Refresh the README: new screenshots, the Builder, and four stale claims - #335
Merged
Conversation
The README had not been updated in months. Its screenshots predated the app bar redesign, the View panel rebuild, the canvas connection rewrite, and the Builder entirely, and several statements had become false. Screenshots are replaced with four new ones, committed to the repo under assets/screenshots/ rather than linked to GitHub upload URLs: the branching canvas (hero), a finished Builder run showing its checklist, budgets and activity log beside the nodes it created, the Builder launch dialog, and a code-to-chart workflow. Each was captured from the real app rendering real scene state - the demo content is authored, the rendering is not. Corrected claims, each verified against the code first: - Provider mode was described as not switchable in the UI. Settings has fired the real setProviderMode intent since ADR-006/ADR-012. - File attachments were described as backend-only with "no way to attach or ingest a file from the interface". The composer has had real staged attachments since R8a. - Test coverage was described as having no browser-driven UI testing. A Playwright suite runs as its own CI job (ADR-015). - CI was described as two jobs. It is four: Python checks, frontend checks, E2E, and a wheel build check. Added: a section on the Builder, and feature/node-kind coverage for the work that had landed without ever reaching the README - the knowledge base and global search, workspaces, undo of a whole agent run, per-branch model pinning, charts, themes, the plugin SDK with capability grants, MCP servers, diagnostics, and the token/cost counter. The node-kind list went from 5 kinds to the real 15, and the plugin directory layout ADR-014 introduced is now documented.
test_evict_idle_session_vetoes_while_autosave_is_preparing_to_write waits for a real SQLite write to land on a background task after a deferred eviction releases it. The poll allowed 2 seconds (200 x 10ms), which is comfortable locally but marginal on CI, where this suite runs under coverage tracing on a shared Windows runner - it failed there on a docs-only change, having passed on the immediately preceding PR. Only the wait budget changes. The assertion is untouched, so an autosave that genuinely never completes still fails the test; it simply no longer fails because a loaded runner needed longer than two seconds.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
The README had not been updated in months. Its screenshots predated the app bar redesign, the View panel rebuild, the canvas connection rewrite, and the Builder entirely. Several statements had also become false, and a substantial amount of shipped functionality had never reached it at all.
Change
Screenshots. Four new ones, committed under
assets/screenshots/rather than linked to GitHub upload URLs so they live with the code that produced them: the branching canvas (hero), a finished Builder run showing its checklist/budgets/activity log beside the nodes it created, the Builder launch dialog, and a code-to-chart workflow. Each was captured from the real app rendering real scene state — the demo content is authored, the rendering is not.Corrected claims, each verified against the code before changing:
setProviderModeintent since ADR-006/ADR-012Added. A section on the Builder, plus feature and node-kind coverage for work that had landed without ever reaching the README: the knowledge base and global search, workspaces, undo of a whole agent run, per-branch model pinning, charts, themes, the plugin SDK with capability grants, MCP servers, diagnostics, and the token/cost counter. The node-kind list went from 5 to the real 15, and the
plugins/package layout ADR-014 introduced is now documented.Test plan
knowledge/search,globalSearch/search,scene/undoRun,scene/setModelOverride,app-settings/setTheme,app-settings/setMcpServers,app-chat-library/createWorkspace,app-chat-library/exportWorkspace,app-plugins/setPluginGrant,diagnostics/exportDiagnosticBundle)Docs and images only — no application code changes.