copilotkit 4.8.4 → 4.9.0

package/onboarding/prompts/credentials/finalize-plan.md

@@ -20,31 +20,44 @@ succeeds. Record that state in the plan.
 If the developer needs a project or another project, do the selection yourself from the
 target directory. Do not send the developer to another terminal.
 
-List the choices first:
+Read the choices first:
 
 ```text
-npx copilotkit@4.8.4 project list --json
+npx --yes copilotkit@4.9.0 project list --json
 ```
 
 With `--json` the payload is the only thing on standard output, so it is safe to parse.
-Show the developer the project names and ask which one to use, or ask for a name for a new
-project. This is the same kind of question as the frontend choice, so ask it here rather
-than ending the turn.
+Show the current selection and the five most recent projects. Sort recency by `createdAt`.
+Offer search, create-new, and list-all options. Do not show every project unless the
+developer asks.
+
+If the developer searches, run:
+
+```text
+npx --yes copilotkit@4.9.0 project list --search <query> --json
+```
+
+If the developer asks for all projects, show the result of the first list command. Ask
+which project to use, or ask for a new project name. Ask this with the frontend and model
+choices when those choices remain unknown.
 
 Then record their answer, from the target directory:
 
 ```text
-npx copilotkit@4.8.4 project select --project <slug-or-id> --json
+npx --yes copilotkit@4.9.0 project select --project <slug-or-id> --json
 ```
 
 Use `--create <name> --json` instead when they want a new project. The two flags cannot be
 combined. A slug that does not exist fails and lists the real ones, so a typo cannot record
 a selection that points at nothing.
 
-Read the JSON result. It reports `api_key_provisioned`. Key provisioning is deliberately
-non-fatal, so this command can persist the selection, warn on standard error, and still
-exit zero with no key written. Treat `api_key_provisioned` as false the same way as an
-error and stop onboarding: a scaffold with no key looks finished and is not.
+Read the JSON result. It reports `selected_project_slug`, `api_key_provisioned`,
+`project_file_written`, and `environment_file_written` at the top level. The result does
+not contain a secret. Report `selected_project_slug` as the project slug that the server
+selected or created. Require `api_key_provisioned`, `project_file_written`, and
+`environment_file_written` to be true. If one is false, report the error and stop
+onboarding. Key provisioning is non-fatal, so the command can persist a selection and
+still exit zero with no key. A scaffold with no key looks finished and is not.
 
 Never print the payload or any secret value.
 
@@ -103,10 +116,10 @@ values.
 Show only the selected framework, frontend, model, planned file changes, validation commands, and proof steps.
 
 If the developer approves the plan, run
-`npx copilotkit@4.8.4 onboard read implementation/build-and-validate`.
+`npx --yes copilotkit@4.9.0 onboard read implementation/build-and-validate`.
 
 If no exact supported path or documentation URL exists, run
-`npx copilotkit@4.8.4 onboard read unsupported/no-validated-path`.
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.
 
 ## Planning subagent brief
 
@@ -120,6 +133,11 @@ Use the selected framework, frontend, model, repository findings, and documentat
 from the main coding agent. Fetch every selected URL in one step rather than one after
 another. Do not use remembered CopilotKit instructions.
 
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Report a
+documentation gap only after a second method also fails.
+
 Name the application the project asks for, and plan that application. Each documentation
 page teaches through one worked example, and that example carries a domain of its own.
 The domain belongs to the page. The project's purpose from the repository findings
@@ -135,11 +153,18 @@ Preserve each agent or frontend that already exists.
 
 Plan the runtime to consume the Intelligence credential. The runtime takes an
 `intelligence` option holding a client built from the project key. A runtime given a
-`runner` option instead is the SSE runtime. It never reads the key, threads are not
-durable, and the Inspector reads the project as locked. The two options cannot be
+`runner` option instead is the OSS runtime. It never reads the Intelligence key, and the
+Inspector reads the project as locked. The default in-memory OSS runner is ephemeral.
+SQLite, custom, or framework persistence can be durable. The two options cannot be
 combined. Take the constructor from the connect-your-runtime page. Where a framework
 quickstart shows a `runner` option instead, the connect-your-runtime page wins.
 
