traffic-management-plan

name	traffic-management-plan
description	Produces a construction traffic management plan for a named site — vehicle and pedestrian routing, segregation, loading/delivery and reversing controls, and the site-access and public-interface arrangements. Use this skill whenever a user asks to write or review a traffic management plan, a TMP, a site traffic plan, vehicle-pedestrian segregation, or construction logistics/access controls for a specific site. It separates people from vehicles by design (one-way systems, segregated routes, eliminating reversing), controls deliveries and the public interface, sets speed and signage rules, and applies the hierarchy of controls — emitting a branded report. Grounded in CDM 2015 Regulation 27 and Schedule 3. Decision-support only; a competent person must review the output.
license	Apache-2.0
metadata	{"author":"eyekyam","version":"1.0","category":"compliance","tier":2,"audience":["M","C","F"],"industry":["Con"],"jurisdiction":["All"],"status":"stable","plugin":"hse-construction","hse_reviewed_by":"","hse_reviewed_date":""}

Traffic Management Plan (CDM 2015 Reg 27 + Schedule 3)

A consultant-grade HSE skill that produces a task/site-specific construction traffic management plan (TMP) for a named site, grounded in CDM 2015 Regulation 27 (organise the site so pedestrians and vehicles can move safely) and Schedule 3 (traffic routes — suitable, sufficient and separated, with warning of approach). It plans the vehicle and pedestrian routing, the vehicle-pedestrian segregation by design, the one-way / turning / reversing controls, the loading & delivery management, the speed / signage rules, and the public / live-highway interface for the named site. It forces the single lever that separates a defensible TMP from copy-paste paperwork: a named site with real access points and named routes plus the segregation-by-design hierarchy — never a vague plan whose only pedestrian control is "hi-vis and a banksman".

Segregation-by-design is the core value. A pedestrian control of "hi-vis and a banksman" with no physical or temporal segregation is flagged PPE/admin-led and pushed up the hierarchy (KB-SNIP-TRAFFIC-SEGREGATION): eliminate the conflict / design out reversing → one-way & turning arrangements → physical segregation (barriers, separated routes) → signage, speed limits & lighting → a banksman / traffic marshal as the LAST resort. Controls are ranked via KB-SNIP-HOC and (where the residual conflict needs scoring) the deterministic controls engine. A generic site, an unnamed access point, or uncontrolled reversing with no Reg 27 / Schedule 3 grounding is a refuse-to-plan gate, not an assumption — the skill never invents the site layout.

When to use this skill

Use this skill when the user needs a traffic management plan for a concrete, named construction site — for example "write a TMP for the Meadowbank Road site: two gated accesses, HGV deliveries reversing into a loading bay, operatives crossing to the welfare cabins", "review this site traffic plan whose only pedestrian control is a banksman in hi-vis", or "plan vehicle-pedestrian segregation and the delivery / reversing controls for our city-centre plot with a footway adjacent to the hoarding". Trigger phrases: traffic management plan, TMP, site traffic plan, vehicle-pedestrian segregation, one-way system, reversing controls, banksman / traffic marshal, loading / delivery management, construction logistics, site access, public / highway interface, CDM 2015 Reg 27, Schedule 3, HSG144. The load-bearing inputs are the named site + its access points and at least the vehicle and pedestrian types that interface; if any is missing, the Workflow intake below refuses to proceed until they are elicited (it never invents a layout, a route, or an access point).

Data Protection & De-identification (MANDATORY — apply before drafting)

Apply this BEFORE you draft anything. Treat injury, illness, and any health detail as the highest sensitivity. Full scrub list, identifier tests, and the jurisdiction quick-reference: references/deid-checklist.md.

