Built against the INACSL Healthcare Simulation Standards of Best Practice and SAMHSA’s trauma-informed principles — enforced in code and checked on every build, not asserted in marketing. Where a standard governs organizations rather than software, we say so and produce the evidence instead. SIMCareReady holds no certification or accreditation from any body named here and does not claim one.
16 met · 3 supported
The platform does this, it is enforced rather than encouraged, and a check proves it on every build. If the behavior regresses the build fails and the claim comes down with it.
COA, CARF and the Joint Commission accredit organizations and programs, not software — and neither this platform nor any other can “meet” them. What a platform can do is produce the evidence an organization puts in front of a surveyor. Saying that plainly is more credible than a claim the buyer already knows to be impossible.
INACSL Healthcare Simulation Standards of Best Practice®
The standard asks for: Simulation-based experiences begin with measurable objectives aligned to the learner’s needs and the organization’s expectations.
What we do: A scenario cannot be released without a competency matrix naming every objective it measures, and every objective must map to a rubric competency that the scorecard actually reports.
Implemented in src/certification.ts, src/rubric.ts · publisher →
The standard asks for: Purposeful design, with fidelity, complexity and modality chosen to serve the objectives rather than the technology.
What we do: Difficulty changes the young person’s behavior rather than his starting numbers. It changes how much ground a correct response earns and how fast a mistake is read as evidence. Starting him angrier is a shorter exercise, not a harder one.
Implemented in src/persona.ts · publisher →
The standard asks for: Learners are prepared and briefed before the experience, including a fiction contract and the psychological safety statement.
What we do: Every session opens with the briefing, the explicit statement that the young person is fictional, and the statement that this is workforce training and not therapy. It cannot be skipped — the session does not start until it is dismissed.
Implemented in app/sim/page.tsx · publisher →
The standard asks for: Facilitation methods are matched to the learner’s level, and psychological safety is maintained throughout.
What we do: The behavioral gauges are hidden from the learner by default. A learner who can see the numbers is playing against a gauge. The observer panel is for a facilitator, and the interface says so when it is on.
Implemented in app/sim/page.tsx · publisher →
The standard asks for: A planned debriefing follows every simulation-based experience — feedback, debriefing and/or guided reflection.
What we do: A session cannot end without one. The scorecard names the turning point and quotes the deciding moment with its timestamp. It gives one change to practice and offers the same scenario one difficulty higher.
Implemented in src/score.ts, app/sim/page.tsx · publisher →
The standard asks for: Evaluation uses tools with evidence of validity and reliability, against criteria set in advance.
What we do: Every score must cite a timestamp and quote the learner’s own words. The schema makes both mandatory, so a judgment nobody can point at cannot be returned at all.
Implemented in src/score.ts · publisher →
The standard asks for: Confidentiality and integrity in performance and in the reporting of results.
What we do: A scorecard states on its face whose rubric produced it. Where no organization framework is loaded it says so in the interface and in the saved record, so a placeholder result can never be mistaken for an organizational one.
Implemented in src/rubric.ts, src/score.ts · publisher →
SAMHSA
The standard asks for: Physical and psychological safety for everyone in the environment.
What we do: Every scenario carries hard ceilings on agitation and perceived threat. Whatever the model proposes, a de-escalation exercise cannot be driven into a crisis the scenario did not authorize.
Implemented in src/state.ts · publisher →
The standard asks for: Choice and shared decision-making rather than control.
What we do: Compliance cannot rise while perceived threat is at or above 70. It is structurally impossible for this simulation to teach a staff member that frightening a young person works, because fear-driven compliance is not modeled as a success.
Implemented in src/state.ts · publisher →
The standard asks for: Operations and decisions conducted with transparency, with the goal of building trust.
What we do: Trust rises at most 8 points a turn and falls at most 30. A promise a staff member cannot keep costs more than an honest no, and the rubric scores that explicitly.
Implemented in src/state.ts, src/rubric.ts · publisher →
The standard asks for: Services that are developmentally appropriate, youth-guided and culturally responsive.
What we do: The behavioral model is built on developmental stage instead of adult conversation. Adolescent escalation outruns de-escalation and working memory narrows under arousal. At fifteen autonomy outranks comfort, and shame turns outward. Register and slang density track the young person’s state and age, and are specified per persona.
Implemented in src/persona.ts · publisher →
The standard asks for: De-escalation and stabilisation in the least restrictive environment.
What we do: The scenario’s terminal condition is settled-enough-to-move-on, never compliance. A session that ends with the young person shut down is reported as a shutdown, not as a success.
Implemented in src/state.ts · publisher →
KnightByrd standard of excellence
The standard asks for: No language model may write the behavioral state of a simulated young person. It may only propose a change, which a separate rules layer applies, constrains or refuses — and every refusal is recorded and inspectable.
What we do: applyDelta() is the only thing that writes state. It caps per-turn movement, enforces scenario ceilings and applies coupling rules, and it returns every override alongside the result so a facilitator can see what was refused and why.
Implemented in src/state.ts, src/engine.ts
The standard asks for: Personas are composites written to exercise a behavioral model. There is no real record or real child’s likeness. No diagnosis is inferred about any real person, and nothing refers to anything outside the scenario.
What we do: Stated in the persona file, enforced in the system prompt, declared in the briefing before a session starts and on the badge throughout it.
Implemented in src/persona.ts, src/engine.ts, app/sim/page.tsx
The standard asks for: The exercise may be as difficult as the work. The feedback must address the response and never the person, and must always name the next thing to practice.