+When the starting state is `both-oss`, use its recorded live baseline evidence. Preserve
+the working agent, frontend, CopilotKit integration, and OSS behavior. Plan only the
+project selection, Intelligence runtime configuration, and authenticated proof needed for
+the conversion. Preserve the existing persistence and user-visible request. Do not
+rebuild a path that already works.
+
 Where the SDK requires an application-level value that the repository cannot supply, such
 as an end-user identity for threads, use one clearly marked local placeholder and name what
 production requires instead. Do not stop onboarding to ask the developer for it.

package/onboarding/prompts/credentials/plan.md

@@ -4,16 +4,29 @@ Use the repository findings to select the agent framework. Ask the developer onl
 choices that the repository does not show. Do not change application code in this phase.
 Do not read, show, store, or request a secret value.
 
-This release validates six agent frameworks:
+These are the agent frameworks and their default vendors:
 
-| Framework                          | Default vendor |
-| ---------------------------------- | -------------- |
-| LangGraph Python                   | OpenAI         |
-| LangGraph TypeScript               | OpenAI         |
-| ADK                                | Google         |
-| Microsoft Agent Framework (Python) | OpenAI         |
-| Microsoft Agent Framework (.NET)   | GitHub Models  |
-| Mastra                             | OpenAI         |
+| Framework                          | Default vendor  |
+| ---------------------------------- | --------------- |
+| AG2                                | OpenAI          |
+| ADK                                | Google          |
+| Agno                               | OpenAI          |
+| Built-in CopilotKit agent          | OpenAI          |
+| Claude Agent SDK (Python)          | Anthropic       |
+| Claude Agent SDK (TypeScript)      | Anthropic       |
+| CrewAI Flows                       | Project-defined |
+| Deep Agents                        | OpenAI          |
+| LangGraph FastAPI                  | OpenAI          |
+| LangGraph Python                   | OpenAI          |
+| LangGraph TypeScript               | OpenAI          |
+| LlamaIndex                         | OpenAI          |
+| Mastra                             | OpenAI          |
+| Microsoft Agent Framework (.NET)   | GitHub Models   |
+| Microsoft Agent Framework (Python) | OpenAI          |
+| MS Agent Harness (.NET)            | GitHub Models   |
+| Pydantic AI                        | OpenAI          |
+| Strands Agents (Python)            | OpenAI          |
+| Strands Agents (TypeScript)        | OpenAI          |
 
 If the project already has an agent in a listed framework, preserve that framework and
 its model setup. Ask only about choices that remain unknown.
@@ -21,18 +34,37 @@ its model setup. Ask only about choices that remain unknown.
 If the project needs an agent, show the options and one recommendation based on repository
 evidence. Match the language the repository already uses. Ask the developer to choose.
 
+If the project is empty, state that no repository-based recommendation exists. Give these
+documented defaults: LangGraph TypeScript, LangGraph Python, ADK, Microsoft Agent Framework
+Python, Microsoft Agent Framework .NET, and Mastra. Explain which default matches the
+developer's stated purpose. Show the other listed frameworks as options. Do not describe a
+documented default as a repository-based recommendation.
+
 Do not offer a framework that is not listed above. Do not port an existing agent to
 another framework. Do not show the internal route.
 