1. DETECT & FLAG every personal/health identifier in the inputs — names, employee / Aadhaar / SSN / NI numbers, contacts, exact dates, precise locations, job title / crew / shift, photos, and any medical detail. List what you found before drafting. If unsure whether something is identifying, treat it as identifying.
2. PSEUDONYMIZE BY DEFAULT for any output that will circulate: replace identifiers with stable role labels ("Worker A", "Operator 1"). Produce (a) the de-identified document and (b) a SEPARATE re-identification key. Never put the key or any name↔label mapping in the document. Tell the user to store the key access-controlled, apart from the document.
3. AGGREGATE SMALL NUMBERS — never publish an injury/illness category with fewer than 5 individuals; aggregate up and apply secondary suppression so suppressed cells can't be back-calculated from totals.
4. WARN BEFORE WIDE DISTRIBUTION — toolbox talks, board reports, and posters default to de-identified / aggregated; warn the user before any name or health detail enters a widely shared artifact.
5. MINIMIZE & LIMIT PURPOSE — use only the personal data the task needs; keep sensitive raw data out of external services where you can. When in doubt, ask before including it.

Knowledge base (read ONE matching file — never load all)

Resolve the user's jurisdiction first. Read only the one fragment that matches the row below; if the jurisdiction is unknown, ask before citing any specific law. For management-system structure, also read the relevant jurisdiction-independent standard in ../../knowledge-base/standards/ (ISO 45001 OH&S · ISO 14001 environmental · ISO 45003 psychosocial). Always apply ../../knowledge-base/prompt-snippets/hierarchy-of-controls.md (KB-SNIP-HOC) to every control recommendation. For any benchmark/figure, look up the ID in the relevant _registry.yaml, then read ONLY the named file — and quote its source+year.

Jurisdiction / scope	Read
UK	../../knowledge-base/regulatory/cdm-2015.md (CDM 2015 Reg 27 organise the site for safe vehicle/pedestrian movement + Schedule 3 traffic routes — suitable, sufficient, separated, warning of approach — the traffic-routes row; pairs with HSE HSG144 The safe use of vehicles on construction sites)
USA	../../knowledge-base/regulatory/osha-1926.md (29 CFR 1926 Subpart O motor vehicles, mechanised equipment & marine operations + 1926.601 / .602 — site traffic / equipment access)
India	../../knowledge-base/regulatory/in-factories-act.md (+ in-state-forms.md for the user's state) — mandatory state detection; defers to hse-india; the state site-traffic obligation is a literal [GAP], never a minted national form number
EU	../../knowledge-base/regulatory/eu-osh.md
Unknown	Ask before citing any specific law
Segregation-by-design (every run)	../../knowledge-base/prompt-snippets/traffic-segregation.md (KB-SNIP-TRAFFIC-SEGREGATION — the segregation-by-design control hierarchy: eliminate the conflict / design out reversing → one-way & turning → physical segregation → signage / speed / lighting → banksman LAST; a "hi-vis + banksman only" pedestrian control is the prevented downgrade)

This skill always grounds in KB-STD-ISO45001 (6.1.2 hazard ID + 8.1.2 hierarchy of controls) and applies KB-SNIP-HOC to every control. The traffic-specific grounding is the segregation-by-design hierarchy in KB-SNIP-TRAFFIC-SEGREGATION (read it every run) plus, for a UK site, CDM 2015 Reg 27 + Schedule 3 via KB-REG-CDM2015 (the traffic-routes row — routes suitable, sufficient and separated, with warning of approach) and HSE HSG144; the bundle clause cross-walk KB-SNIP-CONSTRUCTION-CLAUSE-MAP carries the Reg 27 + Schedule 3 → Traffic Management Plan row that places this skill in the hse-construction document chain. For a US site it cites 29 CFR 1926 Subpart O (+ 1926.601 / .602) via KB-REG-OSHA1926. For an India site it resolves the state via KB-REG-IN-STATEFORMS (mandatory state detection — defers to hse-india; confirm the state before citing any obligation; emit a literal [GAP], never a national form number) with the Factories Act framing in KB-REG-IN-FACTORIES. The rule-9 manifest is references/_skill-kb.md.

Workflow

Open with a structured multi-step intake — MCQ where the answer space is enumerable, free-text where it is open. Ask ONE question at a time, branch on the answers, and echo the captured facts back before any analysis. Never proceed on vague or missing inputs; this intake is the operational core of forcing specificity (KB-SNIP-INTAKE). (Intake is a Workflow convention, not a sixth block.)

Step 0 — Structured intake (run this first, one question at a time)

The full typed, branched intake — the intake-coverage manifest, the question table (Q1 site & layout free-text [named site + access points + constraints — refuse a generic site] · Q2 traffic types multi-select [HGV deliveries · plant / MEWP · forklift / telehandler · light vehicles · pedestrians / operatives · public / highway] · Q3 known conflict points free-text [gates, loading bays, crossings, blind corners, shared routes] · Q4 delivery & reversing profile MCQ [scheduled banked / ad-hoc / continuous — reversing present → push toward elimination via one-way / turning areas, banksman as the LAST resort] · Q5 public & highway interface MCQ [none / footway adjacent / live highway] · Q6 jurisdiction), the mandatory India → state branch (Q6 = India → Q6a + KB-REG-IN-STATEFORMS, defers to hse-india, literal [GAP], never a national form number), the echo-back, and the refuse-on-vague anchors — lives in references/intake.md. Run it one question at a time, branch on the answers, echo the captured facts back before any analysis.

The GATE (refuse-on-vague): no traffic management plan is produced until a named site with its access points (Q1) and at least the vehicle and pedestrian types that interface (Q2) are captured. The skill refuses to plan on a generic site, an unnamed access point, or "the site roads" — ask again, or record [ASSUMPTION] / [GAP]; never invent a layout, a route, an access point, or a conflict point. Uncontrolled reversing with no Reg 27 / Schedule 3 grounding is a citation + specificity failure — name the routes, do not produce a generic plan.

The traffic-management method (CDM 2015 Reg 27 + Schedule 3 + the segregation-by-design hierarchy)

Full method in references/METHODOLOGY.md. The skill reads the segregation hierarchy and ranks controls; it uses no risk-scoring engine for the plan body (the optional controls engine ranks/validates a residual conflict's treatment). Steps:

1. De-identify the inputs — before any drafting (the deid block above + the De-identifier-runs-first orchestration rule). This is the lowest-PII skill in the pack (site / route-level), but any named driver / operative / banksman that arrives in the inputs (e.g. a named driver in a prior reversing near-miss, a fitness-for-duty note) is scrubbed to a role label before any analysis. The TMP is a site/route-level document — it carries no named individual in the circulated plan; see references/deid-checklist.md.
2. Map the site & routes (the specificity anchor) — from Q1/Q3 lay out the named vehicle routes, the named pedestrian routes, the access points, the loading / delivery bays, and the conflict points (gates, crossings, blind corners, shared routes). An unnamed route or a generic "site roads" is a [GAP] and a stop — never invented.
3. Identify every vehicle-pedestrian conflict point — for each Q2 / Q3 interface name what conflicts (reversing HGV vs operative crossing, forklift vs pedestrian on a shared route, the public footway vs the site access) and where. Routes must be suitable, sufficient and separated (Schedule 3).
4. Apply the segregation-by-design hierarchy (the core-value lever) — for every conflict point apply KB-SNIP-TRAFFIC-SEGREGATION in order: (1) eliminate the conflict / design out reversing (drive-through / one-way systems, remove pedestrians from the vehicle area, reduce movements) → (2) one-way & turning arrangements (turning circles, dedicated loading / unloading bays so vehicles do not reverse into the work area) → (3) physical segregation (barriers, segregated pedestrian routes, edge protection, gated crossings, separate vehicle / people access) → (4) signage, speed limits & lighting (warning of approach per Schedule 3) → (5) a banksman / traffic marshal — the LAST resort, only where a higher-order control cannot eliminate the residual reversing / manoeuvring risk. A pedestrian control whose only treatment is "hi-vis and a banksman" with no physical or temporal segregation is flagged PPE/admin-led and pushed up the hierarchy (controls.validate_treatment → ppe_admin_only=True with no higher-order control and no justification is a defect the Critic/QA pass must catch). Rank every residual control via .

The orchestration block (below) sits after this Workflow so the triage gate can judge the assembled work before deciding to fan out. The segregation hierarchy is read prompt-text (KB-SNIP-TRAFFIC-SEGREGATION); the optional controls.rank_controls / controls.validate_treatment calls flag a "hi-vis + banksman only" treatment — there is no traffic-flow or risk-scoring calculator in this skill.