What we do: Scores are 0-4 against named behavioral anchors with the deciding moment quoted, and every debrief ends with one specific change and an invitation to run it again harder.
Implemented in src/rubric.ts, src/score.ts
The standard asks for: The platform never gives clinical or emergency advice, never assesses a real person, and says so wherever a user can see it.
What we do: Declared in the system prompt, in the briefing, in the terminal header and on the public pages.
Implemented in src/engine.ts, src/cli.ts, app/sim/page.tsx
COA Accreditation / Social Current
The standard asks for: Personnel competence is established and maintained through initial and ongoing training, coaching and supportive supervision.
What we do: The record shows demonstrated competency instead of completion. It lists the attempt, the competency measured and the deficiency named. It also gives the one change to practice and the same scenario at a higher difficulty. A pass at difficulty 2 and a failure at difficulty 4 is reported as exactly that.
Implemented in src/score.ts, src/persona.ts · publisher →
CARF International
The standard asks for: Documented, competency-based staff development that a survey can examine.
What we do: Every session is written out in full as a dated record an organization can put in front of a surveyor. It includes each exchange, the proposed and applied state changes, every constraint that fired and the scorecard. CARF accredits programs and not vendors. This produces evidence for an organization’s own file.
Implemented in src/cli.ts · publisher →
The Joint Commission
The standard asks for: Evidence supporting staff competency, training and quality processes.
What we do: The same dated session record serves this. The Joint Commission accredits organizations and not software. The platform’s role is to produce evidence the organization uses in its own processes.
Implemented in src/cli.ts · publisher →
Scenario release gate
A clinical subject-matter expert, a simulation expert and a safety reviewer — each by name and date, not by job title. Software does not approve its own clinical content here.
peer-conflict-dayroom v0.1 declares 6 objectives, every one mapped to a competency the scorecard reports, and claims 16 standards. It carries 0 of 3 required approvals, so the gate refuses it.
That is not an oversight we forgot to tidy up. It is the honest state of a proof of concept with no clinical reviewer yet, and it is exactly what we are asking a design partner for. Filling those three signatures in ourselves to make a demo look finished would be the single most damaging thing this codebase could contain.
Every measurable, and the check behind it
The model proposes and a separate rules layer decides. Every proposal it makes is capped, floored or refused before it touches the simulation, and every refusal is recorded where a facilitator can read it. Nothing about how a young person behaves is left to a model’s mood.
Proven by standards-check: an absurd proposal is constrained rather than obeyed
Compliance cannot rise while perceived threat is at or above 70. It is structurally impossible for this platform to teach a staff member that fear works — not discouraged in a prompt, impossible in the code.
Proven by standards-check: compliance cannot rise while perceived threat is 70 or above
Every scenario carries a hard ceiling on agitation and perceived threat. A de-escalation exercise cannot become a restraint scenario because a model decided to push.
Proven by standards-check: agitation cannot exceed the scenario ceiling whatever the model proposes
A supervisor coaching from this can point at the exact second. A judgment nobody can point at cannot be returned at all — the schema refuses it.
Proven by standards-check: a score cannot be returned without a timestamp or without the quote
No objective exists that nothing measures. If a scenario says it teaches something, the debrief scores it — or the scenario does not release.
Proven by certify(): an objective whose competency is not in the loaded rubric refuses release
A clinical subject-matter expert, a simulation expert and a safety reviewer — each by name and date, not by job title. Software does not approve its own clinical content here. Until all three sign, the scenario is refused.
Proven by certify(): a missing or unnamed approval is a refusal, not a warning
It names the turning point and quotes the deciding moment. It gives one change specific enough to rehearse and offers the same scenario one difficulty higher. A session cannot end without it.
Proven by standards-check: the debrief names one specific change to practice
A learner who can see the numbers plays against the numbers. The observer panel is for a facilitator, it is off by default, and the interface says so whenever it is on.
Proven by standards-check: the behavioral gauges are hidden from the learner by default
Every persona is a composite written to exercise a behavioral model. There is no record, likeness or diagnosis of any real person, and no reference to anything outside the scenario. Stated in the prompt, in the briefing and on screen throughout.
Proven by standards-check: the prompt states the young person is fictional and forbids outside reference
A guard scans our own public copy and fails the build on any wording that implies one of these bodies has certified, accredited, approved or endorsed this software. It also fails on any claim that a study has proven it works. Those bodies accredit organizations and programs, never software, and the audience for this product knows exactly what those words cost to earn.
Proven by standards-check: false certification, accreditation and efficacy claims are all rejected
Every exchange, the proposed state change, the applied state change, every constraint that fired, and the scorecard. That is what an organization puts in front of a surveyor — and it is what turns thousands of sessions into workforce readiness data nobody currently has.
Proven by every session writes sessions/<timestamp>-<scenario>-d<difficulty>.json
When the exercise gets harder or easier, the learner is told which rule fired and the three responses it read, in words assembled from the rule itself. An organization documenting workforce competency cannot defend “the AI decided you were ready.” Generated prose also ends up describing things the system did not do, which is worse than no explanation at all.
Proven by standards-check: every decision in every mode at every difficulty carries an explanation, and src/adaptive.ts is grepped for any model call
Verification requires three certification sessions at 75% or above, across at least two scenarios and two developmental bands. There must be no critical misses and a demonstrated recovery, which means falling below standard and getting back to it in the same conversation. Highest demonstrated and verified are kept as separate facts, because conflating them is how a training record becomes something nobody can rely on.
Proven by standards-check: one strong session is never verified, and recovery is required