 Use exactly one matching internal route:
 
-1. LangGraph Python: `npx copilotkit@4.8.4 onboard read framework/langgraph-python`
-2. LangGraph TypeScript: `npx copilotkit@4.8.4 onboard read framework/langgraph-typescript`
-3. ADK: `npx copilotkit@4.8.4 onboard read framework/google-adk`
-4. Microsoft Agent Framework Python: `npx copilotkit@4.8.4 onboard read framework/ms-agent-python`
-5. Microsoft Agent Framework .NET: `npx copilotkit@4.8.4 onboard read framework/ms-agent-dotnet`
-6. Mastra: `npx copilotkit@4.8.4 onboard read framework/mastra`
+1. AG2: `npx --yes copilotkit@4.9.0 onboard read framework/ag2`
+2. Agno: `npx --yes copilotkit@4.9.0 onboard read framework/agno`
+3. Built-in CopilotKit agent: `npx --yes copilotkit@4.9.0 onboard read framework/built-in`
+4. Claude Agent SDK Python: `npx --yes copilotkit@4.9.0 onboard read framework/claude-sdk-python`
+5. Claude Agent SDK TypeScript: `npx --yes copilotkit@4.9.0 onboard read framework/claude-sdk-typescript`
+6. CrewAI Flows: `npx --yes copilotkit@4.9.0 onboard read framework/crewai-flows`
+7. Deep Agents: `npx --yes copilotkit@4.9.0 onboard read framework/deep-agents`
+8. LangGraph Python: `npx --yes copilotkit@4.9.0 onboard read framework/langgraph-python`
+9. LangGraph FastAPI: `npx --yes copilotkit@4.9.0 onboard read framework/langgraph-fastapi`
+10. LangGraph TypeScript: `npx --yes copilotkit@4.9.0 onboard read framework/langgraph-typescript`
+11. LlamaIndex: `npx --yes copilotkit@4.9.0 onboard read framework/llamaindex`
+12. ADK: `npx --yes copilotkit@4.9.0 onboard read framework/google-adk`
+13. Microsoft Agent Framework Python: `npx --yes copilotkit@4.9.0 onboard read framework/ms-agent-python`
+14. Microsoft Agent Framework .NET: `npx --yes copilotkit@4.9.0 onboard read framework/ms-agent-dotnet`
+15. Mastra: `npx --yes copilotkit@4.9.0 onboard read framework/mastra`
+16. MS Agent Harness .NET: `npx --yes copilotkit@4.9.0 onboard read framework/ms-agent-harness-dotnet`
+17. Pydantic AI: `npx --yes copilotkit@4.9.0 onboard read framework/pydantic-ai`
+18. Strands Agents Python: `npx --yes copilotkit@4.9.0 onboard read framework/strands-python`
+19. Strands Agents TypeScript: `npx --yes copilotkit@4.9.0 onboard read framework/strands-typescript`
 
 If the project has an agent in another framework, or no listed framework fits, keep the
 developer's current agent and run
-`npx copilotkit@4.8.4 onboard read unsupported/no-validated-path`.
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/fallback/best-effort.md

@@ -0,0 +1,29 @@
+# Build from the best available official documentation
+
+Tell the implementation subagent which plan steps lack direct documentation. Give it the
+approved plan, repository findings, and all official documentation that you found.
+
+Tell the subagent to preserve the existing project. It must not invent a CopilotKit API.
+It can combine compatible APIs from different official pages when one page omits the
+selected framework and frontend combination.
+
+Run the approved implementation and validation steps. Repair a failure in new work at most
+three times. Stop if the fix requires changes to the developer's existing agent or
+frontend.
+
+After validation passes, prove the runtime, agent round trip, and real frontend path. Use
+the same proof rules from the approved plan. Keep the development servers running after
+proof.
+
+Do not block core proof on CopilotKit Skills or MCP configuration. Try these tools after
+the application passes proof. Report each tool result separately.
+
+Report the documentation gap and each assumption with the proof evidence. Do not claim
+that the selected documentation proved an inferred step.
+
+When the proof is complete, run `npx --yes copilotkit@4.9.0 onboard complete`, carrying the
+visual-check result:
+
+```text
+npx --yes copilotkit@4.9.0 onboard complete --visual-check <outcome>
+```

