Verifies that .dark CSS selectors in console.css and input.css are
properly scoped to CopilotKit elements and do not leak into the host
application. Tests detect the specific rgb(69, 69, 69) color leak that
occurred when .dark was used as a standalone selector.
The langgraph.json in the starter was copied verbatim from the demo package
with path ./src/agents/main.py:graph, but the starter flattens agents into
the agent/ dir so the correct path is ./main.py:graph.
Also wraps A2UI agent imports in try/except so the starter starts cleanly
even if copilotkit's ToolRuntime patch isn't available — these are optional
features that shouldn't prevent the core agent from running.
Fixes the generate-starters.ts script to rewrite langgraph.json graph paths
when copying from packages to starters, preventing this drift in the future.
The com.ag-ui.community artifacts aren't published to Maven Central,
so the Docker build fails when resolving dependencies. The demo package
Dockerfile already handles this by cloning the ag-ui repo and running
mvn install before building the agent — mirror that in the template
and regenerate the starter.
The V1 CopilotRuntime constructor did `{...endpointAgents, ...agents}`
which silently spread a factory function to `{}`, losing all agents.
Anyone using the V1 API with a factory function got zero agents and no
error. This wraps factory functions so endpoint agents are merged at
resolution time instead of construction time.
The _sanitize_for_json function was applied to state sync events and
get_state responses but not to the raw event stream yielded at line 507.
Events with NaN/Infinity float values in their data payloads would
produce invalid JSON (literal NaN) that downstream parsers cannot handle.
Verifies langchain_messages_to_copilotkit correctly concatenates all text
parts from list-style content blocks (Anthropic models). Includes
regression tests for the exact scenario from #1748 (text + image blocks)
and tests for string lists, mixed content, empty lists, and edge cases.
When Anthropic models return multi-part content lists, only the first
element was used and the rest discarded. Now iterates all parts and
concatenates text blocks, preserving the full message content.
The original fix only covered _emit_state_sync_event but missed the
get_state() code path, which also returns state containing Pydantic
BaseModel instances to callers that will JSON-serialize downstream.
Also fixes filter_state_on_schema_keys returning None implicitly when
schema keys are not set (the except branch returned state but the
non-matching if branch did not).
Adds 17 tests covering _serialize_state, _emit_state_sync_event, and
get_state with Pydantic models (nested, lists, plain dicts, empty).
The original test only tested the _merge_emit_state helper in isolation
using mocks. Replace with integration tests that simulate the actual
state-tracking loop from _stream_events, including:
- Sequential emits with different keys preserve all keys (core bug)
- Proof that the bug manifests without current_graph_state.update
- Three sequential emits accumulate correctly
- Same key emitted twice uses latest value
- Non-dict emit does not corrupt current_graph_state
- Initial state reference is not mutated
The _merge_emit_state call merges emitted state with current_graph_state
into manually_emitted_state, but the `continue` statement skips the
later current_graph_state.update(updated_state). Without this update,
the next manual emit merges against stale current_graph_state and loses
keys from the previous emit.
Also guards the update call against non-dict values from _merge_emit_state
to prevent TypeError on edge-case non-dict emits.
Sequential emit_state calls with different keys now preserve all keys
in the snapshot. Previously, each call would replace the entire
manually_emitted_state, losing keys from earlier calls.
spring-ai: Use maven:3-eclipse-temurin-21 image with mvn instead of
missing ./mvnw wrapper.
ms-agent-dotnet: Replace addgroup/adduser (not available on aspnet:9.0)
with groupadd/useradd. Applied same fix to all four Dockerfile templates
for consistency.
mastra: Make Dockerfile.typescript template use {{AGENT_DIR}} variable
so mastra correctly gets src/mastra/ instead of hardcoded agent/.
Updated generate-starters.ts to substitute variables in Dockerfiles.
Remove orphaned demo-wrapper.tsx (1 file) and error-boundary.tsx (17 files)
that still imported from the deleted @copilotkit/showcase-shared.
Regenerate registry.json, demo-content.json, and starter-content.json.
Generated from template via showcase/scripts/generate-starters.ts.
Each starter is fully self-contained with Sales Dashboard + 5 renderers,
self-contained agent backend, per-language Dockerfile (non-root user),
and deterministic entrypoint with agent health checks.
Marked linguist-generated=true in .gitattributes.
Remove showcase/shared/frontend/ directory and its pnpm-workspace.yaml
entry. All 17 demo packages and 17 starters are fully self-contained
with no shared frontend dependency.
