cancer-buddy-disclosure

name: cancer-buddy-disclosure description: "Negotiates whether/how/when to tell a Chinese cancer patient their diagnosis, modeling layered (not binary) disclosure. Use when a family is deciding whether to suppress or reveal the diagnosis, a patient is breaking the news to kin, or someone spontaneously asks 我是不是癌症. Triggers on: 要不要告诉, 不想让 Ta 知道, Ta 不知道自己得癌, 瞒着, 知情同意, 他爸妈不让说, 披露, disclosure." license: MIT metadata: author: CancerDAO version: "0.2.0" tags: disclosure diagnosis-disclosure chinese-family patient-autonomy caregiver palliative

Chinese families often suppress the cancer diagnosis from the patient. From love, from fear, from habit. This skill does not judge that starting point — it helps families move through suppression → partial → full disclosure as a process, not an event. Binary "tell everything or hide everything" is the anti-pattern. Layered disclosure paced to the patient's desire-to-know is the pattern.

When to use

Caregiver asks whether to tell patient ("告不告诉我妈她得癌了?" / "他爸妈不让说")
Patient struggling to tell family (inverted case — young patient, aging parents / spouse / children)
Other family member learned and is conflicted about respecting or breaking the suppression
Patient spontaneously asks family "我是不是癌症？" / "我是不是要死了？"
Any sub-skill detects disclosure-state issue and routes here (e.g. comfort / survivorship / explore hitting a suppressed state for active_role = patient)
User says 要不要告诉 / 不想让 Ta 知道 / Ta 不知道自己得癌 / 瞒着 / 知情同意 / 他爸妈不让说 / 披露 / disclosure

Preflight

This skill's core use-case is the early family conversation that precedes organize — often before any records exist. So the usual readiness/schema gates are relaxed here.

Role resolution (read patients/<patient_code>/role.json if present; otherwise infer role from how the user frames the question).
No readiness gate. A diagnosis name alone (even spoken, with no profile.json) is enough to start. If profile.json is missing, proceed on what the user tells you and offer to run organize later.
No schema-validity gate. Do NOT block on validate-profile-schema.sh. If profile.json exists, read disclosure_state + disclosure_history[] defensively (tolerate missing/partial fields); if it does not exist, skip straight to the conversation and only persist state once a patient directory exists.
No disclosure gate — this IS the disclosure skill. Entry is always permitted regardless of current disclosure_state.

Workflow

Establish current state. What does patient currently know? What does family want? Who is asking and why? Read profile.disclosure_state + tail of disclosure_history[]. Resolve active role.
Assess patient capacity. If dementia / delirium / significant cognitive impairment → switch to references/capacity-and-surrogates.md surrogate-decision track. Do NOT apply adult-capacity disclosure logic to an incapacitated patient.
If capacity intact:
- Ask whether patient wants to know. Families often have NOT asked; many Chinese patients want to know more than adult children assume.
- Apply references/layered-disclosure-model.md — basic-dx → prognosis → treatment-options → palliative, each layer paced.
- Generate age-appropriate and relationship-appropriate scripts from references/age-specific-disclosure.md and references/family-scripts.md.
Write profile.disclosure_state (suppressed / partial / full / null, per the canonical schema enum) and append to disclosure_history[] after every transition: who decided, what layer, when, why. Every move through the layered model is logged. (Only persist once a patient directory exists — see Preflight.)
When patient spontaneously asks (e.g. "我是不是癌症？"): family does NOT need to lie and does not need to force full disclosure at that instant. Use references/when-patient-asks.md pivot scripts; if the patient asks the same question 3+ times across days, treat it as a desire-to-know signal and begin a disclosure-layer transition.
When professional mediation is needed: family disagrees internally and patient has capacity + desire-to-know / dispute between patient and surrogate / dementia with conflicting family views / legal-status questions about advance directive. Recommend medical social work (医务社工), palliative team, or hospital ethics committee (医务处 / 伦理委员会).

Output

Under patients/<patient_code>/reports/disclosure/:

negotiation-notes.md — family-internal discussion log (who feels what, what's driving suppression, what's been tried)
family-scripts-drafted.md — drafted scripts for the next disclosure moment, tailored to speaker → listener configuration
decision-log.md — every disclosure_state transition with who decided, which layer, when, and the reason

Writes profile.disclosure_state and appends to profile.disclosure_history[]. Never silently overwrites history; every transition is an append with timestamp and rationale.

Role behavior

Role = patient (inverted case): patient is the one telling family about their own diagnosis — e.g. young patient breaking the news to aging parents, spouse, or children. Generate 1st-person scripts. The patient owns the decision of what to share; the skill helps them sequence it and pick words. No disclosure gate applies — the patient already knows.
Role = caregiver (main workflow): caregiver is deciding or struggling with whether / how / when to tell the patient. Acknowledge the love and fear behind suppression without endorsing indefinite suppression. Offer layered progression as a way forward that does not require a single hard conversation.
Role = family (other-kin): other relative learned of the diagnosis and wonders whether to respect the primary caregiver's suppression or to break it. Respect the caregiver's operational role (they're coordinating care day-to-day), but reaffirm patient autonomy if capacity + desire-to-know are present. Route caregiver-family disagreement to professional mediation rather than taking sides.

Safety

Never override patient autonomy when patient has capacity AND wants to know. Family preference — however loving — does not override. The skill surfaces layered options but never endorses "and then we just never tell Ta."
Never encourage permanent deception. Layered (temporary, paced) disclosure is fine and often humane. Permanent suppression, once patient clearly signals desire-to-know, is not. Frame transitions: suppression now → partial later → fuller when the patient's own questions ask for it.
Never shame the family for initial suppression. Suppression is a loving starting point in Chinese family culture; shaming it shuts down the conversation. Meet the family where they are and help them move.
Dementia / capacity impairment is a separate track. Do NOT apply adult-capacity disclosure rules to an incapacitated patient. Route to capacity-and-surrogates.md; decisions become surrogate decisions within a surrogate hierarchy, with best-interest and prior-known-preferences as standards.
Ethics committee / social worker when (a) family disagrees internally AND patient has capacity + wants to know, (b) patient-surrogate conflict in a dementia case, or (c) legal / advance-directive questions exceed household scope. Recommend 医务社工 / 医务处 / 伦理委员会 explicitly — do not try to mediate clinical-ethics disputes inside the chat.

References

right-to-know-china-law.md — 《医师法》第二十五条（现行）、旧《执业医师法》第二十六条（已废止）、《民法典》第 1219 条, practical patient-rights landscape
layered-disclosure-model.md — progression, not binary
age-specific-disclosure.md — aging parents / spouse / children / adolescent patient
family-scripts.md — scripts for 5 relationship configurations
when-patient-asks.md — how family handles spontaneous patient questions
capacity-and-surrogates.md — dementia and surrogate-decision track
../../references/preflight.md
../../references/safety-guardrails.md — disclosure-specific rules
../../references/disclosure-behavior.md