PostHog MCP 0.10.8 keeps tool retries in the same conversation

0.10.8

Patch Changes

• #4433 c514a34 Thanks @gesh! - Deliver the conversation_id session handle on errored tool results, and inject it into the virtual get_more_tools tool. A first call that fails no longer sends the agent's retry into a different conversation, and a reported capability gap now groups with the work that hit it. (2026-08-06)

• #4428 7322893 Thanks @gesh! - Use an agent-supplied conversation_id as the session anchor, so tool calls in one conversation share a $session_id across reconnects, restarts, and per-request server instances. Only a handle the SDK could have minted is accepted; a value the agent invented is replaced with a fresh one, so two callers cannot land in the same session by sending the same string. (2026-08-06)

• #4431 955df8d Thanks @gesh! - Mirror the conversation_id session handle into structuredContent for tools whose output schema declares _mcp_instructions. Clients that read structured results never saw the handle in content, so correlation for those tools was zero. (2026-08-06)

• #4430 e6d9295 Thanks @gesh! - Declare an optional _mcp_instructions property on the output schema of tools that advertise one, when enableConversationId is on. Inert by itself — it is the schema declaration that makes a later change able to mirror the conversation handle into structuredContent without failing client-side validation. (2026-08-06)

• Updated dependencies [4751b33, 64ba193]:

  • posthog-node@5.48.1
  • @posthog/core@1.46.9