package/onboarding/prompts/framework/ag2.md

@@ -0,0 +1,48 @@
+# Configure AG2
+
+Preserve an existing AG2 model setup. Start with OpenAI for a new agent. Offer another
+vendor only after an exact CopilotKit Markdown page names it. The page must name its
+credential variables and setup steps.
+
+Name the default model credential `OPENAI_API_KEY`.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/ag2/quickstart.md
+- https://docs.copilotkit.ai/ag2/inspector.md
+- https://docs.copilotkit.ai/ag2/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/ag2/readables.md
+
+Fetch all four pages in one step rather than one after another.
+
+The documentation scope is `ag2`. Use only the Python examples from these pages.
+Preserve the existing `AGUIStream` endpoint and its route. The quickstart uses a sample
+repository with `uv sync`. It does not name a dependency list for an existing project.
+Do not replace the project's dependency manifest with a guessed list, and do not change
+an existing endpoint to the quickstart's `/weather` sample route.
+
+The fixed-schema page has no AG2 backend schema-loading, render-operations, or runtime
+setup snippet. Do not invent a fixed-schema A2UI backend setup. If the requested work
+needs that missing setup, run the unsupported route below.
+
+The Readables page includes the AG2 backend context path. Use its `ContextVariables` and
+`AGUIStream` examples when the project needs frontend context. Do not substitute another
+framework's context setup.
+
+Record the selected framework, vendor, model, required credential variable names, these
+URLs, and the fixed-schema documentation gap. Do not use remembered CopilotKit
+instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/agno.md

@@ -0,0 +1,48 @@
+# Configure Agno
+
+Preserve an existing Agno model setup. Start with OpenAI for a new agent. Offer another
+vendor only after an exact CopilotKit Markdown page names it. The page must name its
+credential variables and setup steps.
+
+Name the default model credential `OPENAI_API_KEY`.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/agno/quickstart.md
+- https://docs.copilotkit.ai/agno/inspector.md
+- https://docs.copilotkit.ai/agno/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/agno/shared-state/agent-readonly.md
+
+Fetch all four pages in one step rather than one after another.
+
+The documentation scope is `agno`. Use only the Python agent examples from these pages.
+When the repository already has an Agno agent, use the quickstart's
+`Use an existing agent` path. Use `Start from scratch` only when the repository needs a
+new agent. For a new agent, install the documented Python dependencies
+`agno fastapi uvicorn openai ag-ui-protocol`. Preserve an existing `AgentOS` with the
+`AGUI` interface and its endpoint. In the documented runtime, `AgnoAgent` from
+`@ag-ui/agno` targets `/agui`. Do not replace that adapter with a generic agent or change
+the endpoint without repository evidence.
+
+The fixed-schema page has no Agno backend schema-loading, render-operations, or runtime
+setup snippet. The agent-readonly page does not bundle its Agno backend context setup.
+Do not invent either missing backend setup. If the requested work needs one of those
+setups and the project does not already have it, run the unsupported route below.
+
+Record the selected framework, vendor, model, required credential variable names, these
+URLs, the documented dependencies and adapter, and both documentation gaps. Do not use
+remembered CopilotKit instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/built-in.md

