Pexip Claude Skills
lorist/pexip-claude-skills
Skills especializadas para Pexip Infinity: políticas, APIs, SDKs e troubleshooting
Ver no GitHubColeção de skills especializadas para Claude trabalhar com Pexip Infinity, plataforma de videoconferência corporativa. Cada skill captura conhecimento prático sobre configuração de políticas, análise de logs, APIs de management e client, SDKs JavaScript e plugins para webapp. Ideal para engenheiros que constroem e operam infraestrutura de conferência.
Instalação
npx convez-ai lorist/pexip-claude-skills1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Desenhar e implementar servidores External Policy com lógica ABAC em Pexip Infinity
- Analisar raiz de falhas em chamadas usando logs de Management e Conferencing Nodes
- Integrar com Management API REST e Client API para automação e bots de conferência
- Desenvolver plugins webapp3 e aplicações JavaScript com SDKs Pexip nativos
- Configurar webhooks Event Sink para CDR, billing e dashboards em tempo real
Veredito

Ferramenta altamente especializada e documentada para quem trabalha com Pexip Infinity em escala. Excelente para acelerar curva de aprendizado em APIs complexas e padrões operacionais, mas valor concreto limitado a usuários do ecossistema Pexip.
README
A tradução para o português deste README ainda está na fila.
Pexip Claude Skills
A collection of Claude Agent Skills for working with Pexip Infinity and the wider Pexip platform. Each skill captures hard-won, field-tested knowledge that makes Claude meaningfully better at a specific Pexip task — without you having to re-explain context every conversation.
What's in here
| Skill | Purpose |
|---|---|
pexip-external-policy | Designing and implementing Pexip Infinity External Policy servers and local Jinja2 policy scripts. Covers all six request types, ABAC patterns, breakout rooms, and common gotchas. |
pexip-event-sink | Building and operating Pexip Infinity Event Sink receivers — the webhook stream for conference and participant lifecycle events. Covers the wire protocol, the operational patterns (idempotency, fast ACK, durable persistence, replay), and common integrations (CDR/billing, BI, real-time dashboards). |
pexip-call-rca | Root cause analysis for Pexip Infinity call failures from Management Node and Conferencing Node logs (support snapshot, folder of per-node files, or a single log). Bundles a log parser plus reference material on SIP, WebRTC, and Pexip-specific failure patterns. |
pexip-management-api | Working with the Pexip Infinity Management Node REST API (/api/admin/{configuration,status,command,history}/v1/). Ships the full schema for all 154 resources as searchable markdown so Claude can write correct payloads without guessing field names; covers auth, bulk PATCH, rate limits, and the gotchas around DELETE/PUT on list URIs. |
pexip-client-api | Building applications against the Pexip Infinity Client REST API (/api/client/v2/conferences/<alias>/…) — the token-based HTTP+SSE protocol every endpoint and bot uses to join a meeting. Covers the request_token/refresh_token lifecycle, the /events SSE stream and its participant_sync_* bracket, WebRTC negotiation via /calls (transcoded vs direct media), breakouts, host vs guest control, and the operational patterns (refresh cadence, reconnect, idempotency) the docs leave implicit. |
pexip-js-sdks | Choosing between, and building with, the two JS SDK families that sit on top of the Client API: the legacy pexrtc.js browser wrapper and the modern @pexip/* npm packages (@pexip/infinity, @pexip/media, @pexip/components, etc.). Decision tree, full PexRTC reference, the @pexip/signal deep-dive, the media-pipeline composition rules, and runnable examples for both stacks. |
pexip-webapp3-plugin | Building, packaging, and deploying Pexip Infinity webapp3 plugins with @pexip/plugin-api — the sandboxed iframe extensions that add buttons, toasts, and side panels to the stock meeting UI. Covers the plugin SDK (registration, events, UI primitives, sendRequest), the Vite project scaffold, the branding ZIP / manifest.json structure, uploading via the Management API, webapp alias routing, and 22 production-tested gotchas. |
pexip-infinity-debugging | Problem-oriented debugging playbook for live Pexip Infinity deployments — symptom → log source → search term → fix. Covers external policy debugging (silent 404, response validation, PIN rejection), event sink delivery issues, classification state problems, IdP/SAML auth failures, and infrastructure quirks (config replication, licensing, ngrok). Pair with pexip-call-rca for downloaded log archives. |
pexip-policy-server | Production patterns for building and operating Pexip Infinity External Policy servers with participant classification, access control, and conference lifecycle management. Covers service_configuration and participant_properties response validation, classification levels and role ladders (chair/guest/bypass_lock), avatar_request JPEG handling (Keycloak), Client-API-driven conference classification, event-sink integration for admission and breakout migration, and 40 production-tested gotchas. The advanced companion to pexip-external-policy. |
pexip-secure-scheduler | Designing and integrating against the Pexip Secure Scheduler (formerly Pexip Web Scheduler) Portal API — the REST API for scheduling, updating, and inviting participants to Pexip Infinity meetings. Covers the object model (Encounter / Participant / Role / EncounterParticipant / AliasTemplate / BreakoutRoom), the four-step encounter creation recipe, long_lived_token auth, the alias-template Jinja2 surface, breakout-by-role + interpreter routing, iCalendar SEQUENCE for invite updates, and the back-channel that propagates encounters into Infinity VMRs. Ships the v5.1.0 OpenAPI document, 7 topic-specific reference files, and a stdlib Python client. |
mividas-core-rooms | Building, debugging, and operating integrations against the Mividas Core + Rooms REST API — the third-party management platform that sits above Pexip Infinity, Cisco Meeting Server, VCS/Expressway, and individual video endpoints (Cisco CE, Webex Devices, Teams MTR). Covers Basic Auth + X-Mividas-Customer multi-tenancy, the parallel /cospace-acano/ vs /cospace-pexip/ write split, the /calls/... + /call_legs/... live-call control surface, the /endpoint/provision/ → /endpointtask/ provisioning task queue, call/room/meeting/provider statistics, Pexip policy authorization and Call Routing Rule sync, address-book sources, and the /debug/* log surfaces. Ships the full OpenAPI document, 8 topic-specific reference files, and a stdlib Python client. |
| (more to come) |
What is a Claude Skill?
