feat(MCP): Implement the base framework for a RediVL MCP server

[vishal-bala]

This PR implements the base framework for a RedisVL MCP server. In this iteration, the goal is to introduce the scaffolding for the MCP server to the extent that the server is runnable but does not have any meaningful functionality.

---

Note

Medium Risk
Adds a new MCP server surface that manages Redis index creation/validation, vectorizer instantiation, and request concurrency/timeouts; misconfiguration could affect startup behavior and Redis resources. Changes are mostly additive and gated behind an optional mcp extra, reducing impact on existing users.

Overview
Introduces a new optional redisvl.mcp package (enabled via the mcp extra) that provides a runnable MCP server scaffold (RedisVLMCPServer) with lifecycle management for an AsyncSearchIndex, lazy vectorizer construction, and guarded execution with concurrency + timeout limits.

Adds YAML-based MCP configuration loading (load_mcp_config) with ${VAR}/${VAR:-default} env substitution and Pydantic validation for runtime limits and schema field mappings, plus a small error contract (MCPErrorCode, RedisVLMCPError, map_exception) for consistent failure reporting.

Expands test coverage with unit tests for config/env substitution, settings, and error mapping, plus integration tests validating server startup/shutdown behavior (index creation modes, dims mismatch fail-fast, and cleanup on failures). Updates import-sanity to skip redisvl.mcp when optional extras aren’t installed.

^{Written by Cursor Bugbot for commit 8f98c5c. This will update automatically on new commits. Configure here.}

[jit-ci]

🛡️ Jit Security Scan Results

✅ No security findings were detected in this PR

---

^{Security scan by Jit}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cf0f0895a2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Disconnect index on startup exceptions

If vectorizer construction or dimension validation fails in this block (for example, a provider init error or a dims mismatch), startup() exits with an exception after AsyncSearchIndex has already opened a Redis client via exists()/create(), but there is no cleanup path that calls disconnect(). In retrying processes this leaks Redis connections across failed startups and can eventually exhaust connection limits; wrap startup initialization in failure cleanup (or call shutdown() in an exception handler) so the index client is always released.

Useful? React with 👍 / 👎.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

[cursor]

Vectorizer reference leaks when close fails during shutdown

Medium Severity

In shutdown(), self._vectorizer = None is placed after the await aclose() / close() calls. If the close operation raises, that line is never reached and self._vectorizer retains its reference. This contrasts with the index cleanup in the finally block, which correctly sets self._index = None before calling disconnect(). The inconsistency means a failed vectorizer close leaves the reference dangling — subsequent shutdown() retries will re-attempt the failing close, and get_vectorizer() could return a broken instance.