@@ -0,0 +1,43 @@
+# Configure the built-in CopilotKit agent
+
+Preserve an existing built-in agent model setup. Start with OpenAI for a new agent. Offer
+another vendor only after an exact CopilotKit Markdown page names it. The page must name
+its credential variables and setup steps.
+
+Name the default model credential `OPENAI_API_KEY`.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+The built-in agent runs in the CopilotKit Runtime process.
+Do not create or start a separate agent server.
+Keep the agent, model choice, tools, and runtime endpoint in the project's existing
+runtime setup.
+
+## Documentation
+
+- https://docs.copilotkit.ai/built-in-agent.md
+- https://docs.copilotkit.ai/quickstart.md
+- https://docs.copilotkit.ai/inspector.md
+- https://docs.copilotkit.ai/agent-app-context.md
+- https://docs.copilotkit.ai/model-selection.md
+- https://docs.copilotkit.ai/generative-ui/tool-based.md
+
+Fetch all six pages in one step rather than one after another.
+
+These pages use the root documentation scope. Do not add a framework path segment to
+their URLs. Use the tool-based generative UI page for this built-in route.
+
+Record the selected framework, vendor, model, required credential variable names, and
+these URLs. Do not use remembered CopilotKit instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/claude-sdk-python.md

@@ -0,0 +1,49 @@
+# Configure Claude Agent SDK for Python
+
+Preserve an existing Claude Agent SDK model setup. Start with Anthropic for a new agent.
+Offer another vendor only after an exact CopilotKit Markdown page names it. The page must
+name its credential variables and setup steps.
+
+Name the model credential `ANTHROPIC_API_KEY`. The documented model variable is
+`ANTHROPIC_MODEL`.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/claude-sdk-python/quickstart.md
+- https://docs.copilotkit.ai/claude-sdk-python/inspector.md
+- https://docs.copilotkit.ai/claude-sdk-python/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/reference/v2/hooks/useAgentContext.md
+
+Fetch all four pages in one step rather than one after another.
+
+The documentation scope is `claude-sdk-python`. For an existing agent, install the exact
+documented backend set with
+`uv add claude-agent-sdk ag-ui-claude-sdk ag-ui-protocol anthropic fastapi uvicorn python-dotenv`.
+Expose the agent through FastAPI and use `ClaudeAgentAdapter` from `ag_ui_claude_sdk`.
+Keep one long-lived adapter at module scope. Register the remote endpoint in Copilot Runtime
+with `HttpAgent`. Do not replace this transport with a different Claude or AG-UI adapter.
+
+For fixed-schema A2UI, keep the tool in the agent. Set `injectA2UITool` to `false` in the
+runtime. The documented Python setup does not support `action_handlers` on
+`a2ui.render`. Do not add them or claim that the fixed-schema button callback works.
+
+No Claude Agent SDK Python agent-app-context Markdown page is available. The reference
+page above documents the frontend hook only. Do not invent the missing agent-side context
+setup. If the requested work needs that setup, run the unsupported route below.
+
+Record the selected framework, vendor, model, required credential variable names, these
+URLs, and the context and A2UI limits. Do not use remembered CopilotKit instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/claude-sdk-typescript.md

@@ -0,0 +1,53 @@
+# Configure Claude Agent SDK for TypeScript
+
+Preserve an existing Claude Agent SDK model setup. Start with Anthropic for a new agent.
+Offer another vendor only after an exact CopilotKit Markdown page names it. The page must
+name its credential variables and setup steps.
+
+Name the model credential `ANTHROPIC_API_KEY`. The documented model variable is
+`CLAUDE_MODEL`.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/claude-sdk-typescript/quickstart.md
+- https://docs.copilotkit.ai/claude-sdk-typescript/inspector.md
+- https://docs.copilotkit.ai/claude-sdk-typescript/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/reference/v2/hooks/useAgentContext.md
+
+Fetch all four pages in one step rather than one after another.
+
+The documentation scope is `claude-sdk-typescript`. For an existing agent, install the
+documented packages, including `@anthropic-ai/claude-agent-sdk@^0.2.58`,
+`@anthropic-ai/sdk`, `@ag-ui/claude-agent-sdk`, `@ag-ui/core`, `@ag-ui/encoder`,
+`express`, `dotenv`, and `zod`. Keep the Express AG-UI endpoint. Use
+`ClaudeAgentAdapter` from `@ag-ui/claude-agent-sdk` in that endpoint and `HttpAgent` in
+Copilot Runtime. Do not replace either adapter.
+
+For fixed-schema A2UI, expose the backend tool through the in-process Claude SDK MCP
+server. Set `injectA2UITool` to `false` so the agent-owned tool is the only A2UI tool.
+The Claude Agent SDK adapter does not bridge CopilotKit frontend or runtime tools through
+AG-UI. If the requested path needs those tools, run the unsupported route below. Do not
+replace the Claude Agent SDK with the direct Anthropic Messages API fallback.
+
+The adapter appends AG-UI `input.context` entries to the Claude system prompt. Pair that
+backend path with the frontend hook from the reference page. Treat this as read-only app
+context. It does not provide an agent-to-frontend write-back path. Inspect the installed,
+pinned adapter source before using this path. If the package no longer handles
+`input.context`, run the unsupported route below.
+
+Record the selected framework, vendor, model, required credential variable names, these
+URLs, and the adapter and context limits. Do not use remembered CopilotKit instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/crewai-flows.md

