Radiology Research Ops logo

Radiology Research Ops

CommunityPopular
huang-sir1
radiology-research-ops

Plan/audit sites, RACI, milestones, dependencies, resources, risks and closeout; not scientific design.

Overview

Publisherhuang-sir1
Repositoryradiology-skills
Skill nameradiology-research-ops
Stars
2K
Forks
20
Bundled files
9
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 9 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by huang-sir1 on GitHub. Read the source before you install it.

Installation

Install the Radiology Research Ops AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/huang-sir1/radiology-skills/tree/main/radiology-skills/skills/radiology-research-ops
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/huang-sir1/radiology-skills.git /tmp/radiology-skills
mkdir -p .claude/skills
cp -r /tmp/radiology-skills/radiology-skills/skills/radiology-research-ops .claude/skills/radiology-research-ops
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Radiology Research Ops in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Radiology Research Ops on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Radiology Research Ops is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Radiology Research Operations

Use this skill to convert an approved or otherwise frozen scientific plan into a traceable operating system. It owns RACI, dependencies, site readiness, throughput, resource capacity, operational registers, freezes, pause/termination and closeout. It does not decide the scientific question, approve research, execute institutional management authority or claim that a plan was actually run.

Operating modes

ModeUse whenPrimary result
startupa protocol or project is preparing to openoperating charter, RACI, critical path and readiness gates
active-controlrecruitment, acquisition, annotation, assays or analysis are under waystatus dashboard, variance, forecast, actions and evidence
recoverya milestone, site, resource, quality or dependency is failingroot constraint, recovery options, decision deadline and claim consequence
freeze-releasea cohort, labels, analysis set, model, evidence or manuscript package must be frozen/releasedversioned freeze criteria, authorization, manifest and exception ledger
post-award-funder-reportingan awarded project has conditions, amendments, annual/interim/final reports or sponsor deliverablesaward-term passport, obligation/evidence crosswalk, reporting package state and authorized funder-action handoff
pause-stop-closeoutthe project pauses, terminates or completessafety/access handoff, preservation, reconciliation, closure and reusable lessons

For the control model and registers read references/operations-control-model.md. For multicentre and modality-specific readiness read references/site-readiness-and-imaging-operations.md. For risks, changes, freezes and closeout read references/risk-change-freeze-and-closeout.md. Consult references/source-registry.md before policy-facing advice; clinical-trial sources are quality frameworks only when the project is outside their actual scope.

Input passport

Record:

project/study ID | operating mode | scientific protocol/SAP/design digest and owner | study family and intended claim | sites/modalities/work packages | current approvals/contracts/access evidence | start/end and decision milestones | staffing/equipment/assays/compute/budget constraints | critical-to- quality factors | data/annotation/acquisition/analysis freeze states | current risks/issues/deviations | local accountable roles | artifact locations | authority for any update or external action.

Scientific design, ethics, contracts, budget approval, site activation and institutional assignments are author-only or institution-only facts. Use NOT_EVIDENCED until the relevant record is inspected. Advisory work is read-only; do not open a site, alter a schedule, assign personnel, spend funds, contact collaborators or change a protocol without explicit authority.

Operational evidence states

Use PLANNED, EVIDENCED_READY, ACTIVE, AT_RISK, BLOCKED, FROZEN, PAUSED, TERMINATED or CLOSED. A calendar date, verbal assurance, percentage-complete estimate or green dashboard does not establish readiness. Every state cites a record, owner and checked date.

Post-award funder-reporting boundary

For post-award-funder-reporting, bind one exact award and reporting period:

funder/mechanism/cycle/award ID | notice and award-term versions | institutional implementation route | reporting period/deadline/time zone | approved aims/protocol/SAP/budget/timeline versions | conditions and prior amendments | required fields/files/attachments | actual milestone/deliverable evidence and exact locators | deviations/issues/risks | ethics/data/sharing/publication states | scientific owner approvals | institutional/funder action owner.

Use the current award notice, terms, amendment history, reporting instructions and institutional research-office route. A prior application, generic funder guide, planned milestone or percent-complete estimate is not evidence that a deliverable was achieved or accepted.

  • Annual/interim/final reports: map every funder field and deliverable to a source artifact, reporting period, responsible owner and unresolved fact. Research operations assembles and tracks the package; the original scientific owner verifies methods, analyses, results and claims.
  • Amendments: keep a proposed change, authorized amendment and actual protocol/operational deviation distinct. Never backdate approval or use a report narrative to normalize an unauthorized change. Propagate an approved change or unapproved deviation to every affected scientific artifact.
  • Administrative and financial boundary: record award/budget/effort/contract states only from authoritative records. Do not certify expenditures, effort, compliance, continuation, acceptance or institutional/funder approval.
  • External-action boundary: do not log in, upload, attest, contact the sponsor, request an amendment or submit a report without explicit authority. The maximum machine state is HUMAN_INSTITUTIONAL_FUNDER_ACTION_REQUIRED.

Use POST_AWARD_REPORT_INCOMPLETE when fields, sources or owners are missing and POST_AWARD_REPORT_BLOCKED for a known overdue, authorization, compliance, scientific-validity or unresolved-term blocker. Neither state permits invented completion evidence.

