Produces a site-specific health-and-safety induction pack for a named site, project, or contractor cohort, with a competence-verification record and refresher-tracking schedule. Use this skill whenever a user asks to build a site induction, new-starter or contractor induction, orientation pack, or onboarding safety briefing for a specific location. It grounds the induction in the site's real hazards, emergency arrangements, and rules over the mandatory induction baseline (KB-SNIP-INDUCTION-BASELINE); produces a delivery pack plus a signed competence-verification record proving each inductee understood the content; and schedules refreshers by role-risk — emitted as a branded report. A generic induction with no named site or site-specific hazards is refused. Decision-support only; a competent person must review the output.
Produces a site-specific health-and-safety induction pack for a named site, project, or contractor cohort, with a competence-verification record and refresher-tracking schedule. Use this skill whenever a user asks to build a site induction, new-starter or contractor induction, orientation pack, or onboarding safety briefing for a specific location. It grounds the induction in the site's real hazards, emergency arrangements, and rules over the mandatory induction baseline (KB-SNIP-INDUCTION-BASELINE); produces a delivery pack plus a signed competence-verification record proving each inductee understood the content; and schedules refreshers by role-risk — emitted as a branded report. A generic induction with no named site or site-specific hazards is refused. Decision-support only; a competent person must review the output.
A consultant-grade HSE skill that produces a site-specific health-and-safety induction pack
for a named site, project, or contractor cohort — grounded in the site's real hazards,
emergency arrangements, and rules over the mandatory induction baseline
(KB-SNIP-INDUCTION-BASELINE), framed by ISO 45001:2018 clauses 7.2 (competence) & 7.3
(awareness). It layers the named site's specifics onto the baseline (emergency · welfare ·
rules · site-specific hazards · reporting), produces a (per
inductee, role-labelled in any widely distributed copy, the verification level set on
) proving each inductee understood the content, and schedules
as dated SMART actions. It forces the single lever that separates a
defensible artifact from copy-paste paperwork: — never a generic, template-only induction, never a PPE-only treatment.
## Output format
competence-verification record
KB-DATA-COMPETENCE-LEVELS
refreshers by role-risk
every induction topic tied to a named site
arrangement, the real site hazards ranked up the full hierarchy of controls, and a verification
record always present
A generic induction with no named site or site-specific hazards is refused.
When to use this skill
Use this skill when the user needs an induction pack for a concrete, named site, project, or
contractor cohort — for example "build a site induction for the Tilbury distribution
warehouse", "I need a contractor induction for the crews mobilising to Bravo oil terminal",
"write a new-starter orientation pack for our Pune plant", or "set up the visitor safety
briefing and sign-off for our chemicals site". Trigger phrases: site induction, new-starter /
contractor / visitor induction, orientation pack, onboarding safety briefing, induction
verification / sign-off record, refresher schedule. It produces a delivery pack + a signed
competence-verification record + a refresher schedule — distinct from sop-writer (a
procedure) and toolbox-talk (a single pre-task briefing). For the contractor cohort it
pulls site-rules + permit-to-work awareness and follows contractor prequalification
(contractor-prequalification, #16). If the request is vague ("write me an induction"), the
Workflow intake below refuses to produce a pack until the named site + at least the
emergency arrangements + at least one site-specific hazard are captured — a generic induction
is explicitly refused.
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.
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.
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.
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.
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.
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 selector (Q5) → read the matching legal-induction-baseline fragment:
Jurisdiction
Read
India
../../knowledge-base/regulatory/in-factories-act.md (Factories Act 1948 s.111A worker right to information; + in-state-forms.md for the user's state — defers to hse-india, no national form number minted)
UK
../../knowledge-base/regulatory/uk-hswa.md (MHSWR 1999 reg. 10 information for employees + reg. 13 capabilities & training)
Always — for every run — also read the shared induction baseline + competence + method snippets
and apply them over the named site's specifics:
../../knowledge-base/prompt-snippets/induction-baseline.md (KB-SNIP-INDUCTION-BASELINE) —
the mandatory induction-topic baseline (emergency · welfare · site rules · site-specific
hazards · incident/concern reporting) the named site's specifics are layered onto; a generic
induction with no named site/hazards is refused; every induction produces a
competence-verification record.
../../knowledge-base/data-points/competence-levels.md (KB-DATA-COMPETENCE-LEVELS) — the
shared 4-level competence scale (aware / trained / competent / expert) with the evidence
test per level; sets the verification level each inductee/role must reach. Quote its
source+year.
../../knowledge-base/prompt-snippets/ops-clause-map.md (KB-SNIP-OPS-CLAUSE-MAP) — the
bundle clause cross-walk (induction = ISO 45001 7.3 awareness); routes a user who asks for the
wrong artifact for a clause to the owning sibling (e.g. competence-matrix → training-needs-analysis).
../../knowledge-base/prompt-snippets/hierarchy-of-controls.md (KB-SNIP-HOC) — applied to
every site-specific hazard's control in the induction (never a PPE-only line) and to the
refresher schedule's remediation.
