--- oa_review_kit: v1 guide_slug: financial-reporting-workflow-base guide_version: financial-reporting-workflow-base@2026-05-29T07:05:13.592Z archetype: personal_income --- # Review kit: Financial Reporting Workflow Base Thank you for reviewing this Guide. This kit is one file with three parts: how to use it, an interview prompt for your AI, and the Guide itself. ## How to use this kit (3 steps, about 15 minutes) 1. Open the AI you already use (ChatGPT, Claude, Gemini, anything that reads markdown) and paste in everything from "INTERVIEW PROMPT" below, including the Guide at the end. 2. Your AI interviews you like a colleague, one question at a time. Just talk: war stories, walk-throughs, the mistakes you catch. No writing required. 3. Your AI writes your answers up as a single markdown file. Hand it back at openaccountants.com/skills/financial-reporting-workflow-base/handback (also linked from the Guide page: "Hand back your file"). What you added is published under your name and credential. If your AI cannot produce the exact output format, hand back whatever you have: a revised Guide file, a worksheet, or plain notes. We take those too, and a person reviews them by hand. The format below is the one we can apply straight away. --- # INTERVIEW PROMPT (paste from here down into your AI) You are interviewing a practising accountant about how they actually do the work covered by the attached Guide ("Financial Reporting Workflow Base", slug `financial-reporting-workflow-base`). Interview them like a colleague doing a handover. Do not lecture. Ask ONE question at a time and wait for the answer. Chase war stories and specifics: what kind of client, which portal step, how big the penalty was. The rates, thresholds, and citations are our job; we refresh those from primary sources. Capture ONLY what is NOT derivable from law: - order of operations, and what a wrong order corrupts - what to ask a client before computing anything - what to assume when a fact is unknown, and how it gets flagged - the most-missed traps, with penalty size and who falls in - how the portal or filing channel actually behaves - what has to reconcile before anyone signs - when to refuse the work and hand it to a human specialist If the accountant corrects a rate, threshold, or deadline in the Guide along the way, record it in the FACT CORRECTIONS table, but do not steer the interview toward numbers. ## Questions to work through Ask these in order, one at a time. Skip any the accountant has already covered; follow up where a story has specifics worth pinning down. Each question is tagged with the method slot(s) it feeds. 1. [intake_questions] A new personal-tax client sits down. What are your first five questions before you touch a number? 2. [intake_questions] [conservative_default] Which of those answers, if missing, makes you stop rather than estimate? 3. [evidence] Which documents do you insist on seeing, and which do you take the client's word for? 4. [sequence] In what order do you build the return, and what goes wrong when someone does it backwards? 5. [trap] What deduction or relief do clients most often believe they're entitled to but aren't? 6. [trap] What computation does software or AI most often get wrong on returns you've reviewed? 7. [judgment_rule] When the law gives two routes (regime choice, standard vs itemized, allowance vs actuals), how do you actually pick, and what do you write down about the choice? 8. [cross_check] What triggers the tax authority's mismatch letters in your experience, and what do you reconcile up front to prevent them? 9. [edge_case] Tell me about a client whose side income (platform, rental, foreign) changed the whole shape of the return. What did you do differently? 10. [filing_mechanics] Walk me through the e-filing itself: the verification step, the deadline nobody knows, what happens if the client doesn't do their part. 11. [scope_gate] What kind of personal-tax client do you turn away or send to a specialist? 12. [handback_protocol] What exactly do you hand the client before anything is filed? What's in your working paper? ## Method slots (for tagging the write-up) - `scope_gate` (Scope gate and refusals): when to stop and send the client to a human - `sequence` (Order of operations): what order to do things in, and what a wrong order corrupts - `intake_questions` (Client intake questions): what to ask a client before computing - `evidence` (Documents and evidence): which documents to insist on, and what is draft-grade vs file-grade - `judgment_rule` (Judgment rules): how a practitioner actually picks when the law allows two routes - `conservative_default` (Conservative defaults): what to assume when a fact is unknowable at draft time - `trap` (Traps and most-missed items): the mistakes everyone makes, what they cost, and who falls in - `filing_mechanics` (Portal and filing mechanics): how submission actually works: channel, order, what locks - `cross_check` (Cross-checks before signing): what has to reconcile with what before delivery, and how close is close enough - `pattern_library` (Pattern library): how messy real-world data (bank lines, payout platforms) maps to tax categories - `edge_case` (Edge-case playbook): the client situations that change the method, not just the numbers - `unsettled_law` (Unsettled-law flags): what not to finalise right now, and why - `handback_protocol` (Hand-back protocol): what the finished working paper contains and who reviews it ## Output format: oa-handback v1 When the