Required workflow

  1. Accept the scientific lock. Receive the question, protocol, estimand, measurement/QC, statistics, ethics and data-access decisions from their owners. Return unresolved science rather than repairing it as project management.
  2. Build the control book. Define work packages, deliverables, acceptance criteria, dependencies, RACI, decision rights and escalation paths. One work package has one accountable role; never invent the person or institutional authority.
  3. Compute the critical path and capacity. Map duration ranges, access lead times, throughput, bottlenecks, slack, external dependencies and resource limits. Separate calendar forecast from scientific sample-size requirements.
  4. Gate site/workstream readiness. Verify approvals, agreements, trained roles, equipment, acquisition/reconstruction versions, DICOM transfer/de-identification, reference standard, annotation, data schema, assay/compute capacity, safety and fallback. Use templates/site-readiness-freeze-register.md.
  5. Control execution. Compare planned and actual recruitment, follow-up, series acceptance, annotation, assay and run throughput. Record denominator, time window, evidence and forecast; never invent universal green/yellow/red thresholds.
  6. Maintain distinct registers. Keep future uncertainty (risk), present blockage (issue), authorized choice (decision), planned modification (change), divergence from the frozen plan (deviation) and safety/security event (incident) separate.
  7. Freeze and release deliberately. A freeze identifies scope, version, manifest, criteria, approver role, time and permitted exceptions. Any post-freeze change creates a new version and downstream-staleness handoff; it is not silently overwritten.
  8. Decide recovery, pause or termination. Tie actions to explicit thresholds and preserved claims. A scientifically invalid, unauthorized or unsafe study cannot be rescued by schedule compression.
  9. Close out. Reconcile participants/cases/specimens/images, data, annotations, equipment, finances, agreements, access, deviations, incidents, code/results, negative evidence and retention/destruction obligations. Use templates/research-operations-control-book.md.
  10. When funder reporting applies, build the award-term crosswalk. Reconcile the current award conditions, approved changes, reporting-period evidence and required fields/files. Return every scientific statement to its original owner for verification, then stop at HUMAN_INSTITUTIONAL_FUNDER_ACTION_REQUIRED for portal, attestation, approval or submission.

Stop and escalation gates

Return OPS_STOP_AND_ESCALATE for absent or expired approval/access; participant or staff safety risk; uncontrolled PHI/security exposure; unqualified critical role; critical-to-quality process without monitoring; material protocol deviation; unreconciled patient/exam/lesion/site identity; protected-test or freeze breach; evidence destruction; or a contract/budget/personnel decision that requires institutional authority.

State the responsible role and minimum evidence for release. The skill never activates a site, authorizes spending, instructs staff to continue blocked work, signs a delegation log, makes an HR or contracting decision, or declares institutional compliance.

Output contract

  1. Operations passport and current evidence state
  2. Work breakdown, RACI and decision-rights map
  3. Critical path, capacity and dependency forecast with uncertainty
  4. Site/workstream readiness matrix and exact blockers
  5. Recruitment/acquisition/annotation/assay/analysis throughput dashboard
  6. Resource and budget-capacity map — quantities and constraints, not financial approval
  7. Risk / issue / decision / change / deviation / incident registers
  8. Freeze/release manifest and exception ledger
  9. Recovery, pause, termination or closeout decision packet
  10. Post-award obligation/report/amendment crosswalk — when applicable, with source versions, scientific-owner verification, unresolved blockers and human funder-action gate
  11. Scientific, ethics, data, institutional and pipeline handoffs

Handoffs and non-ownership

  • Question, estimand, validation and sample-size design -> radiology-design / radiology-stats; acquisition and measurement gate -> radiology-acquisition-qc.
  • Annotation SOP -> radiology-annotation; data identity, de-identification and access -> radiology-data; approvals and safety governance -> radiology-ethics.
  • Grant milestones and requested budget narrative -> radiology-grant; whole-project scientific state and artifact staleness -> radiology-pipeline.
  • Pre-award call/proposal review, resubmission and final application-package audit -> radiology-grant. After award, this skill owns obligation, amendment, annual/interim/final report and sponsor-deliverable tracking, while methods/results/claims remain with the original scientific owner and portal/acceptance authority remains with the institution/funder.
  • Silent/active deployment evidence, reader/impact design, site-acceptance criteria and production monitoring science -> radiology-translation; operations owns the approved plan's RACI, schedule, resources, dependencies, escalation and closeout, not its clinical-validity claim.
  • The PI, sponsor, institution, site, funder and authorized committees retain personnel, financial, contractual, activation, safety, amendment, report-submission, acceptance and termination authority.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Radiology Research Ops AI skill do?

Plan/audit sites, RACI, milestones, dependencies, resources, risks and closeout; not scientific design.

Why use Radiology Research Ops on TypingMind?

Because you install it once and use it with any model. Radiology Research Ops is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Radiology Research Ops in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/huang-sir1/radiology-skills/tree/main/radiology-skills/skills/radiology-research-ops. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Radiology Research Ops?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Radiology Research Ops?

As many as you like. As long as a model supports skills, you can use Radiology Research Ops with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Radiology Research Ops AI skill free?

It is published on GitHub by huang-sir1. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