Designs a non-punitive behaviour-based-safety (BBS) program for a named site, crew, or operation — ABC (antecedent-behaviour-consequence) analysis, an observable observation-card design, defined metrics (percent-safe, participation, trend-by-category), and at-risk behaviours trended to systemic causes and routed to hierarchy-ranked system fixes. Use this skill whenever a user asks to design, set up, or improve a behaviour-based safety / BBS / behavioural-safety / safe-behaviour-observation program, an observation card, peer-observation or safety-conversation scheme, or to define BBS metrics and feedback loops. Observation cards are non-punitive by design — role-labelled or anonymous, voluntary, used for trending and learning, never individual discipline; observable site-specific behaviours only (never 'work safely'); at-risk behaviours route to system fixes via the controls engine (hierarchy of controls), never to 'retrain the worker'. Decision-support only; a competent person must review the output.
Instrucciones de origen · Vista previa de solo lectura
name
bbs-program-designer
description
Designs a non-punitive behaviour-based-safety (BBS) program for a named site, crew, or operation — ABC (antecedent-behaviour-consequence) analysis, an observable observation-card design, defined metrics (percent-safe, participation, trend-by-category), and at-risk behaviours trended to systemic causes and routed to hierarchy-ranked system fixes. Use this skill whenever a user asks to design, set up, or improve a behaviour-based safety / BBS / behavioural-safety / safe-behaviour-observation program, an observation card, peer-observation or safety-conversation scheme, or to define BBS metrics and feedback loops. Observation cards are non-punitive by design — role-labelled or anonymous, voluntary, used for trending and learning, never individual discipline; observable site-specific behaviours only (never 'work safely'); at-risk behaviours route to system fixes via the controls engine (hierarchy of controls), never to 'retrain the worker'. Decision-support only; a competent person must review the output.
A consultant-grade HSE skill that designs a non-punitive behaviour-based-safety
(BBS) program for a named site, crew, or operation. It forces the single lever
that separates a defensible program from generic safety paperwork: observable,
site-specific behaviours (never "work safely") and the full hierarchy of
controls — at-risk behaviours are trended to causes and routed to
hierarchy-ranked via the deterministic A7 engine, never
to "retrain the worker" or discipline. The program is grounded in (), uses
(percent-safe, participation, trend-by-category from
, with its mandatory small-cell guardrail), and is
: observation cards are ,
, and used for . A
card that records a nameable individual for discipline is a
. It maps to (worker participation) via the
shared . Decision-support only; a competent person
must review the output.
## Output format
systemic
system fixes
controls
ABC
(antecedent–behaviour–consequence) analysis
KB-SNIP-BBS-METHOD
defined metrics
KB-DATA-BBS-METRICS
<5
non-punitive by design
role-labelled or anonymous
voluntary
trending and learning, never individual sanction
de_identification
hard-fail
ISO 45001 clause 5.4
KB-SNIP-LEADERSHIP-CLAUSE-MAP
When to use this skill
Use this skill when the user needs to design, set up, or improve a behaviour-based
safety (BBS) / behavioural-safety program for a concrete site, crew, or operation —
for example "design a BBS observation program for the warehouse picking crew at the
Leeds DC", "set up a peer-observation / safe-behaviour-observation scheme and its
metrics", or "our BBS cards have stalled — redesign the card and the feedback loop so
at-risk behaviours actually drive fixes". Trigger phrases: behaviour-based safety,
BBS, behavioural safety, safe-behaviour observation, observation card, peer
observation, safety-conversation programme, ABC analysis, percent-safe / participation
metrics. This skill designs the program (the card, the ABC routing rule, the
metrics, the feedback loop) — it does not run a full safety-culture maturity
assessment (see safety-culture-assessment), facilitate leadership safety
walks / gemba (see safety-walk-gemba), or build the leading/lagging KPI
framework (see leading-lagging-kpi-framework); intake Q1 disambiguates against
those siblings. If the request is vague, the Workflow intake below forces the
specific site/crew and observable behaviours before any drafting — it refuses a
non-observable card item such as "work safely".
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.
This skill is method-led, not law-led: BBS is a participation method (ISO 45001
clause 5.4), so it always grounds in the BBS method + metrics + the bundle clause
map, then resolves the jurisdiction only for the worker-consultation legal duty.
../../knowledge-base/regulatory/in-factories-act.md (safety-committee worker participation; mandatory state detection for any form, defers to hse-india, never a national form number)
Unknown
Ask before citing any specific law — the BBS method (ABC, non-punitive cards, system-fix routing) still applies
Always apply KB-SNIP-HOC to every at-risk-behaviour treatment: system fixes
(eliminate / substitute / engineer / administrate) rank ABOVE "retrain the worker",
and a discipline-only response is rejected. For any metric, resolve the ID in
KB-DATA-BBS-METRICS and honour its <5 small-cell suppression — a percent-safe for a
4-person crew is suppressed. 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, the mandatory India→state branch, 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, and refuse to
proceed on a vague request. The two hard refuse anchors:
Q1 (program scope + sibling disambiguation) — a named site / crew /
operation to design the BBS program for. Refuse "improve our safety culture"
in the abstract; if the user actually wants a culture-maturity assessment, a
leadership gemba walk, or a KPI framework, route to the sibling
(safety-culture-assessment / safety-walk-gemba /
leading-lagging-kpi-framework) instead of building a BBS card.
Q3 (observation-card behaviours) — the card items must be observable and
site-specific (tied to a named task/area). Refuse a vague, non-observable item
such as "work safely" / "be careful" — record [GAP] and elicit the observable
behaviour; never publish a card item that cannot be observed and counted.
