Create and update a practical brief to help patients and caregivers advocate for themselves.
fuck-cancer.mp4
Since I published my mom's story, I’ve heard from many people supporting a loved one through cancer.
The last thing a patient or caregiver needs is a pile of confusing medical terms, hype, and scattered information. Instead, they need clear answers to three questions: What should we do next? What do we know? How can we advocate for ourselves?
Here's what patients and caregivers have told me:
"You have to be a huge patient advocate. Push for answers. Push for biopsies. Push for proper testing. Ask questions."
"I need help navigating difficult conversations and advocating for myself."
"The volume of doctors, documents, and insurance paperwork can quickly become overwhelming."
This skill creates a concise, fact-based brief with 5 sections:
- Patient information. Essential identifiers, care team contacts, and more for easy reference during phone calls.
- What to do next. Up to 3 specific actions naming the next steps.
- What we know. The shortest useful summary of what's confirmed vs. what's still ambiguous. Can also include research on clinical trials.
- Medical terms. Plain-English explanations of medical terms and acronyms.
- Care log. A compact history of recent updates.
It creates this brief based on documents and context that you share with it and trusted medical sources like NCI's PDQ summaries and the ClinicalTrials.gov API.
On first use, the skill creates the brief in a Google Doc you can share, or a local Markdown file if you ask for one. You can run the skill with more context to update the brief over time.
The easiest way to install the skill is to paste this into ChatGPT, Claude Code, or your favorite agent:
Install the /fuck-cancer skill globally from https://github.com/petergyang/fuck-cancer
You can also install it with npx:
npx skills add petergyang/fuck-cancer --skill fuck-cancer --global --yesStart with a brain dump:
/fuck-cancer My dad was just diagnosed. Here’s what I know.
Paste or upload a report:
/fuck-cancer Explain this report and help me prepare for doctor meeting.
Research trials or second opinions:
/fuck-cancer Find relevant trials near Toronto for this diagnosis.
Share new information as it arrives:
/fuck-cancer Update our brief with this new biomarker report.
The skill grounds its research in trusted medical sources:
- Evidence and treatment context: NCI's PDQ summaries or the official cancer agency for the patient's country or province.
- Drug approvals and labels: The patient's national regulator, such as the FDA, Health Canada, EMA, MHRA, or TGA.
- Testing and clinical guidance: Current official publications from groups such as ASCO, ESMO, CAP, or NICE when they directly apply.
- Clinical trials: The official ClinicalTrials.gov API, including the status of the individual site.
- Medical definitions: The NCI Dictionary of Cancer Terms or an equivalent official national source.
- Emerging questions: Peer-reviewed primary research indexed in PubMed, with early or indirect evidence labeled as such.
- Specialists and services: Official academic cancer-center pages for their own clinicians, programs, and trials.
The skill supports decisions with current evidence. It does not diagnose cancer, choose treatment, determine trial eligibility, or replace the patient's medical team.
SKILL.mdcontains the complete care-navigation and research workflow.search_trials.pysearches ClinicalTrials.gov API v2, filters sites by distance from home with--near LAT,LON, labels each site's recruitment status, and ranks likely matches first.eval.mdchecks the destination, brief structure, action specificity, medical boundaries, and concise care-log format.test_search_trials.pycovers location normalization, distance filtering, site-status handling, pagination, criteria previews, and relevance ordering.openai.yamlcontains the Codex skill metadata.
Google Doc mode, the default, uses the user's connected Google Drive integration; if it is not connected, the skill asks the user to enable it before writing. Local Markdown mode requires no account, database, or API key. The trial-search helper requires no API key, uses only Python's standard library, and requires Python 3.8 or newer.
MIT