Token-efficient Reason MCP v2

[IsraelAraujo70]

What changed

• adds compact response envelopes and bounded projected page reads
• adds exact ordered block queries with signed cursors
• adds compact canonical operation batches with deterministic IDs and local references
• filters tool discovery by token scopes and returns structured correctable errors
• configures MCP cursor signing and documents the compatibility contract

Why

The existing MCP duplicated results, returned full trees, relied on semantic search for deterministic discovery, and required verbose operation metadata. This keeps legacy clients working while reducing agent context usage and preserving the canonical operation engine, authorization, workspace isolation, idempotency, ordering, and LWW semantics.

Validation

• make test
• cargo fmt
• git diff --check
• docker compose config --quiet
• local PostgreSQL and MCP acceptance for exact ordering/filtering, cursor continuation and invalidation, trash visibility, workspace grants, canonical replay/conflict behavior, and compact snapshots
• OpenCode 1.18.10 connected to the local MCP endpoint

Production MCP_CURSOR_SIGNING_KEY configuration and deployment are intentionally not included and still require separate confirmation.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.