This skill always grounds in KB-STD-ISO45001 (cl. 7.2/7.3) + KB-SNIP-INDUCTION-BASELINE +
KB-DATA-COMPETENCE-LEVELS, and reads ONE jurisdiction fragment for the legal-induction baseline
(Q5). 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 audience · Q2 named site · Q3 site-specific hazards & arrangements available ·
Q4 delivery mode · Q5 jurisdiction · Q6 verification method · roles/owners · refresher
cadence), the contractor-cohort branch (Q1 = contractors → pull site-rules + PTW awareness,
follow contractor-prequalification #16), 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 drafting, and refuse to produce a pack until the named
site (Q2) + at least the emergency arrangements + at least one site-specific hazard (Q3) are
captured — record [ASSUMPTION] / [GAP], never emit a generic induction. A generic
induction with no named site or site-specific hazards is explicitly refused.
The site-specific induction-design method (ISO 45001 7.2/7.3)
Full method in references/METHODOLOGY.md; the topic baseline is KB-SNIP-INDUCTION-BASELINE
and the verification scale is KB-DATA-COMPETENCE-LEVELS. Steps:
De-identify the inputs — before any drafting (the deid block above + De-identifier-first
orchestration rule). Inductee names belong only on the legitimate signed
competence-verification record; any widely distributed copy of the pack uses role
labels, and no inductee health/medical detail enters the pack. The de-id block warns
before any name enters a shared artifact.
Resolve the audience + the named site — from Q1/Q2; refuse a generic site. The contractor
cohort (Q1) branches to pull site-rules + permit-to-work awareness and follows
prequalification (contractor-prequalification).
Lay the mandatory baseline — apply KB-SNIP-INDUCTION-BASELINE: every induction covers
emergency arrangements · welfare & first aid · site rules · site-specific hazards &
controls · incident/concern reporting. The baseline is the floor, never the whole pack.
Layer the named site's specifics (the core lever) — for every baseline topic, tie it to
this site's real arrangement (the named muster point, the actual traffic plan, the real
permit systems) and each site-specific hazard to its control ranked via KB-SNIP-HOC
(Elimination → Substitution → Engineering → Administrative → PPE) — never a generic line,
never PPE-only without justification. A topic with no named site arrangement is a [GAP].
Build the competence-verification record — per inductee (role-labelled in any shared
copy), set the verification level required for the role on KB-DATA-COMPETENCE-LEVELS
(aware / trained / competent / expert) and the verification method from Q6 (quiz / supervised
sign-off / competence demonstration). An induction with no verification record fails the
quality gate.
Schedule refreshers by role-risk — turn the refresher cadence into SMART actions via
smart_actions.validate_register — each with a named owner + ISO due date (higher-risk
roles refresh more often). No anonymous actions, no "ASAP".
Validate against references/QUALITY_CHECKLIST.md — no generic-only content; every topic
tied to a named site arrangement; the verification record present; de-id applied; every legal
citation traced to the KB.
Assemble the branded report — build report.json (see
assets/induction-pack-report.template.json) and run the canonical report-output call below.
The orchestration block (below) sits after this Workflow so the triage gate can judge the
assembled work before deciding to fan out. The induction is content assembly, not a
calculation — smart_actions (step 6, the refresher schedule) is the only A7 script call.
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.
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 single-thread-by-default roster: a short induction-assembly artifact passes the
Step-0 triage as low-complexity and runs single-threaded — no subagents — but the de-id
scrub, the KB-SNIP-INDUCTION-BASELINE baseline, the KB-DATA-COMPETENCE-LEVELS verification
level, and the SME + Critic/QA passes are still made. It fans out to 2 only for a large
multi-audience pack (e.g. permanent + contractor + visitor variants of one site), never
exceeding MAX=6. Archetypes: KB-SNIP-ARCHETYPES.
Single-threaded by design — for a single-audience induction the orchestrator does it
itself: de-identify first, lay the baseline, layer the named site's specifics, build the
verification record and refresher schedule.
De-identifier (fan-out only) — runs FIRST (sequential gate, not a fan-out peer); scrub
every inductee name, contact, and any health/medical detail to role labels before any
assembly. Inductee names survive only on the legitimate signed verification record; every
widely distributed copy is role-labelled. The Content-Assembler consumes scrubbed text only.
Content-Assembler (fan-out only) — assemble the per-audience induction pack: the
KB-SNIP-INDUCTION-BASELINE baseline layered with this named site's real hazards
(controls ranked via KB-SNIP-HOC), the competence-verification record (level on
KB-DATA-COMPETENCE-LEVELS), and the refresher schedule (smart_actions). SCOPE-OUT: de-id
(the De-identifier owns it), SME sign-off (the SME Reviewer owns it).
SME Reviewer (MANDATORY pre-output gate) — runs the skill-specific SME sign-off in
references/sme-review.md (Site HSE Manager) before any output: no generic-only
content, every topic tied to a named site arrangement, the verification record present, no
inductee health detail or names in a widely distributed copy.
Critic/QA (MANDATORY) — adversarial final pass: no generic-only induction, every
site-specific hazard's control ranked (no PPE/admin-only without justification), the
competence-verification record present, every legal-induction citation traced to the KB, zero
PII/health leak (no inductee name/medical detail in a shared copy, no re-id key in the
output). PASS/FAIL.
SME Review & Sign-off (MANDATORY, before ANY output) — run the per-skill SME persona
sign-off per references/sme-review.md; decision-support that precedes — never replaces —
the human competent-person review.
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:
Build report.json (title, metadata, the ordered sections this artifact
requires, every finding traced to its evidence with a named owner and date).
Resolve branding: the user's brand.yaml overrides the Eyekyam default.
Render both DOCX and PDF from the one report.json via the shared engine.
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.