De-identify the inputs FIRST — any prior observation cards, incident notes, or
participation logs are scrubbed to role/group labels before any analysis (the
deid block above + references/deid-checklist.md + the De-identifier-runs-first
orchestration rule). No nameable individual is ever recorded on a card for
discipline — observation data is role-labelled or anonymous. Everything
downstream consumes only the scrubbed text.
ABC analysis (KB-SNIP-BBS-METHOD) — for each target behaviour, map the
antecedent (the real trigger, not a slogan), the observable behaviour, and
the consequence, and design the system so the safe behaviour carries the
soon / certain / positive consequence.
Design the observation card (non-punitive, observable) — each item is
observable and site-specific, tied to a named task/area; the card is
role-labelled or anonymous, voluntary, and its data is used for trending and
learning, never individual sanction. A non-observable item ("work safely") is
refused at this step.
Define the metrics (KB-DATA-BBS-METRICS) — percent-safe = (safe ÷ total) ×
100; participation = (active ÷ trained pool) × 100; trend by behaviour category,
never by person. Apply the <5 small-cell suppression to any team breakdown
(a 4-person crew's percent-safe is suppressed; secondary suppression applied).
Route at-risk behaviours to SYSTEM fixes (the hierarchy lever) — an at-risk
behaviour is a signal of a system gap. Apply KB-SNIP-HOC and call
controls.rank_controls + controls.validate_treatment to route it to a
hierarchy-ranked system fix (eliminate / substitute / engineer / administrate
before PPE). Never "retrain the worker", "be more careful", or discipline — a
treatment whose only response to a trended at-risk category is retraining or
sanction is a defect the Critic/QA and SME passes must catch (the hard
enforcement of the core value).
Trend at-risk categories to systemic causes — aggregate cards by behaviour
category, identify the recurring systemic driver, and feed it to owned/dated
actions (smart_actions.validate_register: a named role-label owner + an ISO due
date + a measure). No anonymous actions, no "ASAP".
Close the observer-feedback loop — two-way, immediate feedback (positive AND
at-risk discussed on the spot); learning fed back to observers and the workforce so
the program produces systemic improvement, not a tally.
Validate against references/QUALITY_CHECKLIST.md — observable site-specific
behaviours only; cards non-punitive / role-labelled / voluntary; no <5 cell
published; at-risk behaviours routed to system fixes (no retrain-the-worker-only
treatment); metrics defined; every action owned + dated; ISO 45001 clause 5.4 cited
accurately; de-id applied.
Assemble the branded report — build report.json (see assets/report.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 deterministic control-ranking
step (5 via controls) and metric computation (4 via KB-DATA-BBS-METRICS) are
A7/script-grounded in every case — never a fan-out judgement call.
Scope note — this skill designs the BBS program; it does not run a
safety-culture maturity assessment (safety-culture-assessment), facilitate a
leadership gemba walk (safety-walk-gemba), or own the leading/lagging KPI
framework (leading-lagging-kpi-framework). It supplies the BBS percent-safe
leading indicator to that KPI framework, but does not build it here.
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 moderate roster (A6 "moderate = 2–3 fan-out") with the De-identifier
as the sequential first gate — here it is critical, because observation cards and
participation data are exactly the place a nameable individual can be re-identified and
punished. Control-ranking is a deterministic A7 controls call (Workflow step 5)
and metric computation is grounded in KB-DATA-BBS-METRICS (step 4) — never an LLM
fan-out job. Archetypes: KB-SNIP-ARCHETYPES.
De-identifier — runs FIRST (sequential gate, not a fan-out peer). Before any
analysis: scrub every nameable individual on any input card / log to a role/group
label; ensure no card records a named individual for discipline; apply the
<5 small-cell suppression + secondary suppression to any team breakdown. Return
the re-identification key SEPARATELY to the orchestrator — never to a sibling, never
in the document. Every job below consumes only this scrubbed output.
SCOPE-OUT: does not design the card (Program-Designer) or rank the fixes (the A7
controls call).
Program-Designer — build the ABC map, the non-punitive observation card
(observable, site-specific items only), the defined metrics (percent-safe /
participation / trend-by-category from KB-DATA-BBS-METRICS), and the
observer-feedback loop. Refuse any non-observable item ("work safely").
SCOPE-OUT: the deterministic control-ranking of at-risk behaviours (the A7 controls
call), the SMART action register (Action-Planner).
Action-Planner — turn each trended at-risk category into a system fix
routed through the controls engine (hierarchy-ranked, never "retrain the worker"),
then into a SMART action with a named role-label owner + an ISO due date + a
measure (smart_actions.validate_register). SCOPE-OUT: designing the card
(Program-Designer), de-identification (De-identifier).
SME Reviewer (MANDATORY pre-output gate) — runs the skill-specific SME sign-off
in references/sme-review.md (Behavioural Safety Consultant) before any
output: cards non-punitive and observable, at-risk behaviours routed to system fixes
(no retrain-the-worker-only treatment), metrics defined with <5 suppression, no
individual named on a card.
Critic/QA (MANDATORY) — adversarial final pass: every card item observable and
site-specific, cards role-labelled/voluntary, at-risk behaviours routed to
hierarchy-ranked system fixes (no retrain-the-worker-only treatment), metrics defined
and <5-suppressed, every action owned + dated, ISO 45001 clause 5.4 cited
accurately, and ZERO de-identification leak (no named-individual discipline card,
no <5 cell, no re-id key in the output). PASS/FAIL.
A single crew with one card set may run single-threaded — no subagents — but the
De-identifier scrub + <5 suppression, the A7 controls ranking call, and the
Critic/QA pass are still made.
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.
