radiology-grant
Develop/review radiology grants against a live call, including current NSFC rules, code/call routing, funder criteria and imaging feasibility; not papers or decks. CN: 标书、国自然、基金评审
Install / Use
npx skills add huang-sir1/radiology-skills --skill radiology-grantInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
Customer SupportSupported Platforms
Our assessment of radiology-grant
radiology-grant scores 85/100 on our quality scale, 234th of 319 Customer Support skills we index.
Its SKILL.md is 13 KB long, split into 7 sections and no code examples: a thorough specification that gives an agent plenty to work with.
With 1,785 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so radiology-grant is actively maintained.
- Our last check on 2026-11-04 found the source still online.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 88/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-02. Automated pattern scan on 2026-10-02. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
radiology-grant compared with similar skills
All 4 of these similar skills score higher than radiology-grant; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| radiology-grant (this skill)by huang-sir1 | 85 | 1.8k | 56d ago | SKILL.md |
| taste-skillby Leonxlnx | 100 | 116.8k | 8d ago | CLAUDE.md |
| Claude-Code-Game-Studiosby Donchitos | 100 | 39.7k | 7d ago | CLAUDE.md |
| x402by browser-use | 100 | 121.0k | 3d ago | SKILL.md |
| algorithmic-artby anthropics | 100 | 177.9k | 55d ago | SKILL.md |
Frequently asked questions
- How do I install radiology-grant?
- Run
npx skills add huang-sir1/radiology-skills --skill radiology-grant. The install tabs above show the steps for each supported agent. - Which AI agents does radiology-grant work with?
- It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is radiology-grant safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 88/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is radiology-grant still maintained?
- The repository was last updated about 2 months ago, so radiology-grant is actively maintained.
Skill content
View source on GitHubname: radiology-grant description: "Develop/review radiology grants against a live call, including current NSFC rules, code/call routing, funder criteria and imaging feasibility; not papers or decks. CN: 标书、国自然、基金评审"
Radiology Grant Review and Development
Use this skill for funding proposals, not for manuscript peer review or presentation production. Treat a grant as a call-bound argument for a consequential, answerable question and a feasible research programme. A polished narrative cannot cure an ineligible application, an unsupported premise, an underpowered design, or an uncontrolled imaging measurement chain.
Non-negotiable boundaries
- Lock the live call before compliance advice. Record the funder, mechanism, cycle, programme,
official sources and retrieval date in a grant call passport.
If those cannot be verified, emit
CALL_VERSION_UNRESOLVED, restrict the review to scientific content, and do not claim submission readiness. - For NSFC, lock six routing fields. Record application year, programme/type and subtype,
science division/office, year-bound application code candidate, research attribute and any
special call/clinical track. Read
nsfc-2024-2026-change-map.md; current official-source
conflicts become
CONFLICTING / ADMIN_CONDITIONAL, never an inferred eligibility answer. - Separate two gates. Administrative acceptance (eligibility, limits, forms, attachments, budget regime, ethics/security) is not scientific merit. Passing either gate does not imply passing the other.
- Use the target funder's current criteria. Do not translate every call into a generic NSFC, NIH or numeric rubric. Do not invent scores, weights, deadlines, page limits or eligibility.
- Evidence before rhetoric. Preliminary findings must point to real source artifacts. Keep verified facts, user attestations, calculations, hypotheses, planned work and missing evidence visibly distinct. Never turn planned work into preliminary data.
- Protect claim ceilings. Prediction is not measurement, mechanism, causality or clinical utility. A model-performance aim cannot silently license any of those stronger claims.
- Review is read-only by default. Critique and propose changes; modify the submitted file only when the user explicitly requests revision.
- NSFC generative-AI boundary. For a live NSFC application, read call-and-compliance-gate.md. The 2026 annual application and completion notice permits bounded research-tracking/reference assistance only with human verification, applicant responsibility, truthful declaration and marking; it prohibits directly generated applications and unverified content. Label AI-assisted text as advisory for applicant verification and rewriting; never represent it as applicant-authored or submission-ready.
Route by task
| User need | Mode and resources | Main output |
|---|---|---|
| Live NSFC rule, year comparison or submission-readiness question | read nsfc-2024-2026-change-map.md, call-and-compliance-gate.md and source-registry.md | six-field current-year lock, exact-source rule map and conflict states |
| NSFC title, abstract, 立项依据, 研究内容, 研究基础 or innovation development | read nsfc-application-content-playbook.md and grant-architecture.md | scientific-function map and author decision points, not legacy headings |
| NSFC imaging code, H18/H27/H28/H29 or clinical/special-call choice | read nsfc-radiology-code-and-call-router.md | year-bound candidate-code/call record and unresolved confirmation gate |
| Fast fatal-flaw or one-section check | triage/targeted; read review-modes-and-criteria.md | bounded findings, missing inputs and next gate |
| Whole application review | full audit; also read call-and-compliance-gate.md, radiology-feasibility-audit.md, and evidence-contract-and-stopping-rules.md | criterion-by-criterion report |
| Simulated panel or pre-submission challenge | mock panel; read review-modes-and-criteria.md | independent views, disagreement record, panel synthesis |
| Response to prior critiques / resubmission | revision/resubmission; read review-modes-and-criteria.md | critique-to-change matrix and residual risks |
| Final funder-system fields, files and institutional handoff | funder-package-finalization; read call-and-compliance-gate.md and use the submission contract in grant-call-passport.md | closed field/file/attachment inventory, unresolved blockers and human institutional action gate |
| Develop question, aims or architecture | read grant-architecture.md and reframe-and-innovation.md | proposal spine and author decision points |
| Feasibility, milestones, budget or contingency | read radiology-feasibility-audit.md and feasibility-and-pitfalls.md | aim/milestone/risk register |
| Choose a framework for imaging AI, diagnostic accuracy, prediction, radiomics, generative AI or quantitative imaging | read imaging-methodology-router.md | task-specific design-function map; no checklist-as-validity claim |
| NIH, ERC, Wellcome or another international route | read international-grants.md, then verify the named call live | funder-specific mapping; never generic substitution |
Do not load all references for a narrow request.
