Repository navigation
Integrations: ADK/LlamaIndex/AutoGen backends, 7 client setup pages; test cleanup; API ref index - #108
Merged
Conversation
test_apply.py's category-and-event test repeated test_slots.py's (same setup and assertions; the slots version also covers alias folding). The always-skipped Postgres placeholder in test_db.py had no body; the real Postgres suite is tests/test_postgres.py. Coverage unchanged at 87%.
The Memory page 404s on the live site although GitBook imported it, lists it in the sitemap and logs no finding; the other four tool pages, with the same schema shapes, render. It was the group's first page, which the group URL redirects to. The new first page is a generated index of all 32 tools with their capability, and has no OpenAPI blocks. Unverified until it syncs from main.
…ntinue, Codex CLI and Gemini CLI Each page gives the client's MCP config for Vayl, a health check, and says how it was checked. Codex CLI 0.163.0 (add, list, Vayl chosen unprompted, writes need approval) and Gemini CLI 0.65.0 (add with a $OPENAI_API_KEY reference, list: Connected) were tested; code --add-mcp was tested with VS Code 1.142. Windsurf, Zed, Cline and Continue configs follow each client's official docs and are marked as not run.
…toGen - vayl.integrations.adk.VaylMemoryService: an ADK BaseMemoryService. Session events, event deltas and explicit entries go to remember (user turns only by default, each event once); search_memory answers from recall. ADK's app_name maps to Vayl's agent_id. - vayl.integrations.llamaindex.VaylMemory: block() is a BaseMemoryBlock (flushed messages -> remember, latest user message -> recall into the prompt) and tools() gives FunctionTools. - vayl.integrations.autogen.VaylMemory: autogen-core Memory (add -> remember, query and update_context -> recall). clear() raises instead of calling delete_all. All three ran end to end against vayl-mcp with gpt-6-luna. Pydantic AI (MCPToolset with include_instructions) and Agno (MCPTools with OpenAIResponses) need no adapter and are documented with tested examples. Extras [adk], [llamaindex], [autogen]; llama-index-core and autogen-core join the dev extra so CI runs their tests (pip-audit clean); mypy skips numpy's 3.12-only stubs that llama-index pulls in.
One table for the agent frameworks (install command and import, marked memory backend or tools) and one line of links for the MCP clients' setup pages. The new client pages go live when this branch reaches main.
Merged
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.
Memory backends (memory runs on every turn, no tool call needed):
vayl.integrations.adk.VaylMemoryService: Google ADKBaseMemoryService(user turns →remember, each event once;search_memory→recall)vayl.integrations.llamaindex.VaylMemory(...).block(): LlamaIndex memory block, plustools()vayl.integrations.autogen.VaylMemory: autogen-coreMemory;clear()raises instead of callingdelete_all[adk],[llamaindex],[autogen]; llama-index-core and autogen-core added todevso CI runs their tests (pip-audit clean); mypy skips numpy's 3.12-only stubsMCP, no adapter: tested examples for Pydantic AI (
MCPToolset,include_instructions=True) and Agno (MCPTools+OpenAIResponses).All five ran end to end against vayl-mcp with gpt-6-luna (Enterprise → Pro, then now/before).
Client setup pages: VS Code, Windsurf, Zed, Cline, Continue, Codex CLI, Gemini CLI. Codex CLI 0.163 and Gemini CLI 0.65 tested;
code --add-mcptested on VS Code 1.142; the rest follow official docs and are marked as not run.README: install table for every integration.
Also: