Policy Compliance Assessment

Policy compliance assessment

Work out which planning policies govern an application, and whether it accords with each one. The skill does three things:

1. Identify the adopted development plan for the relevant local planning authority, and verify it — the essential first step, because determination starts there and nothing downstream is safe if this is wrong.
2. Assess the proposal against each relevant policy, on the evidence in the application documents and the policy's own words.
3. Score and report — a policy-by-policy table with an accordance score from -2 to +2, then a reasoned conclusion on accordance with the development plan read as a whole.

National policy (NPPF/PPG) and any emerging plan are assessed too, but as material considerations carrying less weight than the adopted plan — never as a substitute for it.

When to use

The user asks: "does this comply with the local plan?", "which policies does this application breach?", "assess this against the adopted policies", "what does the local plan say about this site?", "is this development plan compliant?" — or a representation needs a policy foundation before it is drafted.

Use it after application-triage has identified which considerations are engaged (or alongside it — triage's decision-framework step hands off to this skill), and before policy-representation drafts anything.

Not this skill: the deep technical evaluation of a topic's evidence base (the ecological-, transport-, heritage- and flood-representation skills own that), the current NPPF edition and policy codes (national-planning-policy owns those), or the final harms-vs-benefits weighing (planning-balance).

Two layers, no duplication

This skill owns the procedure: how to find and verify an adopted development plan, how to read a policy, and how to score accordance. It owns no citations of its own. The current NPPF/PPG edition register, the verify-before-citing protocol, s.38(6) and plan primacy, the presumption in favour of sustainable development (NPPF S3–S6), emerging-plan weight and the conditions/obligations tests all live in the companion national-planning-policy skill — take them from there. The policies themselves are per-instance data: they come from the council's adopted plan at run time, and are quoted, never remembered.

The integrity principle (read first)

• The adopted plan is the local authority's own document, and it has primacy. Adopted policies are made and published by the local planning authority — on its own website, under its own numbering, adopted on its own date. They are not published by central government, and they are not the NPPF. This matters twice over: it tells you where to look (the council's planning-policy pages, never a national source), and it fixes the hierarchy — the development plan is the statutory starting point, and national policy is a material consideration that informs the decision without displacing the plan. An assessment that leads on the NPPF and treats the local plan as background has the hierarchy upside down.
• "Adopted" is a formal term — establish it, don't assume it. A plan is adopted when the council has formally resolved to adopt it following independent examination. A published, submitted, consulted-on or examined-but-not-yet-adopted plan is not adopted, and neither is a supplementary planning document. Getting this wrong is the single most damaging error available here: an assessment against draft policy numbers, or against policies that a later plan superseded, is worthless and visibly so.
• Quote the policy; never recall it. Policy wording and numbering vary between councils and between editions of the same plan, and draft numbering rarely survives adoption. Every policy in the output must be quoted from the adopted document, with the document name, adoption date and policy reference recorded.
• A score is a reasoned judgement, not a measurement. It is a shorthand for an argument that must be stated alongside it. A score with no quoted requirement and no evidence from the application is not an assessment.
• Never total or average the scores. Section 38(6) requires accordance with the development plan read as a whole, which is a planning judgement, not arithmetic. One -2 against a policy central to the proposal can outweigh five +1s on peripheral ones; the reverse is also true. Presenting a sum or a mean would manufacture false objectivity — and would let a genuinely fatal conflict be averaged away.
• Missing evidence is not conflict. Where a policy requires something the application has not supplied, the honest output is "cannot be assessed" (flagged ?), not a negative score. That distinction is what separates "the proposal conflicts with Policy X" from "the Council cannot yet conclude whether it complies with Policy X" — a different ask entirely.
• Score the proposal against the policy, not against a preference. Record accordance where it exists, at full strength. An analysis that finds only conflict will be read as advocacy and discounted.

Workflow

Step 1 — Identify and verify the adopted development plan (do this first)

Check the application type before anything else. If it is an amendment — a s.96A non-material amendment, or a s.73 variation or removal of conditions — there is a baseline to reconstruct before the assessment means anything (see the Step 4 discipline "On an amendment application, reconstruct the permission before scoring anything"). It bears on this step directly: a s.73 grants a new permission, so the plan you identify here is the plan as it stands now, not the one the parent permission was granted under.

Work through references/finding-the-development-plan.md. In outline:

1. Confirm the local planning authority. Usually the district, borough or unitary council — but a National Park authority or the Broads Authority is the LPA for its own area, and a development corporation may be. In two-tier areas, minerals and waste policy sits with the county council. The wrong LPA means the wrong plan.
2. Assemble the development plan. Under s.38 of the Planning and Compulsory Purchase Act 2004 it comprises the adopted development plan documents taken as a whole plus any made neighbourhood plan for the area; in Greater London it also includes the Mayor's spatial development strategy (the London Plan). It is often several documents — a core strategy plus a site-allocations and/or development-management document, sometimes a joint plan, plus the county minerals and waste plans. ⏳ Verify the composition against the current statute at run time — see the reference file's note on pending reforms.
3. Record the adoption date and status of each document, and find the schedule of superseded and saved policies — a new plan usually replaces only some of its predecessor's policies. Never cite a policy without checking it has not been superseded.
4. Read the policies map for the site: settlement boundary, allocations, designations and constraints. The map is part of the plan and often decides which policies apply.
5. Note what is not the development plan but may still be a material consideration: supplementary planning documents, design guides and codes, conservation area appraisals, council strategies, and emerging or withdrawn plans. Keep these in a separate tier.

Output a short plan register — document, adoption date, status, where obtained — before assessing anything.

Step 2 — Fix the plan's status

Record the plan's age and review position: its adoption date, the plan period, whether a review or replacement is underway, whether it was examined and adopted against the current (August 2026) NPPF, and whether the council can demonstrate the required housing land supply and Housing Delivery Test performance. This governs how much weight a conflict carries — and the mechanics changed in the August 2026 edition (take the tests themselves from the national-planning-policy skill rather than restating them here):

• there is no longer a general "out-of-date plan" presumption switch: the old tilted balance is gone, and the national presumption now turns on location (within/outside a settlement, NPPF S4/S5), not plan status;
• Annex A(2): a plan policy (or part of one) materially inconsistent with the Framework's national decision-making policies carries only "very limited weight" unless the plan was examined and adopted against this Framework — but a policy is not given reduced weight merely because it pre-dates the Framework. Record which applies, and why;
• Annex A(3): a recently adopted plan is shielded — with a five-year supply, HDT above 75% and adoption within the last five years, a standard-method figure above the adopted housing requirement does not count as evidence of unmet need for the S5(1)(j) gateway;
• housing land supply / HDT below 75% now feed the S5(1)(j) unmet-need gateway outside settlements (and grey belt) rather than triggering a general presumption.

A conflict with a reduced-weight policy still counts; say so, and state the reason for the reduction, rather than silently discounting it.

Step 3 — Select the relevant policies

Use references/policy-families.md to work from the proposal and site to the policy families that ought to exist in any local plan, then find each one in this plan. Two disciplines:

• Relevance, not volume. Include a policy because it applies to this development, on this site, at this scale — not to lengthen the list. Check each policy's own scope: many apply only to a development type, size threshold or designated area.
• Flag the "most important" policies for determining this application — the handful the decision turns on (typically the spatial strategy or settlement-boundary policy, the site-specific allocation or designation, and the principal topic policies). The flag still matters even though the August 2026 NPPF's presumption no longer hinges on whether the "most important policies" are out-of-date: it is the set the accordance statement leads on, and where any Annex A(2) consistency question bites hardest.

Step 4 — Assess each policy

For each policy: quote the requirement in its own words, then state what the application documents show, then conclude. Reading discipline:

