Repository navigation
Installation
Panel 1.1.0 requires Grafana 11.6 or later and digitalrcs-intelligencegateway-datasource 1.1.0 or later. Install the companion first. The optional Grafana LLM app has its own requirements: app 1.2.0 requires Grafana 13. The Current View Exporter plugin is not required for image capture.
Use Node.js 22, the declared npm version, Go 1.26.10 or later as required by the companion's go.mod, Mage, Docker, and Docker Compose. Keep the panel and companion data-source repositories as sibling directories.
cd digitalrcs-intelligencegateway-datasource
npm install
npm run build
go run github.com/magefile/mage@v1.17.2 -v build:linux
cd ..\digitalrcs-intelligencegateway-panel
npm install
npm run build
docker compose up --buildOpen http://localhost:3004. The development Compose configuration mounts both unsigned plugins, starts a credential-free deterministic mock provider, and provisions DEMO ONLY - Mock AI (no inference). The supplied default returns a fixed receipt: no model runs and no data or image analysis occurs. The TestData chart supplies synthetic rows, not an AI model. Follow the two-phase walkthrough to create a separate real-provider instance, then select that instance and model under the panel's AI provider settings. Read the current AI connection label after switching; the introduction describes only the supplied default.
Obtain the matching companion and panel archives from their release pages and follow the artifact and signature instructions for that exact version. Install the companion first. The plugin IDs are digitalrcs-intelligencegateway-datasource and digitalrcs-intelligencegateway-panel. Restart Grafana after installing a build or changing either plugin.json, then fully reload the browser page.
Verify release checksums and provenance against the downloaded archives. A provenance attestation establishes build origin; it is not a Grafana plugin signature or catalog approval. Do not assume a release is signed or approved solely because source code or release documentation exists. Use the compatibility and certification guide for the recorded release status. Do not enable unsigned-plugin loading in production as a substitute for signing.
When upgrading from 1.0.1, retain the secure AI instance and existing data-only workflow. Image support is optional and the backend vision policy is off by default. To enable it, approve a vision-capable model in both model lists, then use Image Analysis. Review provider and network policy before sending new types of evidence.
Private instances can be signed with npm run sign -- --rootUrls https://grafana.example.com/. Catalog releases must meet Grafana's publishing and signature criteria.
If an old UI remains after a verified installation, use a full-page reload or a fresh private session to compare. Dashboard Refresh reloads data, not plugin code. For same-version local builds, follow cache guidance; do not clear unrelated browser credentials or weaken server settings.
- Home
- Reviewer Walkthrough
- Panel Setup and Configuration
- Connecting Data from Other Panels
- Image Analysis
- Copilot Studio and Enterprise Authentication
- Configuration: AI Providers
- LM Studio Setup
- OpenAI Setup
- Prompt Engineering Guide
- Security Considerations
- Secure Backend and Secret Storage
- Installation
- Release notes: 1.1.2
- Release notes: 1.1.0
- Grafana Compatibility and Certification
- Development and Contributing
- Roadmap