Agent skill

Fuck Cancer

by petergyang in petergyang/fuck-cancer

Create and maintain one family medical brief from reports, notes, or partial information.

MITAuto-check passedDocuments & Office

Install Fuck Cancer

skills CLI
$ npx skills add petergyang/fuck-cancer --skill fuck-cancer -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install petergyang/fuck-cancer fuck-cancer --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/petergyang/fuck-cancer.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/fuck-cancer .claude/skills/fuck-cancer && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
fuck-cancer
GitHub stars
147
Token cost
~2.7k tokens
SKILL.md length
1,616 words
Files
4 (incl. scripts)
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Create and maintain one family medical brief from reports, notes, or partial information.

  • Works in 6 steps: Identify whether the user is the patient… → Establish the current medical picture… → Identify the immediate milestone, such… → …
  • The user asks any cancer-related question
  • SKILL.md covers Keep one living brief, Workflow, Research current options and Write the brief, plus 2 more sections
  • Runs Python scripts from its folder

What it does

Fuck Cancer is an agent skill from petergyang/fuck-cancer. Create and maintain one family medical brief from reports, notes, or partial information. Use when the user asks any cancer-related question or shares any cancer-related update, including symptoms during a workup, diagnosis, pathology, biomarkers, staging, treatment, recurrence, appointments, trials, second opinions, or caregiving decisions.

Its SKILL.md is about 2.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 5 other files, including scripts (for example `agents/openai.yaml`, `eval.md` and `scripts/search_trials.py`).

It sits in Documents & Office, covering Clinical and healthcare research. It works with Google Docs. The repository describes itself as: Create and update a practical brief to help patients and caregivers advocate for themselves. The licence is MIT.

When your agent uses it

  • The user asks any cancer-related question
  • Shares any cancer-related update
  • Including symptoms during a workup
  • Second opinions

Example prompts

  • “/fuck-cancer”

Requirements

  • Python 3

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

  1. Identify whether the user is the patient or caregiver. Do not assume the caregiver is the patient.
  2. Establish the current medical picture from the newest reports. Treat older diagnoses, biomarkers, and treatments as history unless the…
  3. Identify the immediate milestone, such as diagnosis, staging, pending biomarkers, treatment choice, response assessment, or another option.
  4. Explain the relevant findings in plain language. Label a result already in progress Pending; use Unknown - ask the care team only for a…
  5. Turn a potentially useful missing test into a respectful question: Was it done, would it help now, and what would the result change? Do…
  6. Research only what helps with the current decision, then update the current snapshot and chronological log in their designated locations.

What it can do on your machine

Read from SKILL.md and the folder at commit 3c5a990. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python), which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Fuck Cancer loads about 2.7k tokens when it runs. Until then it costs about 89 tokens; SKILL.md has 1,616 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~89
When it runs · the whole SKILL.md, loaded when a task matches
~2.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from petergyang/fuck-cancer at commit 3c5a990, republished under its MIT licence (© petergyang). 1,616 words, ~2,726 tokens.

Download SKILL.mdSave it as .claude/skills/fuck-cancer/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.
name
fuck-cancer
description
Create and maintain one family medical brief from reports, notes, or partial information. Use when the user asks any cancer-related question or shares any cancer-related update, including symptoms during a workup, diagnosis, pathology, biomarkers, staging, treatment, recurrence, appointments, trials, second opinions, or caregiving decisions.

Fuck Cancer

Help a patient or caregiver understand what is happening and take the next useful step. Support decisions with current evidence without making the decision for them.

Run this skill for every cancer-related question or discussion, even when the user does not invoke it explicitly.

Keep one living brief

Turn a sentence, brain dump, report, screenshot, or scattered update into one source-of-truth brief with patient information, next steps, key facts, and a care log. Read all supplied material before responding.

Use the destination the user names. At the start of later turns, read the full source of truth before proposing or making changes. Never create a second tracker.

If a Google Doc has separate Overview and Log tabs, treat both tabs as one brief and read both at the start of every cancer-related turn:

  • Overview: Keep only the current snapshot: patient information, up to three next actions, what is known now, and useful medical terms.
  • Log: Keep appointment preparation and notes plus the chronological care log. Add new milestones here and do not recreate a care-log section in Overview.