interview is done, write the answers up as ONE markdown file in exactly this shape. Fill in the reviewer's real name, credential, and email (ask for them at the end if they have not come up). Every method block gets a `### [method:]` heading where `` is one of the 13 slot ids above. Keep `guide_slug` and `guide_version` exactly as given. Omit any section the interview produced nothing for, but keep the headings that remain exactly as shown. The `fact_key` column may be left blank when unknown. ```markdown --- oa_handback: v1 guide_slug: financial-reporting-workflow-base guide_version: financial-reporting-workflow-base@2026-05-29T07:05:13.592Z reviewer_name: reviewer_credential: # free text: CPA, EA, ACCA, Steuerberater... reviewer_email: verdict: --- ## METHOD ### [method:filing_mechanics] ### [method:intake_questions] - - ... ## FACT CORRECTIONS | fact_key | current | correct | source | |---|---|---|---| | | | | | ## FLAGS - [unsettled] - [refer] ## NOTES ``` If for any reason you cannot produce this exact format, output the accountant's corrections and methods as clear plain notes instead. The hand-back page accepts plain notes and revised Guide files too; this format is an optimization, never a gate. --- # THE GUIDE UNDER REVIEW --- name: financial-reporting-workflow-base description: Foundation workflow base for financial-reporting (US GAAP and IFRS) content skills. Contains the universal runbook, the two-layer output contract (reference layer + executable computation layer), the journal-entry format, the dual-standard convention, the AUDIT FLASH POINT marker convention, conservative-default principle, the structured question form, and the universal self-checks. This skill provides workflow architecture only — it contains no standard-specific recognition or measurement content. It MUST be loaded alongside a topic content skill (e.g. us-gaap-asc606-revenue, ifrs15-revenue) that provides the actual recognition, measurement, presentation, and disclosure rules. This base is the foundation every financial-reporting content skill loads on top of. jurisdiction: GLOBAL domain: income-tax tax_year: 2025 --- # financial-reporting-workflow-base ## Financial Reporting Workflow Base v0.1 ## What this file is **This file is the foundation workflow base that every financial-reporting content skill loads on top of.** It is the shared contract. It carries no recognition or measurement rules of its own — those live in the topic content skills (revenue, leases, debt vs. equity, business combinations), each issued in a US GAAP (ASC) edition and an IFRS edition. A content skill is useless without this base, and this base computes nothing without a content skill. Load both. **Currency.** Content skills state their own standard-effective-date currency. This base is standard-agnostic and does not expire. **The preparer and the reviewer are the customers of this output.** These skills assume a qualified preparer (controller, financial accountant, audit associate) is applying the skill to a real transaction, and that a credentialed reviewer (CPA, ACCA, Chartered Accountant, or equivalent) reviews and signs the resulting working papers. The skill produces working papers, journal entries, and a reviewer brief — **not** filed financial statements and not audit evidence. ## Section 1 — The two-layer output contract Every financial-reporting content skill is built in **two layers**. An agent invoking a skill produces both unless the user asks for only one. ### Layer A — Reference layer (the rules) - **Layer A — Reference layer (the rules)** — A faithful, citation-anchored statement of the standard's recognition, measurement, presentation, and disclosure requirements for the topic, expressed as decision trees and rules an agent can reason over. The reference layer answers: "What does the standard require, and where is it written?" Every rule cites the governing paragraph (e.g. ASC 606-10-25-1, IFRS 15.31, IAS 32.16, ASC 842-10-15-3). _(ASC 606-10-25-1, IFRS 15.31, IAS 32.16, ASC 842-10-15-3)_ ### Layer B — Executable layer (the computation) - **Layer B — Executable layer (the computation)** — A step-by-step procedure that takes a specific transaction's facts and produces: 1. The recognition/measurement conclusion (with the reference-layer rule that drives it). 2. The journal entries (see Section 3 format), at initial recognition and over the relevant subsequent periods. 3. The amounts that flow to the primary statements (B/S, P&L / OCI, cash flow classification). 4. The disclosure checklist triggered by the transaction. Layer B answers: "Given these facts, what do I book, and what do I disclose?" Rule: Layer B steps must reference the Layer A rule number they execute. No computation without a cited rule. ## Section 2 — The dual-standard convention Each topic ships as two files: a US GAAP (ASC) edition and an IFRS edition. They are separate skills, never merged, because the citations and — frequently — the answers differ. Each content skill MUST include a "Divergence from [the other framework]" section that flags, for the topic, where US GAAP and IFRS reach different recognition, measurement, presentation, or disclosure outcomes. This is the single most valuable section for a dual-reporter (e.g. a US subsidiary of an IFRS parent) and the section reviewers scrutinise most. When a user's facts touch both frameworks (group reporting under one, statutory under the other), an agent loads both editions for the topic and runs Layer B twice, presenting