A live demo or admin page shipped with every feature whose results the user would otherwise not see. Use when building or extending a server-side capability, service, registry, or pipeline, or adding a backend endpoint or feature with no UI, whose output only shows in logs, the DB, or an agent run, or when the user says there's no way to see it or asks 'how do I test this'.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
A feature the user can't see is a feature you'll both burn turns guessing about. Every new capability ships with a place the user can open to watch it work — built in the SAME change, not "later." This single rule kills the most common waste in this codebase: multiple turns where the human has no window into what the server is doing. Build the page in the SAME change — it's the next file, not a ticket.
When you add or change a server-side capability, also build a live page that renders its real results and lets the user trigger/test it with simple controls. Not a mock — it reads the live backend. The user opens it once and re-checks it after each of your changes.
(dev)/demos/<name>/page.dev.tsx (auth
shell) or (public-demos)/demos/public/<name>/page.tsx (no auth). Auto-listed at /demos.
Use the new-route-scaffold skill to place it.(admin)/administration/<name>/page.tsx (super-admin gated at the layout; lower with an
in-page selectIsAdmin seam if org-admins should reach it later).When unsure, prefer the lower bar (demo) so the user can reach it without ceremony.
GET /<feature>/...
returning a typed Pydantic model (the FE generates TS from it). If the feature does
something, add a small authed POST /<feature>/execute-style endpoint that runs ONE
unit and returns a typed receipt — that's what makes the page a test bench, not just
a viewer. Bare router prefix, response_model=, no dict[str, Any] (see
aidream/api/FEATURE.md). Read endpoints that mirror non-sensitive metadata can be
unauthenticated; anything that writes runs acting_as_user (RLS) and is authed.features/<name>/
feature + the route. Reuse, don't fork: the canonical base URL (selectResolvedBaseUrl
/ apiConfigSlice — never hardcode), the component-library table/select/loading/empty/
error states, the admin gate, and any existing renderer (e.g. the matrx-envelope
block renderer for live chips). Runtime-guard the response; show a real loading + error
state; add a manual Refresh (and light polling) so it reflects the live server with no
redeploy. A few dropdowns / inputs to drive it beat a wall of text.The page itself lives in matrx-frontend (the next section); from aidream you ship piece 1 above — the thin typed endpoint — in the same change, and tell the user the URL of the frontend page that reads it.
Self-contained features/<name>/:
selectResolvedBaseUrl (apiConfigSlice) — the same value every backend call uses, so
the admin server toggle routes it. Runtime-guard the JSON (untrusted). For authed
POST/execute calls use the canonical backend client (lib/api/backend-client.ts →
createAuthenticatedClient), not a hand-rolled fetch.Select/inputs to drive the feature;
show the REAL result (and, for actions, the receipt). Reuse existing renderers — e.g. the
features/matrx-envelope block renderer for live reference chips — never fork one.ui-dense / data-dense-panels). No new barrels; import from source.The base-URL resolver, the auth/backend client, the component library, the admin gate, and the envelope renderer all already exist. A visibility surface is assembled from them in minutes — if you're writing a second fetch layer or a second renderer, stop.
/administration/directive-catalog (matrx-frontend features/directive-catalog/) ← GET /directives/catalog + POST /directives/execute (aidream services/directive_catalog +
services/directive_apply): a live grid of every capability, with dropdowns to build and run
one and see the receipt. That is the bar — a place the user goes to instantly see and test
the whole surface.
/demos index) and uses real
loading/empty/error states.If you built a server capability and there's no such page, you're not done — that's the next file, not a follow-up ticket.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Admin debug system wiring for the floating AdminIndicator's live debug data and Copy Full Context. Use when adding debug visibility to a route or feature, wiring useDebugContext or the debug panel, capturing console errors, or enabling the copy-context workflow for a page.
The English summary is being prepared. Showing the original description.
Compact two-icon Copy / Copy-for-AI controls (components/agent-copy). Use when adding copy buttons to a row, card, list, or record; merging duplicate Copy/JSON/AI controls; continuing the copy rollout; or writing a Copy-for-AI payload. NOT for markdown content actions (use rich-document-actions).
The English summary is being prepared. Showing the original description.
Disclosing a surface's existing fixed AI jobs in the shell's top Agents menu. Use when a page, panel, overlay, or window already runs a mandate behind a button, assist, automatic action, or mode; when the agent-disclosure guard names a file; or during a surface check.
The English summary is being prepared. Showing the original description.
The watch-fix-rerun method for making an existing platform agent or automated process efficient and correct: baseline its ledger, run one unit yourself, fix the class behind every wasted call, rerun, record. Use when asked to improve, optimize, watch, or 'make efficient' an agent, a sandbox session, a sync job, or any recurring automated process.
The English summary is being prepared. Showing the original description.
Redux state for agent execution and firing agent shortcuts. Use when editing features/agents/redux/, building agent UI, creating a conversation, touching assembleRequest or NDJSON stream state, wiring agent-state selectors, adding a per-conversation capability, or triggering a shortcut from a button, menu, or code (useShortcutTrigger, launchAgentExecution).
The English summary is being prepared. Showing the original description.
The provision for a call site: the exhaustive menu of values that place in the code can realistically produce. Use before creating or fixing a mandate or agent that reasons about more than its own input, when deciding whether an agent can answer at all with what it is sent, or when its output 'looks right' but nobody checked it could be. NOT for building the agent (use create-agent).
The English summary is being prepared. Showing the original description.