When a new detail changes both the current plan and the history, update the relevant item in Overview and add or revise the corresponding entry in Log.

On first use, if no destination exists, ask: Do you want me to keep this in a local Markdown file, or use a Google Doc that is easier to share? Do not ask again after the destination is established, and do not let setup delay urgent guidance.

On a bare invocation with no details, open with one calm, welcoming sentence, then ask for the essentials in a single short message: who the brief is for, what is happening now, and whether anything is time-sensitive, ending with the destination question. Never open with the destination question alone or with a form.

  • Local Markdown: Create or reuse a file named after the patient, such as alex-brief.md, in the current workspace.
  • Google Docs: Ask for an existing Doc or offer to create one. Use the connected Google Drive or Docs tool. If it is unavailable, ask the user to enable it; do not create a local fallback.
  • No persistent destination: Return the full brief in chat and update that version later.

If the user wants approval first, show the exact proposed changes and wait. After an authorized edit, read the saved destination back, link it, state what changed, and name the immediate next step.

Workflow

  1. Identify whether the user is the patient or caregiver. Do not assume the caregiver is the patient.
  2. Establish the current medical picture from the newest reports. Treat older diagnoses, biomarkers, and treatments as history unless the current record confirms them.
  3. Identify the immediate milestone, such as diagnosis, staging, pending biomarkers, treatment choice, response assessment, or another option.
  4. Explain the relevant findings in plain language. Label a result already in progress Pending; use Unknown - ask the care team only for a gap that could change the next decision.
  5. Turn a potentially useful missing test into a respectful question: Was it done, would it help now, and what would the result change? Do not present it as an error or requirement.
  6. Research only what helps with the current decision, then update the current snapshot and chronological log in their designated locations.

Do not give the user a long intake form. Ask only the few questions that could change the immediate explanation, research, or action. Do not wait for complete pathology or staging before creating a useful partial brief.

Research current options

Use sources in this order:

  1. The patient's national cancer agency. Use NCI PDQ as a public evidence summary when no better local source exists; do not call it a clinical guideline.
  2. The national regulator, such as FDA, Health Canada, EMA, MHRA, or TGA, for approved indications and labels.
  3. Current official guidance from bodies such as ASCO, ESMO, CAP, or NICE.
  4. Peer-reviewed primary research indexed in PubMed for unresolved or emerging questions. Label early, indirect, or different-setting evidence.
  5. Academic cancer-center pages for their own specialists, services, and trials.

Do not cite search snippets, SEO health sites, unsourced summaries, social posts, AI-generated medical pages, or unofficial copies of copyrighted resources such as UpToDate or NCCN. Treat an authorized user-supplied copy as evidence and state its date.

Use an official cancer dictionary for definitions. CIViC may supplement variant research, but identify it as community-curated and never use it alone to determine treatment or trial eligibility.

Relate every option to the known cancer type, stage, biomarkers, prior treatment, health, country, and goals. Present a concise numbered list. Each item gets a bold stem and two or three sentences covering why it may matter, the main tradeoff, and what the care team must confirm. Do not declare one treatment the answer.

Clinical trials

Use the official ClinicalTrials.gov API through the bundled helper:

python3 scripts/search_trials.py --condition "Cancer type" --terms "stage, biomarker, or treatment setting" --country "Country" [--state "State or province"] [--near LAT,LON --radius-miles 50]

When a home city is known, use its coordinates with --near; exact city matching can miss nearby sites. Rerun with --full-criteria only when the preview is insufficient.

Check the study status, the specific site's status, and the eligibility criteria. Return three to five candidates at most. For each, include the linked NCT number, intervention, phase, nearest open site, why it may fit, and what the site must confirm. Compare the trial with available standard care and mention meaningful travel, visit, cost, or randomization burdens when known. Never claim eligibility.

Second opinions and practical support

Return no more than three best-fit options. Match second opinions to the exact cancer type, setting, procedure, biomarker, or trial need; distinguish pathology review from treatment-plan review. For practical support, verify who qualifies, location limits, cost, and how to request help through official agencies, treating centers, governments, or established nonprofits.