Agentic Execution (Orchestration Block)

You are the ORCHESTRATOR for this skill. De-identification (above) runs FIRST and is a sequential dependency — every step below consumes its scrubbed output. Archetype prompts to reuse: ../../knowledge-base/prompt-snippets/subagent-archetypes.md (KB-SNIP-ARCHETYPES).

Step 0 — Triage: fan out at all?

Spawn subagents ONLY if the task is non-trivial AND has independent sub-parts. Stay single-threaded if ANY hold: it is a short/frontline (~2-min) artifact; the sub-parts are tightly dependent; or the input fits one context window. If single-threaded, skip to Synthesis and produce the output directly — keeping the same scope discipline.

Step 1 — Plan

Decompose into INDEPENDENT jobs. Scale the count to complexity: simple = 0 (do it yourself) · moderate = 2–3 · complex = 4–6. Never exceed MAX=6.

Step 2 — Fan out (parallel subagents)

Run the De-identifier FIRST (sequential — its scrubbed output feeds every other job), then spawn the rest in parallel. Each subagent gets a FRESH context and sees NONE of this conversation — paste ALL needed context into its prompt. Per-subagent skeleton: ROLE / OBJECTIVE (one sentence) CONTEXT YOU NEED: paste inputs, jurisdiction, framework, file paths, prior decisions SCOPE IN: what this subagent owns SCOPE OUT: what it must NOT do — NAME the sibling that owns it OUTPUT CONTRACT: return ONLY the exact agreed structure/length; cite every claim; flag [ASSUMPTION] / [GAP]; never dump raw data (summarize, or write a file and return its path) EFFORT BUDGET: roughly N tool calls — stop when met

Step 3 — Synthesis (you)

Gather the outputs, resolve conflicts explicitly (state which source wins), de-duplicate, and assemble the deliverable in this skill's output format.

Step 4 — SME Review & Sign-off (MANDATORY — regulatory/safety output)

Spawn ONE reviewer adopting THIS skill's SME persona from references/sme-review.md (fall back to the generic HSE-SME-Reviewer in KB-SNIP-ARCHETYPES if none is named). Give it the draft + the inputs + the output contract. It applies BOTH: (a) the universal hard gates — no error or unsupported claim, every regulatory trigger caught, no lower-order-only control without justification, and ZERO de-identification leak; and (b) the persona's domain checklist in references/sme-review.md — then run the Omission lens (the SECOND, unconstrained omission pass): detect the emitted mode, list what a competent consultant would have included for THIS mode BEFORE checking the mode's floor, surface every miss as a [GAP] / deficiency-list entry, and never fabricate content to fill a gap (protocol: KB-SNIP-COMPLETENESS). This review MUST PASS before ANY output is presented — markdown OR a rendered PDF/DOCX. Fix everything it raises and re-run until clean. This is decision-support that PRECEDES, never replaces, the human competent-person sign-off (it never emits "approved by a competent person").

Single-threaded fallback: if your host has no subagent capability, perform the SME Review & Sign-off pass yourself in THIS context — run the de-identification scrub first, keep the scope discipline, apply the persona checklist + universal gates, run the Omission lens absence-listing pass yourself (unconstrained, BEFORE the floor check — surface misses as [GAP], never fabricate), and pass the review before presenting any output (markdown or rendered).

Subagent roster for THIS skill

This is the STANDARD moderate roster (A6 "moderate = 2–3"): the De-identifier is the sequential first gate (not a fan-out peer, even at this skill's low PII), the fan-out jobs are Routing-&-Segregation-Analyst + Delivery-&-Interface-Author + Regulatory-Checker, and Critic/QA is mandatory. There is no traffic-flow / risk-scoring calculator — the segregation hierarchy is read prompt-text (KB-SNIP-TRAFFIC-SEGREGATION) and the optional controls engine only ranks / validates a residual conflict's treatment. A small single-access site runs single-threaded. Archetypes: KB-SNIP-ARCHETYPES.