- Remove @copilotkit/showcase-shared webpack alias from all 17 next.config.ts
- Remove @copilotkit/showcase-shared tsconfig paths from all 17 tsconfig.json
- Update all 17 manifest.yaml starter sections to point to showcase/starters/<slug>
with Sales Dashboard name, npx degit clone command, and Railway demo URLs
- Add OPENAI_BASE_URL/OPENAI_API_KEY env passthrough to all 17 playwright configs
for aimock-backed e2e testing
Replace shared-import demo pages across all 17 showcase packages with
the original self-contained implementations from git history. Each demo
is now fully readable in the Code tab with zero @copilotkit/showcase-shared
imports, no demonstrationCatalog, and no shared error boundaries.
9 demos x 17 packages = 153 page.tsx files reverted.
- Replace custom { name, props } schema format with spec-aligned inline
catalog format (allOf + properties) so the LLM sees the same flat
structure it must produce — eliminates "props" nesting confusion.
- Restructure generation prompts: inline literal values are the default,
path binding is a narrow schema-driven exception for form inputs.
- Export InlineCatalogSchema type from a2ui-renderer.
LLMs sometimes use path bindings (e.g. {"path": "/chartData"}) on
component properties that only accept literal values, causing silent
render failures. The new guideline tells the LLM to check the schema's
anyOf type before using path bindings.
## Summary
- The `generate-starters.ts` script was flattening
`src/main/{java,resources}` into `agent/{java,resources}`, breaking
Maven builds which expect the standard `src/main/` layout
- Added a post-copy restructure step that moves `agent/java/` to
`agent/src/main/java/` and `agent/resources/` to
`agent/src/main/resources/`
- Removed the now-redundant explicit resources copy that was duplicating
work
- Updated tests to verify the correct Maven layout and assert the
flattened directories no longer exist
## Test plan
- [x] `npx vitest run
showcase/scripts/__tests__/generate-starters.test.ts` — all 251 tests
pass
- [x] Verified regenerated `showcase/starters/spring-ai/agent/` has
`src/main/java/` and `src/main/resources/` (not flattened `java/` and
`resources/`)
- [x] Verified Dockerfile `COPY agent/ ./` is compatible with the new
layout (no changes needed)
## Summary
- Replace TODO placeholders in all 73 QA checklist files across 17
showcase packages with specific, actionable test steps derived from
reading actual demo page source code
- Each checklist now references real data-testid selectors, hook names
(useFrontendTool, useRenderTool, useLangGraphInterrupt,
useHumanInTheLoop, useAgent, useConfigureSuggestions), expected UI
states, and interaction flows
- Stub demos (shared-state-write, shared-state-streaming, subagents) are
marked as stubs with basic chat verification steps
- Add `showcase/QA-COVERAGE.md` with a comprehensive test coverage
matrix covering manual QA, Vitest unit tests, Playwright E2E (smoke +
interaction), per-package E2E, aimock fixtures, and CI automation for
all 9 demos plus the Sales Dashboard starter hero
## Test plan
- [ ] Verify QA checklist files no longer contain generic "TODO: Add
checks specific to X" placeholders
- [ ] Verify each checklist references the correct demo-specific
features (e.g., agentic-chat mentions background change + weather card,
hitl mentions step selector + accept/reject)
- [ ] Verify QA-COVERAGE.md matrix entries are accurate by
cross-referencing with actual test files
- [ ] Verify framework names in titles are preserved correctly across
all 17 packages
The generate-starters script was flattening src/main/{java,resources} into
agent/{java,resources}, breaking Maven builds. Add a post-copy step that
moves them to agent/src/main/{java,resources} as Maven expects.
## Summary
- The langgraph-python starter's agent loops endlessly on `query_data`
instead of calling `pieChart`/`barChart` frontend tools to render charts
- Root cause: system prompt doesn't mention the frontend chart tools,
and the `query_data` tool description encourages calling it ("Always
call before showing a chart") without saying what to do after
- Updated system prompt to explicitly instruct the agent to call
`query_data` once then use `pieChart` or `barChart`
- Updated `query_data` tool description to reinforce single-call
semantics
- Fixed in both the source package (`showcase/packages/`) and the
generated starter (`showcase/starters/`)
## Test plan
- [ ] Deploy langgraph-python starter and click "Pie Chart" suggestion
pill
- [ ] Verify agent calls `query_data` once, then calls `pieChart`
frontend tool
- [ ] Verify no infinite loop on `query_data`
- [ ] Repeat for "Bar Chart" suggestion pill
The agent's system prompt didn't mention pieChart/barChart frontend
tools, so gpt-4o-mini would call query_data repeatedly instead of
rendering charts. Updated the system prompt to explicitly instruct
the agent to call query_data once then use pieChart/barChart, and
updated the query_data tool description to reinforce single-call
semantics.
Jordan Ritter