the two answers side by side with the divergence called out. ## Section 3 — Journal entry format - **Journal entry format template** — [Date / period] — [Event description] — driving rule: [ASC/IFRS ref] Dr Dr Cr Cr (memo: how each amount was derived) Rules: - Debits must equal credits on every entry. State the check. - Every account caption must map to a primary-statement line (asset, liability, equity, revenue, expense, OCI). - Where an amount is a present value, an allocation, or an estimate, the memo line shows the inputs (rate, term, standalone selling prices, fair values) and which Layer A rule produced it. - Show the subsequent-measurement entries, not only day 1 — unwinding of discount, amortisation, remeasurement, reclassification. _(Section 3 — Journal entry format)_ ## Section 4 — AUDIT FLASH POINT convention - **AUDIT FLASH POINT marker convention** — Mark every position that auditors, regulators (SEC, FRC, ESMA), or peer reviewers actively challenge with a bold marker: > **⚑ AUDIT FLASH POINT —** [the judgement at issue, why it is contested, what evidence supports the position taken, and the disclosure that defuses it.] These are not edge cases for completeness; they are the judgements a reviewer must personally own. An agent must surface every flash point a transaction triggers in the reviewer brief, never bury them. Typical flash-point families across topics: estimates and significant judgements, related-party and non-arm's-length terms, cut-off and period-end timing, principal-vs-agent and gross-vs-net, classification that changes a key metric (EBITDA, leverage, current ratio), and anything that moves an amount between P&L and OCI or between periods. _(Section 4 — AUDIT FLASH POINT convention)_ ## Section 5 — Conservative-default principle - **Conservative-default principle** — When the facts are incomplete and a judgement could go either way, the skill takes the position that is harder to challenge and easier to unwind, states that it has done so, and lists the specific fact that would change the answer. It does not silently pick the favourable treatment. Aggressive positions require an affirmative, documented instruction and a named reviewer. _(Section 5 — Conservative-default principle)_ ## Section 6 — Structured intake (the question form) - **Structured intake question form** — If the transaction facts needed to run Layer B are missing, the agent asks for them in a single structured block before computing — never piecemeal, never guessing. Group questions by the Layer A decision they unlock. Example shape: To apply [topic] under [standard] I need: 1. [Fact] — used in [step / rule ref] 2. [Fact] — used in [step / rule ref] ... If you don't have [fact], I will assume [conservative default] and flag it. _(Section 6 — Structured intake (the question form))_ ## Section 7 — Universal self-checks - **Universal self-checks checklist** — Before delivering output, the agent verifies (the content skill adds topic-specific checks on top): - [ ] Both layers produced (reference rule cited for every Layer B step) - [ ] Every journal entry balances; debits = credits stated - [ ] Every account caption maps to a primary-statement line - [ ] Subsequent-measurement entries shown, not just day 1 - [ ] Every figure traces to a source input or a cited rule — no unsourced numbers - [ ] Standard edition matches the entity's reporting framework (ASC vs IFRS) - [ ] Divergence section checked where the entity is a dual-reporter - [ ] Every triggered AUDIT FLASH POINT surfaced in the reviewer brief - [ ] Conservative default applied and flagged wherever facts were incomplete - [ ] Disclosure checklist produced for the transaction - [ ] Effective-date / transition correct for the reporting period - [ ] Output marked with its verification tier; reviewer sign-off requested _(Section 7 — Universal self-checks)_ ## Section 8 — Output specification - **Output specification order** — Deliver, in this order: 1. Conclusion — one paragraph: recognition/measurement outcome and the rule that drives it. 2. Reference-layer trace — the decision path through Layer A, with citations. 3. Journal entries — Section 3 format, day 1 + subsequent periods. 4. Statement impact — what hits B/S, P&L, OCI, cash flow; effect on key metrics. 5. Disclosure checklist — every disclosure the transaction triggers, with reference. 6. Reviewer brief — the judgements taken, every AUDIT FLASH POINT, conservative defaults applied, and the open questions the credentialed reviewer must personally clear before sign-off. 7. Divergence note — where the other framework would differ (if relevant). _(Section 8 — Output specification)_ ## Section 9 — Verification status and disclaimer - **Verification status tiers and disclaimer** — Each content skill carries a quality tier. Research-verified means drafted from the authoritative standards and awaiting credentialed sign-off; Accountant-verified means a named, licensed practitioner has reviewed and signed off the section. These skills provide computational and interpretive guidance on accounting standards. They are not an audit, not an assurance engagement, and not a substitute for professional judgement. Recognition, measurement, and disclosure under US GAAP and IFRS frequently turn on entity-specific facts and significant judgement. Always have outputs reviewed and signed by a qualified accountant before they are reflected in financial statements relied upon by third parties. _(Section 9 — Verification status and disclaimer)_