futrx-com/remote.futrx/remote-application
Claude Code and agent skill by futrx-com/remote.futrx
- community
Checking that the SKILL.md file is still in the repository.
Create or extend an installable remote.futrx application from product requirements. Use when asked to scaffold or build a catalog application, design its application.json, or add application-owned infrastructure, container programs, a host backend, publishers, UI extensions, or project skills. Do not use for unrelated Remote core features.
Install
npx vskill@latest install futrx-com/remote.futrx/remote-applicationSKILL.md
Remote Application
Turn the user's requested behavior into a complete, minimal Remote application. Build only capabilities backed by real implementation; never add demonstration fields, credentials, database metadata, services, or events that the application does not actually have.
Establish the target
- Locate the
remote.futrxrepository root and inspect its instructions, Git status, and existing application/package context before editing. - Determine the application ID, display name, supported scope, delivery form, and observable behavior from the request. Ask only when a missing choice would materially change the product; otherwise make a narrow, explicit assumption.
- Decide whether the user wants a built-in catalog entry under
applications/<id>/or a standalone directory/ZIP for upload. Preserve the repository's policy when the location is already clear. - Read references/capability-selection.md before selecting folders or manifest fields.
Read the live contract
The repository documentation is authoritative and may have changed since this skill was written. Always read these files from the current checkout before implementing:
applications/README.mddocs/dev/installable-applications/01-overview.mddocs/dev/installable-applications/02-application-json.mddocs/dev/installable-applications/03-application-capabilities.mddocs/dev/installable-applications/11-testing.mddocs/dev/installable-applications/13-security-model.md
Then read only the capability-specific documents that apply:
- container provisioning, commands, a manifest service, ports, or health checks:
04-install-scripts.mdand17-versions-and-upgrades.md - UI contributions:
05-slots.md,06-extension-api.md,08-scoping-and-visibility.md, and09-styling-and-icons.md - host backend:
15-application-backends.md - uploaded package or ZIP:
16-uploaded-packages.md - publishers or subscriptions:
18-application-events.md
Use applications/hello-remote/ as the worked architectural example when the request involves multiple layers, a service, or events. Do not copy its unrelated showcase capabilities.
Design the smallest honest application
- Start with
README.mdandapplication.json. The directory name and manifestidmust match;versionis a non-empty string; scopes must reflect where the behavior can actually run. - Declare only inputs a user can meaningfully configure and the code consumes. Omit
connectionunless a real user, password, or database exists. - Treat secrets as secrets across the manifest, logs, UI responses, health checks, and documentation.
- Prefer platform-owned mechanisms over application-owned plumbing: manifest services own systemd, declared ports own LXD proxying, health checks own readiness, and host tools own verified downloads.
- Put only genuinely custom, idempotent container provisioning in
infra/install.sh. - Keep host execution and container execution separate. The host backend root is
backend/main.go; request handling belongs inbackend/api/; container programs belong inbackend/container/. - Keep
backend/main.goas the composition root. It constructs dependencies and serves the API; it does not own request behavior or publisher payloads. - For each application publisher, keep its consumer-facing interface, concrete emitter, event identity, and payload construction together under
backend/lifecycle/. Construct each publisher separately inbackend/main.gofrom core-ownedRuntime.Eventsand inject its interface into the API. - The manifest is the publisher registry. Application code chooses when business events occur; core validates and dispatches emissions. Do not add
subscriptionsunless the application truly consumes another publisher's events. - Keep UI modules disposal-safe and scoped. Use Remote's extension APIs and theme tokens; do not reach into undocumented SPA internals.
- Add an application-owned skill under
skills/<name>/SKILL.mdonly when installing the application should publish that workflow into a project.
Implement a vertical slice
Build the shortest end-to-end path that proves the requested behavior:
- Write the manifest and package README.
- Add only the selected capability folders.
- Connect each declared field to implementation and an observable outcome.
- Add focused tests at the owner of each behavior. For events, test publisher name, event name, version, payload, and the API trigger separately.
- Update application documentation with install, use, verification, upgrade, and cleanup behavior that actually exists.
- Bump an existing application's version when installed copies must reconverge. A brand-new application starts at the version chosen by the user or the repository convention.
Avoid speculative abstractions, placeholder screens, fake resources, and broad showcase behavior. A user asking for one tool should get one coherent tool, not another Hello Remote.
Validate proportionally
Run focused checks first, then the repository suites affected by the package:
jq empty applications/<id>/application.jsonbash -nfor each shell script- application package tests and the catalog registry tests
go test ./...from the catalog module when Go source is presentgo test ./...frombackend/for catalog, installer, host-builder, or core integration changesnpm testandnpm run buildfromfrontend/when UI behavior or its contract changedgo vet ./...in every changed Go module when the change is ready to hand offgit diff --checkand a final search for stale names, placeholder fields, secret exposure, and undeclared or unused manifest capabilities
When the app has a host backend, include a test that compiles it through Remote's generated-module builder rather than relying only on the repository module. When it provisions a container service, test the manifest contract and provide safe runtime verification steps.
Do not push, upload, install, deploy, or mutate a live catalog unless the user explicitly asks. Report the files created, capabilities selected, checks run, and any intentionally omitted capability.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingbackend/internal/agent/provisioning/assets/skills/remote-application/SKILL.md
Popularity
- Repository stars
- 699
- Repository forks
- 93
- 7-day trend score
- 38
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/futrx-com/remote.futrx/remote-application)