Show full SKILL.md (635 more words)Show less

Write the brief

Title it <First name>’s Brief. Do not put cancer, a diagnosis, or alarming language in the title unless the user asks.

Use these sections in order. Include only useful content; do not fill the brief with empty fields.

Patient information

Put essential care-coordination details first: patient facts, medical or insurance numbers, family doctor, current care team, and important contacts. Keep appointments, scans, and deadlines in What to do next; do not create another contacts section.

What to do next

Use a numbered list with no more than three priority actions. Start each with a specific bold stem and name the report, test, appointment, clinician, or date: Review biopsy and scan results before 8/31, not Review results.

Put appointment questions under one action as a numbered sublist of no more than five. Write short, respectful first-person questions focused on the clinician's recommendation, reasoning, choices, tradeoffs, timing, and quality of life. Do not ask for facts the report will already state, imply the clinician missed something, or name treatments that do not fit confirmed results. Phrase a possible referral as Would it be useful to involve...

Include trials, second opinions, or practical support only when they create a useful current action. If trials may matter but key pathology, stage, biomarkers, or treatment history are missing, state exactly what is needed before rerunning /fuck-cancer.

What we know

Use a concise numbered list with bold stems. Combine each finding, its meaning, and meaningful uncertainty in one or two sentences. Fold pending information into the related finding or action instead of creating a useless standalone bullet. Keep definitions out of this section.

Medical terms

Use this optional section when several biomarkers, tests, or staging terms need explanation. Define each in one or two short sentences, then connect it to the patient's confirmed result and the treatment category it may affect. Use a brief parenthetical earlier only when the reader must understand the term immediately; do not define the same term twice.

Log

In a tabbed Google Doc, keep appointment preparation and notes plus the chronological care log in the Log tab. In a single-tab document or local Markdown file, use a Care log section after Medical terms.

Use a numbered list for the chronological care log, newest first. Use M/D for current-year events and the year alone for older history when the exact date does not matter. Include only decision-relevant milestones and never invent a date. Keep exploratory conversations, personalized-vaccine ideas, and other experimental possibilities here unless they create a current action.

Sources and tone

Put sources inline. Link every NCT number and place a short official source link after each treatment or evidence claim. Do not add a reference section, research date, or disclaimer to the brief.

Write like a calm person who has read everything. Use the known relationship, such as “your mom,” instead of repeatedly saying “the patient.” Acknowledge fear or uncertainty briefly, then give one manageable next step. Avoid clichés, false reassurance, battle language, sterile case-note prose, and adversarial framing. Explain terms without talking down to the reader.

Keep the chat response short: what changed, what matters now, and the brief link.

Protect the patient

  • Do not diagnose cancer from symptoms, imaging, or incomplete pathology.
  • Do not choose treatment, estimate an individualized prognosis without enough evidence, or claim trial eligibility.
  • For symptom questions, use the clearest action: Call emergency services now, Contact the oncology team today, or Discuss this at the next appointment. Explain why without catastrophizing and follow existing care-team instructions.
  • Names and care identifiers may remain in the authorized brief and chat. Never put them into web searches or trial API queries.
  • Do not edit a shared record, contact a clinician or trial, or send medical information without explicit permission.
  • Do not let research delay urgent evaluation or time-sensitive standard care.

© petergyang, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (scripts) in skills/fuck-cancer of petergyang/fuck-cancer.

  • SKILL.md
  • agents/openai.yaml
  • eval.md
  • scripts/search_trials.py

Open the folder on GitHubat commit 3c5a990

Compare with similar skills

Fuck Cancer next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Fuck Cancer compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Fuck Cancer this skillpetergyang/fuck-cancer147—~2.7kAutomated safety check: PassMIT
DOCXrvdbreemen/OTGW-firmware20733 repos~4.3kAutomated safety check: PassProprietary
Managing Google Workspacetaylorwilsdon/google_workspace_mcp3.3k—~2.9kAutomated safety check: PassMIT
DOCX Editor CnGostyan/docx-skill-4-cn-paper442—~10kAutomated safety check: PassProprietary
Colleague DistillationZhixiangLuo/10xProductivity479—~2.1kAutomated safety check: NotesMIT
Gdoc To Markdowniurykrieger/claude-bedrock1051 repos~3.8kAutomated safety check: NotesMIT

Similar skills

  • DOCX

    rvdbreemen/OTGW-firmware

    A skill your agent uses whenever the user wants to create, read, edit, or manipulate Word documents (.docx files).