@@ -0,0 +1,56 @@
+# Configure CrewAI Flows
+
+Preserve the project's existing CrewAI Flow, model, and deployment. The selected pages do
+not name a default model vendor or model credential variable. Do not choose a vendor or
+credential name for the developer. Ask only for a choice that the repository does not
+show and the documentation supports.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/crewai-crews/quickstart.md
+- https://docs.copilotkit.ai/crewai-crews/inspector.md
+- https://docs.copilotkit.ai/crewai-crews/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/crewai-crews/shared-state/in-app-agent-read.md
+
+Fetch all four pages in one step rather than one after another.
+
+The public onboarding route ID is `crewai-flows`. The canonical documentation scope is
+`crewai-crews`. The quickstart and feature pages also contain older `/crewai-flows/`
+links. Use the canonical Markdown URLs above and keep the project's CrewAI Flow name.
+
+The quickstart offers `Use the CopilotKit CLI (NextJS only)` and `Code along`. It assumes
+an existing Flow. Preserve the vendor, model, and credential variable that the project
+already uses. The new quickstart example uses GitHub Models. An existing OpenAI Flow can
+keep OpenAI and `OPENAI_API_KEY`. Do not change vendors to match the example.
+
+The current Python base adapter is `ag-ui-crewai==0.3.0`. Add the endpoint to the
+project's FastAPI app with
+`add_crewai_flow_fastapi_endpoint(app, ExistingFlow(), "/chat")`. Keep the existing Flow
+and its regular, non-conversational mode. Use the generic `HttpAgent` for the `/chat`
+endpoint. Do not invent a `CrewAIAgent` runtime class.
+
+The fixed-schema page extends this base adapter. Keep the fixed-schema tool on the CrewAI
+Flow path, call `copilotkit_emit_tool_result`, and set `injectA2UITool: false` in the
+runtime catalog. Do not replace the existing Flow with a Crew, Agent, or Task graph.
+
+The shared-state page is not an agent-app-context guide. Do not use it to invent a
+`useAgentContext` backend setup. If the requested work needs app context rather than the
+documented shared-state path, run the unsupported route below.
+
+Record the selected framework, project-defined vendor and model, repository-proven
+credential variable names, these URLs, the adapter, and both documentation limits. Do not
+use remembered CopilotKit instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/deep-agents.md

