Menu
Grafana on the yai stack — datasource UIDs, file provisioning (not TF), dashboard conventions, scrape targets page, Grafana plugin setup. Auto-load on Grafana, dashboard, datasource, DS_*, panel, templating variable, provisioning questions.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
agent-browser CLI — headless Chromium for AI agents. Snapshot-and-ref workflow, core commands, and yai service interaction patterns. Auto-load when asked to open a browser, navigate a URL, click, fill, screenshot, or interact with any yai web UI (Grafana, n8n, Windmill, Langfuse, MinIO, etc.).
Browserless headless Chromium service on the yai stack — REST endpoints (screenshot, PDF, scrape, function), WebSocket CDP/Puppeteer/Playwright patterns, token auth, sizing. Auto-load on Browserless, headless browser, Puppeteer, Playwright, CDP, screenshot, PDF, scrape-script questions.
Generate clean architecture / flow diagrams as PNG+SVG for the yai stack. Hand-laid-out Python→SVG→PNG pipeline (no Mermaid auto-layout). Auto-load when asked to create, update, or render an architecture diagram, flow chart, system diagram, or "draw the stack".
Firecrawl web crawling & scraping API on the yai stack — scrape, crawl, extract endpoints, job status, concurrency knobs, proxy config. Auto-load on Firecrawl, web scrape, crawl, extract, markdown, web content questions.
How to control every yai web UI — login recipes, navigation patterns, and common actions for Grafana, Langfuse, MinIO, n8n, Windmill, LiteLLM, Qdrant, and Traefik. Auto-load when asked to log in, navigate, click, or perform any action in a yai service UI.
Langfuse on the yai stack — LLM observability (traces, evals, prompt mgmt), REST API, SDK usage, Langfuse–LiteLLM callback wiring. Auto-load on Langfuse, LLM traces, evals, prompt management, observation, session questions.
| name | grafana |
| description | Grafana on the yai stack — datasource UIDs, file provisioning (not TF), dashboard conventions, scrape targets page, Grafana plugin setup. Auto-load on Grafana, dashboard, datasource, DS_*, panel, templating variable, provisioning questions. |
| when_to_use | Load when the user asks about Grafana dashboards, panels, datasources, Grafana API calls, provisioning files, plugin setup, or wants to view/create/update any dashboard or panel on the yai stack. |
URL in env: YAI_GRAFANA_URL (port 22000). API token: YAI_GRAFANA_TOKEN.
Default admin login: admin / password set in grafana/.env.
Grafana in the yai stack uses file-based provisioning, not the
Terraform provider. Datasources are provisioned via files mounted into
the container at /etc/grafana/provisioning/datasources/. Dashboards are
currently managed via the Grafana UI (first-time setup) — file
provisioning for dashboards is a reasonable follow-up.
The Grafana UI is write-enabled — edits you make in the UI persist in
grafana/data/grafana/. They are not committed to the repo.
Datasource URLs use compose internal service names (resolved from inside the Grafana container, not from the host):
| Datasource | Type | Internal URL |
|---|---|---|
| VictoriaMetrics | victoriametrics-metrics-datasource | http://victoriametrics:8428 |
| VictoriaLogs | victoriametrics-logs-datasource | http://victorialogs:9428 |
| VictoriaTraces | jaeger | http://victoriatraces:10428/select/jaeger/ |
Don't substitute localhost:<host_port> — those don't resolve from
inside the Grafana container.
source ./env.sh
H="Authorization: Bearer $YAI_GRAFANA_TOKEN"
# All dashboards
curl -s -H "$H" "$YAI_GRAFANA_URL/api/search?type=dash-db" \
| jq '.[] | {uid, title, folderTitle}'
# One dashboard JSON
curl -s -H "$H" "$YAI_GRAFANA_URL/api/dashboards/uid/<UID>" | jq
# All datasources
curl -s -H "$H" "$YAI_GRAFANA_URL/api/datasources" \
| jq '.[] | {uid, name, type}'
# Health check
curl -s "$YAI_GRAFANA_URL/api/health" | jq
The easiest way to see which scrape jobs are up (served by VictoriaMetrics built-in scraper):
open "$YAI_VMETRICS_URL/targets"
# or as JSON
curl -s "$YAI_VMETRICS_URL/api/v1/targets" | jq '.data.activeTargets[] | {job: .labels.job, health}'
vlogs skill).:pipe format.GF_INSTALL_PLUGINS in grafana/.env triggers download from grafana.com
on container start. In an air-gapped setup, bake a custom Grafana image.
The two required Victoria* plugins:
victoriametrics-metrics-datasource (VictoriaMetrics)victoriametrics-logs-datasource (VictoriaLogs)