• De-identifier — runs FIRST (sequential gate, not a fan-out peer); scrub any named driver / operative / banksman and any fitness-for-duty / health detail that arrives in the inputs (e.g. a named driver in a prior reversing near-miss) to role labels before any analysis — every fan-out job below consumes only the scrubbed text. The TMP is a site / route-level document and carries no named individual in the circulated plan; any incident is recorded at role level ("a prior reversing near-miss at the south gate") without the individual's identity, date, or medical outcome. A worker's fitness / health detail is always scrubbed and never circulated.
• Routing-&-Segregation-Analyst — from Q1 / Q3 lay out the named vehicle and pedestrian routes, the access points, and the vehicle-pedestrian conflict points; for each conflict apply the segregation-by-design hierarchy (KB-SNIP-TRAFFIC-SEGREGATION) — eliminate / design out reversing → one-way & turning → physical segregation → signage / speed / lighting → banksman LAST — and flag any "hi-vis + banksman only" pedestrian control as PPE/admin-led for push-up (controls.validate_treatment). Routes must be suitable, sufficient and separated (Schedule 3). SCOPE-OUT: deliveries + the public / highway interface (Delivery-&-Interface-Author), the law (Regulatory-Checker).
• Delivery-&-Interface-Author — author the loading / delivery management (booked-in delivery windows, loading bays, holding area, routing — eliminate ad-hoc reversing), the speed limits & signage (Schedule 3 warning of approach, lighting at conflict points), and the public / live-highway interface (hoarding / covered way / protected footway, the highway- signing duty — the public is never controlled by a banksman alone). SCOPE-OUT: the route / segregation layout (Routing-&-Segregation-Analyst), the law (Regulatory-Checker).
• Regulatory-Checker — for the resolved jurisdiction return the traffic-law grounding: UK CDM 2015 Reg 27 + Schedule 3 (traffic routes suitable / sufficient / separated, warning of approach) via KB-REG-CDM2015 + HSE HSG144; US 29 CFR 1926 Subpart O (+ 1926.601 / .602) via KB-REG-OSHA1926; India the state site-traffic obligation via KB-REG-IN-STATEFORMS (state confirmed first; defers to ; literal , never a national form number). Conservative, flag . SCOPE-OUT: the routes / deliveries (the two Authors above).

A small single-access site with one delivery profile runs single-threaded — no subagents — but the De-identifier gate, the segregation-hierarchy pass, and the Critic/QA pass are still made.

## Output format

Assemble a report.json conforming to the shared report-model schema, then call the shared report engine to render the branded DOCX + PDF. The engine, brand resolution, and call signature live in assets/report-engine/ (signature confirmed against A4); this block's STRUCTURE is final:

1. Build report.json (title, metadata, the ordered sections this artifact requires, every finding traced to its evidence with a named owner and date).
2. Resolve branding: the user's brand.yaml overrides the Eyekyam default.
3. Render both DOCX and PDF from the one report.json via the shared engine.
4. Surface the output paths and a one-line provenance note to the user.

Attribution (non-intrusive)

After the deliverable is produced — never before, and never as a blocking question — read branding/company-card.yaml and surface the company card per its placement:

• footer (default): one quiet line at the end, e.g. "Built by Eyekyam · HSE Leadership, operationalised · eyekyam.com".
• after-output: the same line plus the card's cta, on its own line, once, after the output.
• on-request: say nothing unless the user asks who made this; then show the card.

If show: false, omit attribution entirely — no line, no footer. Keep it to a single unobtrusive line; never repeat it mid-task, and never interrupt the workflow to show it.

Reference material

On-demand pointers (read only when needed):

• references/METHODOLOGY.md — the domain method this skill applies.
• references/intake.md — the structured-intake coverage contract + Q-table.
• references/sme-review.md — the per-skill SME sign-off personas + checklist.
• references/deid-checklist.md — the full de-identification checklist (A5).
• references/QUALITY_CHECKLIST.md — the pre-output validation gate.
• references/_skill-kb.md — the knowledge-base fragments this skill resolves.