@@ -0,0 +1,45 @@
+# Configure Deep Agents
+
+Preserve an existing Deep Agents language and model setup. Start with OpenAI for a new
+agent. Offer another vendor only after an exact CopilotKit Markdown page names it. The
+page must name its credential variables and setup steps.
+
+Name the default model credential `OPENAI_API_KEY`.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/deepagents/quickstart.md
+- https://docs.copilotkit.ai/deepagents/inspector.md
+- https://docs.copilotkit.ai/deepagents/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/deepagents/shared-state/agent-readonly.md
+
+Fetch all four pages in one step rather than one after another.
+
+The documentation scope is `deepagents`. Do not use `deep-agents` in a documentation URL.
+The quickstart has Python and TypeScript tabs. Preserve the project's current language.
+Preserve the project's existing deployment path. Choose the same `Deep Agent` or
+`FastAPI` tab in both the quickstart and runtime setup. Do not change a Deep Agent
+deployment to FastAPI or a FastAPI deployment to Deep Agent.
+
+The fixed-schema A2UI backend example is Python-only. No Deep Agents agent-app-context
+Markdown page is available. Do not invent a TypeScript A2UI backend or an
+agent-app-context setup. If the requested work needs either missing setup, run the
+unsupported route below.
+
+Record the selected framework, language, vendor, model, required credential variable
+names, these URLs, and each documentation gap. Do not use remembered CopilotKit
+instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/google-adk.md

@@ -34,8 +34,13 @@ Record the selected framework, vendor, model, required credential variable names
 these URLs.
 Do not use remembered CopilotKit instructions.
 
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
 If the pages support the selection, run
-`npx copilotkit@4.8.4 onboard read frontend/plan`.
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
 
-If a page does not load or support the selection, run
-`npx copilotkit@4.8.4 onboard read unsupported/no-validated-path`.
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/langgraph-fastapi.md

@@ -0,0 +1,42 @@
+# Configure LangGraph FastAPI
+
+Preserve an existing LangGraph FastAPI model setup. Start with OpenAI for a new agent.
+Offer another vendor only after an exact CopilotKit Markdown page names it. The page must
+name its credential variables and setup steps.
+
+Keep `INTELLIGENCE_API_KEY` separate from model-vendor credentials. Do not read, show,
+store, or request a secret value.
+
+## Documentation
+
+- https://docs.copilotkit.ai/langgraph-fastapi/quickstart.md
+- https://docs.copilotkit.ai/langgraph-fastapi/inspector.md
+- https://docs.copilotkit.ai/langgraph-fastapi/generative-ui/a2ui/fixed-schema.md
+- https://docs.copilotkit.ai/langgraph-fastapi/agent-app-context.md
+
+Fetch all four pages in one step rather than one after another. Sequential fetching
+cost about nine minutes of the first proven journey.
+
+Read the FastAPI tab of the quickstart.
+Preserve an existing FastAPI transport and its persistence.
+The FastAPI graph must have a checkpointer.
+Keep the existing checkpointer. For a new local agent, use `MemorySaver` only for
+development.
+Use `add_langgraph_fastapi_endpoint` for the FastAPI agent endpoint.
+Use `LangGraphHttpAgent` in the Copilot Runtime.
+Do not copy the LangSmith path, its no-checkpointer rule, or its `LangGraphAgent` adapter.
+
+Record the selected framework, vendor, model, required credential variable names, and
+these URLs.
+Do not use remembered CopilotKit instructions.
+
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
+If the pages support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
+
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.

package/onboarding/prompts/framework/langgraph-python.md

@@ -21,8 +21,13 @@ Record the selected framework, vendor, model, required credential variable names
 these URLs.
 Do not use remembered CopilotKit instructions.
 
+A fetch tool that refuses a URL, or fails to reach it, reports a limit of the tool and
+not a fact about the page. Retrieve the same URL a second way before you judge it. Run
+`curl -fsSL <url>`, or read the same page without the `.md` suffix. Treat a page as
+unavailable only after a second method also fails.
+
 If the pages support the selection, run
-`npx copilotkit@4.8.4 onboard read frontend/plan`.
+`npx --yes copilotkit@4.9.0 onboard read frontend/plan`.
 
-If a page does not load or support the selection, run
-`npx copilotkit@4.8.4 onboard read unsupported/no-validated-path`.
+If a second method also fails, or a page that loads does not support the selection, run
+`npx --yes copilotkit@4.9.0 onboard read unsupported/no-validated-path`.