A skill is a folder containing a SKILL.md (with YAML frontmatter describing when to trigger it) plus optional reference files and scripts. Claude consults relevant skills automatically based on the conversation. Skills work with:
- Claude Code (CLI) — drop the skill into
~/.claude/skills/<skill-name>/or<project>/.claude/skills/<skill-name>/ - Claude.ai apps (Pro/Max/Team/Enterprise) — upload a packaged
.skillfile via Settings → Capabilities → Skills - Claude API / Agent SDK — load via the code execution tool
The format is plain markdown plus YAML frontmatter — portable, reviewable, and version-controllable.
Installation
Install all skills (recommended)
Clone the repo and run the install script:
git clone https://github.com/lorist/pexip-claude-skills.git
cd pexip-claude-skills
./scripts/install.sh
This copies every skill in skills/ to ~/.claude/skills/, making them available in Claude Code immediately. Override the destination with CLAUDE_SKILLS_DIR=/custom/path ./scripts/install.sh.
Install a single skill
./scripts/install-one.sh pexip-external-policy
Windows (PowerShell)
The .sh scripts above are Bash and need Git Bash or WSL. In native PowerShell or cmd.exe, use the PowerShell installer instead:
git clone https://github.com/lorist/pexip-claude-skills.git
cd pexip-claude-skills
.\scripts\install.ps1 # all skills → %USERPROFILE%\.claude\skills
.\scripts\install.ps1 -Skill pexip-external-policy # just one
.\scripts\install.ps1 -Symlink # symlink (needs Developer Mode or admin)
.\scripts\install.ps1 -Dest C:\custom\skills # custom destination
If PowerShell blocks the script with an execution-policy error, run it for the current process only: powershell -ExecutionPolicy Bypass -File .\scripts\install.ps1.
Install for Claude.ai (web/desktop)
Build packaged .skill files and upload them via the Claude.ai settings:
./scripts/package.sh # outputs dist/<skill-name>.skill for every skill
Then in Claude.ai: Settings → Capabilities → Skills → Upload skill and pick the .skill file. (Available on Pro / Max / Team / Enterprise plans.)
Manual install (no script)
A skill is just a folder. Copy it into your skills directory:
cp -r skills/pexip-external-policy ~/.claude/skills/
Or, for active development against a single machine, symlink it:
ln -s "$(pwd)/skills/pexip-external-policy" ~/.claude/skills/pexip-external-policy
Symlinking means git pull instantly updates the installed skill.
Using a skill in Claude Code
Once installed, you can invoke a skill two ways.
Just ask — Claude picks it up automatically
Each skill's SKILL.md declares the topics it covers in its frontmatter description. Natural prompts on a relevant topic pull the skill in without any ceremony:
"Help me build a FastAPI policy server that overrides participant roles based on IdP groups."
"What's the correct response shape for a Pexip breakout room?"
"My
media_locationresponse isn't being applied — what should I check?"
Claude reads the relevant SKILL.md plus any references/ files it needs and answers with field-accurate guidance.
Invoke explicitly with a slash command
To force-load a skill at the start of a turn — useful when the topic isn't obvious from the prompt, or when you want to make sure Claude is grounded in the skill before you start — type the skill name as a slash command:
/pexip-external-policy create a sample external policy server that rejects calls from unauthenticated IdP groups
Claude loads the skill and treats the rest of your message as the request.
Confirming a skill is installed
ls ~/.claude/skills/
You should see one directory per installed skill, each containing a SKILL.md.
Worked end-to-end examples
For complete builds — code, README, and a case study evaluating how the skill performed during the actual build — see skills/pexip-external-policy/examples/.
Contributing a new skill
See docs/CONTRIBUTING.md. Short version: copy docs/skill-template/ into skills/<your-skill-name>/, fill in the SKILL.md and any references, and run ./scripts/validate.sh before opening a PR.
Versioning
This repo uses single-tag semantic versioning across all skills. Tagged releases (v1.0.0, v1.1.0, …) on GitHub include built .skill files as release assets for easy distribution to non-developers.
Licence
CC BY-NC 4.0 — free to share and adapt for non-commercial use, with attribution. See LICENSE for the full notice.
Acknowledgements
These skills are distilled from real-world Pexip deployments — production policy servers, ABAC enforcement, Webapp3 plugins, partner integrations, and a lot of trial and error. If you find a gap, please open an issue or PR.