• Distinguish mandatory from permissive wording. "Will not be permitted", "must" and "will be required" are requirements; "should", "will be encouraged", "where possible" and "have regard to" are weaker. The strength of the wording sets the ceiling on the score.
• Criterion-based policies are assessed criterion by criterion. Where a policy permits development only if a list of criteria is met, failing one is conflict with the policy even if the rest are met — identify which criterion and why.
• Read the whole policy, including any exception or flexibility limb the proposal might rely on. Supporting text and the reasoned justification are not policy, but are legitimate aids to interpreting it — label them as such.
• Designated areas raise the bar — score them that way. In a conservation area, or where a listed building or other designated heritage asset (or its setting) is affected, the plan's character, design and scale policies are not ordinary detail policies: a statutory duty and national policy's "substantial weight" stand behind them (take the citations and the harm framework from the heritage-representation skill — s.66/s.72 and the NPPF heritage tests live there). Two consequences for this skill's procedure: (a) assess the cumulative volume of physical change quantitatively from the drawings — footprint, depth, height, plot coverage, extensions and alterations taken together, measured against the plot and its neighbours, not merely described; (b) where that change fails a character or scale criterion of a policy applying to the designated area, treat the conflict as going to the heart of the policy (-2 territory, and normally one of the "most important" policies) rather than softening it to a tension because only one limb fails or because each individual alteration looks modest.
• Mitigation that is not yet secured is not compliance. If a policy's requirement is met only by something a condition or obligation would have to secure, say so — that is a (C) point, and it belongs in the score's reasoning.

Evidence disciplines — habits observed in officer practice that the assessment must match:

• Audit the openings window-by-window. Compare existing and proposed elevations opening by opening: every new or altered window and door, which neighbour it faces, and whether it creates overlooking a condition (obscure glazing, non-opening) would have to control. A narrative read of the elevations misses exactly the opening the decision turns on.
• Sweep the site's planning history first, and assess against any fallback. Check the register for the site's (and close precedents') history before assessing: an extant permission is the controlling baseline, and the assessment narrows to the delta between it and the current proposal; past refusals and appeals on the site or its immediate context are weight-bearing precedent.
• On an amendment application, reconstruct the permission before scoring anything. A s.96A non-material amendment or a s.73 variation or removal of conditions supplies only the delta — a covering letter and a drawing or two — and cannot be assessed on those documents alone. Retrieve the parent permission's decision notice, full condition set, approved plans, officer report and s.106, and every earlier amendment, and rebuild what the permission consists of today; the application-triage skill's references/amendment-applications.md carries the method, the routes and their limits. Four consequences for this assessment:
  • the baseline is the permission as already varied, and the proposal is scored as the change against it — not as if the site were undeveloped and unpermitted;
  • a s.73 grants a new permission, so it is assessed against the plan and the evidence base as they stand now, not as they stood when the parent was granted — check whether any policy relied on then has since been superseded;
  • where an earlier officer report concluded a policy was complied with because of a particular measure, an amendment that removes or dilutes that measure puts the policy back in issue, and the conflict is recorded now even though the parent scheme complied;
  • assess the cumulative effect of the whole chain as well as this change — for a s.96A that is the statutory question (s.96A(2) TCPA 1990), and for any amendment it is what catches a scheme that has drifted a defensible increment at a time.
• Measure against the neighbour as well as the plot. Amenity geometry is relative: projection beyond the neighbour's rear building line, orientation to their windows and garden, and relative levels — computed from the drawings, not asserted.
• Verify the basic site facts from more than one source. Attachment status (detached / semi / terrace), plot orientation and constraints — state them with their evidence, and check the council's own GIS/policies-map layers as well as national datasets; local designations (minerals belts, ecology zones) often appear only on the council's layers.
• Read the representations. Consultee responses and neighbour/third-party comments are part of the evidence: they surface issues, site knowledge and precedents the application documents omit (handle any personal data minimally).
• Consultee positions are evidence, not conclusions. Re-derive each element's assessment from the documents; decision-makers routinely depart from their own specialists in both directions, and a consultee's general dispensation still has to survive the policy's own wording at this site's scale.
• Prefer a condition to a refusal reason for separable detail. Where an element is acceptable in principle and only its detail is missing, the officer's instinct is to reserve it by condition — a (C) point — not to refuse; reserve refusal reasons for harm. Likewise a missing supporting document on a minor scheme is a proportionate-information ask (B), not automatically a refusal reason.

Step 5 — Score

Apply the rubric in references/scoring-rubric.md:

Alongside each score record the weight tier (adopted development plan / reduced — e.g. an Annex A(2) material inconsistency, or a time-expired policy / national policy / emerging plan / guidance) and, where the downstream skills need it, the A/B/C classification the repo uses: (A) demonstrated unacceptable impact, (B) insufficient evidence, (C) resolvable by condition or obligation. Every ? is a (B).