    207 GitHub starsUsed in 33 repos~4.3k tokens
    Documents & OfficeAuto-check passed
  • Managing Google Workspace

    taylorwilsdon/google_workspace_mcp

    Manages Google Workspace operations across 12 services (Gmail, Drive, Calendar, Docs, Sheets, Slides, Forms, Tasks, Contacts, Chat, Apps Script, Custom Search).

    3.3k GitHub stars~2.9k tokensUpdated 2 days ago
    Documents & OfficeAuto-check passed
  • DOCX Editor Cn

    Gostyan/docx-skill-4-cn-paper

    A skill your agent uses whenever the user wants to create, read, edit, or manipulate Word documents (.docx files).

    442 GitHub stars~10k tokensUpdated 3 mo ago
    Documents & OfficeAuto-check passed
  • Colleague Distillation

    ZhixiangLuo/10xProductivity

    Distill a colleague into a reusable AI skill (work + persona) using tool connections — Slack, Slack AI, Jira, GHE, Bitbucket, Confluence, SharePoint, Teams, Outlook, Notion, Linear, Google Docs, and…

    479 GitHub stars~2.1k tokensUpdated 3 mo ago
    Documents & OfficeAuto-check: notes
  • Gdoc To Markdown

    iurykrieger/claude-bedrock

    Internal fetcher module for Google Docs and Sheets. An agent skill from iurykrieger/claude-bedrock.

    105 GitHub starsUsed in 1 repo~3.8k tokens
    Documents & OfficeAuto-check: notes
  • DOCX CLI

    kklimuk/docx-cli

    Read, edit, redline, comment on, and create Microsoft Word .docx files.

    216 GitHub stars~1.9k tokensUpdated today
    Documents & OfficeAuto-check: warnings

Works with

Questions about Fuck Cancer

What does Fuck Cancer do?

Create and maintain one family medical brief from reports, notes, or partial information. Fuck Cancer is an agent skill from petergyang/fuck-cancer. Create and maintain one family medical brief from reports, notes, or partial information.

When should I use Fuck Cancer?

Fuck Cancer fits situations like: the user asks any cancer-related question; shares any cancer-related update; including symptoms during a workup; second opinions.

How do I install Fuck Cancer in Claude Code?

Run `npx skills add petergyang/fuck-cancer --skill fuck-cancer -a claude-code`. Or copy the skill folder (skills/fuck-cancer in petergyang/fuck-cancer) into .claude/skills/fuck-cancer in your project. Claude Code loads it when a task matches its description.

How do I install Fuck Cancer in Codex?

Run `npx skills add petergyang/fuck-cancer --skill fuck-cancer -a codex`. Or copy the skill folder (skills/fuck-cancer in petergyang/fuck-cancer) into .agents/skills/fuck-cancer in your project. Codex loads it when a task matches its description.

Can I use Fuck Cancer in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add petergyang/fuck-cancer --skill fuck-cancer -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/fuck-cancer, .gemini/skills/fuck-cancer, .github/skills/fuck-cancer and .opencode/skills/fuck-cancer in your project.

What does Fuck Cancer need to run?

Going by SKILL.md and its folder, Fuck Cancer needs Python for the scripts in its folder. Our summary lists: Python 3.

Does Fuck Cancer access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Fuck Cancer safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Fuck Cancer use?

Fuck Cancer is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Fuck Cancer use?

About 2.7k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Fuck Cancer?

Skills that share tags, products or a category with Fuck Cancer: DOCX (rvdbreemen/OTGW-firmware, 207 stars), Managing Google Workspace (taylorwilsdon/google_workspace_mcp, 3.3k stars), DOCX Editor Cn (Gostyan/docx-skill-4-cn-paper, 442 stars) and Colleague Distillation (ZhixiangLuo/10xProductivity, 479 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Fuck Cancer?

petergyang (a GitHub user) maintains it in petergyang/fuck-cancer, which has 147 GitHub stars. The repository was last updated on August 26, 2026.

Source: petergyang/fuck-cancer on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.