Funder-package finalization boundary
Use funder-package-finalization only for one exact funder + mechanism + cycle + programme/subcall +
applicant/host configuration. Recheck the live call, amendments, current electronic form/system and
institutional route. Freeze:
call-passport ID/version | proposal/source versions | field/section/attachment inventory | exact file locators and hashes where files exist | page/character/format checks | budget/ethics/security/AI-use handoffs | applicant/co-applicant attestations | institutional steps and deadlines | unresolved blockers | responsible human role.
This mode audits a package; it does not log in, paste fields, upload files, attest for an applicant, confirm institutional eligibility, authorize a budget or press a submission control. Scientific or budget content that changes during finalization returns to the grant/scientific owner and creates a new source version before the inventory is rechecked.
Use only these machine states:
FUNDER_PACKAGE_INCOMPLETE— required fields, files, current rules or source versions are missing;FUNDER_PACKAGE_BLOCKED— a known compliance, authority, confidentiality or unresolved-live-rule condition prevents institutional handoff; orHUMAN_INSTITUTIONAL_SUBMISSION_REQUIRED— the bounded machine audit is complete and every remaining portal, attestation, approval and submission action belongs to authorized humans.
HUMAN_INSTITUTIONAL_SUBMISSION_REQUIRED is the maximum machine state. Never report SUBMITTED,
ACCEPTED, ELIGIBLE_CONFIRMED, official receipt, funding probability or funder/institutional
approval without the corresponding authoritative external evidence.
Review workflow
- Intake and version lock. Identify the review mode and material version. Build the call passport. Record absent proposal sections instead of inferring them.
- Administrative gate. Check eligibility, application limits, required forms/attachments,
research category/code, budget route, ethics/security and AI-use rules only against current
official sources. For NSFC, freeze all six routing fields before any compliance verdict. Preserve
official-text conflicts instead of resolving them by convenience. Return
ADMIN_PASS,ADMIN_CONDITIONAL,ADMIN_FAIL, orADMIN_NOT_ASSESSABLEwith evidence. - Freeze the proposal spine. Extract clinical/scientific need → gap → central question → hypothesis → aims → expected inference/impact. Flag aims that do not serve the same question.
- Apply the funder criteria. Create one criterion finding per material issue: criterion, verdict, evidence anchor, consequence, actionable repair and residual uncertainty. Preserve the funder's language and scoring model.
- Run the radiology gate. Audit the clinical-phenomenon → image-phenotype → biological/clinical proposition bridge; intended use and unit; acquisition/reconstruction; multimodal intersection denominator; reference standard; annotation; data/model-lineage leakage; aim-level sample size; external-independence vector; statistics; transportability; mechanism evidence; clinical pathway; data/compute/ethics; and measurement-to-claim consistency.
- Test execution. Map every aim to personnel, data, dependencies, milestone, decision rule, fallback, time and budget. Use the aim/milestone/risk register.
- Synthesize without laundering disagreement. Prioritize fatal compliance defects and inference-breaking weaknesses before competitive or editorial improvements. In panel mode, retain minority concerns.
- Close with a decision boundary. State what was assessed, what remains unverified, which
claims are supportable, and whether the next state is
READY_FOR_AUTHOR_REVISION,READY_FOR_INSTITUTIONAL_CHECK,REVIEWABLE_BUT_NOT_READY, orSTOP_AND_REFRAME.
Output contract
For a full review use grant-review-report.md and include:
Review scope— mode, proposal version, sections actually read, target call and limitations.Call passport status— current official evidence and unresolved live rules.Administrative gate— separate from merit, with no implied funding judgment.Proposal spine— question, hypothesis, aims, claim ceiling and logic breaks.Criterion findings—PASS / CONDITIONAL / FAIL / NOT_ASSESSABLE, each evidence-anchored.Radiology feasibility gate— measurement, data, reference standard, analysis and translation.NSFC code/call and methodology route, when applicable — year-bound candidate code, clinical or special-call boundary, selected design frameworks and explicit non-applicability.Aim–milestone–risk–budget alignment— go/no-go logic and scientifically meaningful fallback.Prioritized revision plan— P0 compliance, P1 inference-threatening, P2 competitive, P3 editorial.Evidence and uncertainty ledger— verified, attested, planned, inferred, missing and stale items.Readiness boundary— never a guarantee of eligibility, score, review outcome or funding.
Handoffs
- Study design and validation →
radiology-design; acquisition/reconstruction/QC →radiology-acquisition-qc; ground truth →radiology-annotation; data →radiology-data. - Sample size and inference →
radiology-stats/radiology-method-evaluation; clinical reader-study design →radiology-translation(stats supplies MRMC analysis and sizing); ethics and governance →radiology-ethics; clinical pathway →radiology-clinical-domain. - Live call, deadline, eligibility or frontier verification →
radiology-search/radiology-frontier; project-level provenance →radiology-pipeline.
Truncated for display — read the full file on GitHub.
Related Skills
taste-skill
116.8kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Claude-Code-Game-Studios
39.7kTurn Claude Code into a full game dev studio — 49 AI agents, 72 workflow skills, and a complete coordination system mirroring real studio hierarchy.
x402
121.0kSet up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key.
algorithmic-art
177.9kCreating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
