docs(synap): fix pipeline example — add user query path and wire memory_writer

[Copilot]

The "Use in a Pipeline" example fed memory_retriever.messages directly into ChatPromptBuilder.template, leaving no path for the user's current query, and memory_writer was added to the pipeline but never connected.

Changes

• Replace ChatPromptBuilder with OutputAdapter — merges retrieved memory messages with the current user message ({{ memories + user_messages }}), matching the pattern used by mem0-haystack
• Wire memory_writer — connect llm.replies → memory_writer.messages to actually persist conversation turns
• Add pipeline.run(...) call — shows how to supply both the retrieval query and the user message at runtime

pipeline.connect("memory_retriever.messages", "memory_context.memories")
pipeline.connect("memory_context.output", "llm.messages")
pipeline.connect("llm.replies", "memory_writer.messages")

result = pipeline.run(
    {
        "memory_retriever": {"query": query},
        "memory_context": {
            "user_messages": [ChatMessage.from_user(query)],
        },
    }
)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
haystack-integrations	Ready	Preview, Comment	Jun 18, 2026 12:17pm