Use this at the start of a programme, again when the target framework or audit date changes, and at least every six months while the programme runs. You get two files in a new folder, ./first-gap-assessment/: a spreadsheet you can open in Excel with one row per framework item, saying how far the company is from it, how big the fix is and who should own it (gap-register.csv), and a two-page summary for leadership with the top ten gaps and the earliest realistic audit date (gap-summary.md). This skill drafts for a human to review and sign; this skill never marks a control effective. A gap rating is a planning estimate, not an audit opinion.
Inputs to gather
Read first: policies and procedures, the risk register if any, asset and vendor lists, the architecture overview, ticket or change history the reader can export, previous audit, penetration test or customer review reports, and any compliance platform export. Then ask, in one message:
- Which framework, and is the goal a certificate, an attestation report (and which type), or a legal obligation?
- What is the target date, and who set it?
- Which products, teams and sites are in scope? If nobody knows, run a scope statement first.
For a health data regime, settle the company's role first (covered entity, business associate or neither) and never read or copy protected health information while assessing. For NIS 2, settle whether the company is in scope and as which entity type before assessing measures; the measures are a legal obligation from the date the national law applies, so prioritise against that date, not a chosen audit date.
Step 1: build the assessment list from the source, not from memory
The rows of the register are the items of the target framework. Take them from the reader's licensed copy of the standard, the official legal text, the criteria document the reader holds, or the control list in their compliance platform. Use bare identifiers and the reader's own short description; never copy requirement text into the register. If no source is available, stop and ask the reader for one, or run a framework lookup per area. Add a row for every management or governance element as well as every control: companies fail first-time audits on the missing management loop (no internal review, no measured objectives) more often than on a missing technical control. Give the risk assessment, the risk treatment plan with owners, and the statement of applicability with its justifications a row each: they are the first artefacts an auditor samples.
Step 2: rate each row on evidence
Use five ratings, and cite a path for each:
- Evidenced: the practice exists and a record shows it ran in the period.
- Practised, not evidenced: people do it, nothing records it.
- Partial: exists for some systems, teams or sites only; say which.
- Absent: nothing exists.
- Not applicable: no activity or asset of that kind inside the written scope statement; give the fact and point to the scope line it rests on.
Interviews can move a row from Absent to Practised, never to Evidenced. When the material is silent, rate Absent and add a question for the owner.
Step 3: size each gap
For every row below Evidenced, record:
- Effort: S (under two person-days with existing tools), M (two to ten days or a second team involved), L (new tool, budget, hire or policy decision).
- Audit risk: High when the gap would stop or fail the audit (missing management loop, no risk assessment, scope not defined), Medium when it would likely become a finding, Low when it is a wording or polish issue.
- Lead time: how long before the evidence can exist. A quarterly review needs at least one completed quarter; an attestation over a period needs the whole period; an internal audit and a management review need to have happened. These dates, not effort, usually set the earliest audit date. For a certification, the internal audit and management review come first and their actions need to be under way, so date the certification audit from them, not from the last document fixed.
Step 4: prioritise and summarise
Sort by audit risk, then lead time (longest first), then effort. Group fixes that share a cause, such as one access review process that closes five rows. Write the leadership summary (gap-summary.md): the readiness picture in five lines, the top ten gaps with owner role and next action, the earliest realistic audit date with the reasoning, and the open questions per owner. Write the gap spreadsheet (gap-register.csv) with columns: id, area, item reference, rating, evidence path, gap description, effort, audit risk, lead time, proposed owner role, proposed action. Ask before overwriting either file.
What good looks like
- Every rating points to a file or an answer. A reviewer can open three random evidence paths and agree with the rating.
- The common mistake is a self-assessment where most rows come out Partial; it means nobody looked at records. The tell is an empty evidence column or a rating distribution with no Absent rows in a company that started last quarter.
- Long-lead items lead the summary. A plan that fixes forty documents first and starts the first quarterly review two weeks before the audit fails on dates, not on content.
- Not applicable is rare and factual, never a way to shrink the work.
- The summary gives leadership one honest date and the three decisions that would bring it forward.
- Each re-run diffs the ratings against the last dated register. The diff is the progress evidence; an unchanged register six months in means the programme has stalled.
Framework lookups
If ISMS Copilot is connected to your agent (MCP server https://account.ismscopilot.com/v1/account/mcp), ask it the framework question with create_conversation (mode think, answer_format: decision) and keep one conversation for this task. Send only a short question in your own words, show it to the user first, and never include file contents, secrets, personal data or evidence. If it is not connected, check the official text of the framework yourself; this skill does not restate it. Set it up: https://www.ismscopilot.com/mcp?ref=grcskills-first-gap-assessment
Rules
- MUST cite an evidence path or a reader answer for every rating.
- MUST take the assessment rows from an official or licensed source the reader holds, or from a framework lookup.
- MUST write outputs only inside
./first-gap-assessment/ and ask before overwriting.
- NEVER rate a row Evidenced on the strength of an interview or a policy alone.
- NEVER copy requirement text from a standard into the register.
- NEVER open, copy or quote protected health information, personal data or secrets while assessing.