Step 6 — Add national and emerging policy, at their proper weight

Assess the relevant NPPF/PPG policies and any emerging plan policies the same way, in a separate section of the table, explicitly marked as material considerations that carry less weight than the adopted plan:

• National policy — a material consideration, influential but not above the plan (though under Annex A(2) it claws back weight from materially inconsistent plan policies — see Step 2). Verify every citation through the national-planning-policy skill before citing: the August 2026 NPPF replaced paragraph numbers with coded policies, so any paragraph-number citation is stale on its face; ⏳ re-verify the edition at run time.
• Emerging plans — weight depends on the stage of preparation, the extent of unresolved objections, and the degree of compliance with the Framework's plan-making policies (NPPF DM4(1)). Record the stage and reason the weight; an emerging policy at early consultation carries very little, and saying so is part of the assessment.
• Guidance (SPDs, design codes, technical standards) — weight as guidance that supplements plan policy; it cannot create policy the plan does not contain.

Step 7 — Conclude on the plan read as a whole

Write the accordance statement. Open with one sentence stating the conclusion — whether the proposal accords with the development plan read as a whole (no arithmetic). Then support it with short bullets, not a run-on paragraph:

• the plan documents that govern;
• the policies conflicted with, each with a one-line severity;
• the policies accorded with (a compact list is fine);
• anything that cannot yet be assessed, and why.

Close with one sentence on what follows — whether material considerations (national policy, emerging policy, the scheme's benefits) might indicate a decision otherwise than in accordance with the plan — and hand off to planning-balance for that weighing and to policy-representation for drafting.

Output format

1. Plan register — the development plan documents, adoption dates, status, source.
2. Site's plan status — designations from the policies map; allocation or settlement position; plan age and review position.
3. Policy table — one row per policy: reference · source document · weight tier · quoted requirement · assessment (with the document evidence) · score · most-important flag · A/B/C. Keep the assessment cell to one or two short sentences. If a policy needs more, put the detail in a bulleted note below the table keyed to the policy reference — never a semicolon-run inside the cell.
4. A second table for national and emerging policy, marked as lower-weight material considerations. Same cell discipline.
5. Accordance statement — conclusion sentence, supporting bullets, what-follows sentence (Step 7); no totals.
6. Verification note — what was checked, when, and what needs re-checking at run time.

Output style. This report is read by busy people: any enumeration of three or more items goes in a bulleted or numbered list, never strung through a paragraph with semicolons. One point per paragraph; keep paragraphs to three sentences or fewer. The representation skills' references/house-style.md files state the same discipline for drafted representations — this skill's analysis output follows it too.

Reference files

• references/finding-the-development-plan.md — how to identify and verify the adopted plan: what counts, where to look, the traps (superseded policies, draft numbering, SPDs, the wrong LPA), and emerging-plan stages.
• references/scoring-rubric.md — the -2 to +2 anchors with worked illustrations, the evidence-gap flag, the weight tiers, and the no-aggregation rule.
• references/policy-families.md — the policy families to look for by proposal type, and which skill owns each one's technical evaluation.

Scope and limitations

• Not legal advice, and no warranty. Policy interpretation is ultimately for the decision-maker, and contested wording is a matter for a planning professional or a solicitor; provided "as is", with no guaranteed outcome.
• Human review is necessary before use. A person must check every quoted policy against the adopted document and every assessment against the application documents.
• England-focused. The development-plan framework cited is England's; Wales, Scotland and Northern Ireland differ — flag and adjust outside England.
• Evidence-bound. Quote policies and application documents; never invent a policy reference, a plan date or a requirement. Where the plan is unclear, say it is unclear.
• Time-sensitive. Plans are adopted, superseded and reviewed continuously, and national policy editions turn over (most recently the August 2026 recoding) — re-resolve the plan and re-verify every citation at run time.
• A UK planning representation is public and in the submitter's name — this skill produces an analysis rather than a submission, but carry that warning through to policy-representation or whichever skill drafts from it.
• "It complies" is a valid, valuable output. So is "the plan is silent on this."