Everything below was opened and read on 4 November 2026.

1. HM Revenue and Customs, "Guidance on tips, gratuities, service charges and troncs" (E24)

https://www.gov.uk/government/publications/e24-tips-gratuities-service-charges-and-troncs/guidance-on-tips-gratuities-service-charges-and-troncs, read 4 November 2026.

Section 8 states tips do not attract National Insurance contributions where "the employer does not determine, directly or indirectly, the allocation of those tips," with allocation defined as "deciding who should receive what amount by way of tips." Section 6 states that where an employer, partner or company official acts as troncmaster, "they are considered to be making payments as if they were the employer."

Three worked examples draw the line precisely. Example 10: even though the employer appointed the troncmaster, "Mr Z has played no part, directly or indirectly, in the allocation of the tips," so no NI is due. Example 11: "Mr C is indirectly allocating the tips" where the troncmaster carries out the employer's wishes, so NI is due. Example 14: the employer influenced only who could join the tronc, not how much each person received, so "they are not directly or indirectly allocating the tips," and no NI is due.

Used for all five steps: this document is the whole test the skill runs.

2. Code of Practice on Fair and Transparent Distribution of Tips

https://www.gov.uk/government/publications/distributing-tips-fairly-statutory-code-of-practice/code-of-practice-on-fair-and-transparent-distribution-of-tips-html-version, published 29 July 2024, in force 1 October 2024, read 4 November 2026.

Paragraph 30 warns that "care is needed to maintain independence" where the operator is an external payroll or accountancy firm or a staff member elected or agreed by workers.

Used as a cross-check against the HMRC test, since the Code's own warning about the same risk, an employer quietly steering a tronc it calls independent, matches the pattern HMRC's Example 11 describes.

Where the skill departs: the Code's fairness test (paragraphs 21 to 26) and HMRC's tax independence test are two different questions that can produce different answers for the same tronc. This skill answers only the tax independence question; the fairness question is covered by tip-pool-split and tronc-record in this pack.

Best public prompt we found for this job

The closest public artefact is the `sox-testing` skill in Anthropic's `knowledge-work-plugins` repository, folder path `finance/skills/sox-testing`, referenced from the repository's own tree listing at https://api.github.com/repos/anthropics/knowledge-work-plugins/git/trees/main?recursive=1. The repository has 34,654 stars, read from api.github.com on 4 November 2026. A control-testing skill is built to test whether a control actually operates the way it is documented to operate, which is the exact shape of testing whether a tronc that is labelled independent is actually controlled independently.

We took the discipline of testing the described control against a named example of failure, rather than accepting the label on the control.

We did not copy its assumption that the test produces a pass or fail against an internal control framework the organisation itself wrote. Here the test is HMRC's own worked examples, a published external standard the business cannot redefine, so the skill names which specific HMRC example the facts match rather than scoring against a locally written control description.

Want this running in your business?

I optimise how businesses run — your sales, your visibility, your social media — and build bespoke software where nothing off the shelf fits. The first conversation is free. Work starts from £150 a day.