platform-engineering

The citation, test, and outcome discipline this repo enforces: every test cites a vendor/seed/rubric source via @cite, every commit ends with an outcome ID like (O7), every PR body declares Outcomes + Test plan + Citations sections, and TDD red/green/refactor is surfaced in commit titles. Use when bootstrapping a new repo that should adopt the same discipline, when writing a new test and unsure what @cite to point at, when composing a PR body, or when picking a @tdd stage tag.

Drive Claude in Chrome to screenshot claude.ai/code environment env vars and diff against the OSEC1 parity table. Use when the secrets-parity verifier reports clean for local/gh planes but the cloud_env plane is unaudited (verifier cannot reach the cloud plane via API).

Wire `.lsp.json` for a Claude Code plugin so the agent gets symbol navigation, hover info, and diagnostics from a real Language Server Protocol server (typescript-language-server, pyright, gopls, rust-analyzer, …). Use when adding a new language to a plugin, when Claude is missing definition/refs/hover for a project, or when debugging an LSP that fails to start. References the Claude Code plugin spec's LSP section and the canonical LSP-servers taxonomy.

Declare a Claude Code plugin Monitor (`monitors/monitors.json`) that listens for PR/issue/CI events and dispatches a local handler. Use when wiring an event-driven workflow into a plugin, when an existing monitor isn't firing, or when designing the local handler shape for a GitHub-triggered routine (the trigger itself stays web-UI-only per the routines spec; the monitor declares what runs locally).

Discover and install Claude Code plugins from a marketplace, audit plugin dependencies, and decide whether to add a new plugin vs extend an existing one. Use when adding a new capability to the chassis, when an upstream plugin gets updated, when checking a plugin's transitive dependencies, or when picking between two competing marketplace plugins.