Posts
Building tools for platform engineering — AI plugins for Backstage, auth and multi-cluster work in Kubernetes, and the systems around them. Build-first, blog the journey.
-
One login, every cluster: when is it actually safe to broadcast an OIDC token?
Adding opt-in multi-cluster token broadcast to Headlamp — and why issuer + client-id + audience is the whole ballgame.
-
Generating Backstage scaffolder templates from one sentence
Building template-authoring-backend — zod-constrained LLM output, semantic post-validation, and the future ActionsRegistry hook.
-
Backstage as an LLM agent: a 5-cap tool-calling loop with propose-only tools
Building the AgentLoop for incident-copilot-backend — what earns the 'agent' framing, and the five stop conditions that keep it cheap.
-
Building AI plugins for Backstage: a four-part roadmap
Where the work fits — the RFCs, the BEP, the plugins shipped, and the ones still ahead.
-
Catalog-aware LLM Q&A for Backstage (grounded, with citations)
Building catalog-assistant-backend — natural-language questions over your Software Catalog, deliberately wired for the AI Provider Service when it lands.
-
Two open scaffolder bugs, fixed — and what I read on the way there
A field report from contributing to backstage/backstage: PRs #34480 and #34481, plus the latent bugs and CI surprises I bumped into.
-
Backstage as an MCP client: calling any AI tool from a scaffolder template
Building scaffolder-backend-module-mcp — the inverse of mcp-actions-backend. One generic action, any MCP server.