Drasi skill package
This package contains one bundle-first Drasi agent skill.
Installation
For GitHub Copilot project skills, copy the drasi folder to one of the supported project skill locations:
.github/skills/drasi/.agents/skills/drasi/.claude/skills/drasi/
For personal skills, copy it to a supported personal skill location such as ~/.copilot/skills/drasi/ or ~/.agents/skills/drasi/.
The top-level SKILL.md is the router. Load bundle guides only when the task needs them.
Bundles
azure-hosting: Drasi Server on Azure, Container Apps, azd, ingress, revision health.sources: Source providers, CDC prerequisites, credentials, identity, lifecycle.continuous-queries: Query authoring, Cypher/GQL, joins, result contracts.reactions: Reaction setup, delivery reliability, Event Grid, SignalR, MCP, webhooks.operations: Runtime triage, inactive resources, missing events, logs, API/CLI drift.observability: Logs, metrics, dashboards, alerts, SLOs, telemetry.delivery: CI/CD, GitOps, promotion, release evidence.recovery: Safe recovery and destructive-change controls.security: Auth, private ingress, secrets, identity, supply chain.validation: End-to-end testing and acceptance evidence.agent-integration: MCP and agent-facing contracts.examples: Minimal examples and scaffold guidance.currency-and-ai-context: Official context, docs freshness, release verification, drift control.synthetic-user-testing: Naive-user validation for tutorials, scaffolds, and Dev Container/Codespaces flows.scaling-and-capacity: Capacity planning, load tests, HPA/KEDA, backpressure, scaling evidence.drasi-lib: Embedded Rust change detection withdrasi-lib.drasi-server: Drasi Server standalone runtime. Web UI, REST API, Solution Templates, Instances, plugins, webhook mode, Docker/binary deployment.custom-plugins: Custom Source/Reaction plugins, plugin SDK internals, and query-language extension guidance.developer-experience: Local workflow, VS Code extension, Dev Containers, Codespaces, onboarding.
Templates
The templates/ folder contains reusable handoff templates for currency checks, acceptance evidence, architecture decisions, and risk registers. Use them when a task needs durable PR, release, or customer-facing evidence.
Quality checks
Run the package validator after editing:
python scripts/validate-skill-package.py --root .The validator checks frontmatter, bundle catalog consistency, guide coverage, stale checked dates, and common anti-slop issues such as production latest usage.