Redis MCP

# Fix list: Redis MCP

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/redis-mcp, the October 2026 research run, assessed 9 October 2026. Grade C, 55.7 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Redis MCP: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Agent ergonomics, 39 out of 100, up to 9.9 more on the total

Why it scored 39: 53 tools load at once, about 15,700 characters of descriptions, with no toolsets, dynamic loading or read-only subset. Pull request #161 for an allowlist has been open since 28 July 2026 (5 of 25). `scan_keys` pages with a cursor, `lrange` and `zrange` take ranges, `xrange` and `xreadgroup` take a count, `read_messages` caps at 100 and the vector tools take `k`. `scan_all_keys`, `hgetall`, `smembers` and `json_get` return everything with no limit (12 of 20). Failures come back as ordinary results, a string starting `Error` or a dictionary with an `error` key, carrying the Redis error text. Input checks return clear messages, such as the refusal of `block_ms=0`. Nothing marks a result as an MCP error and no error list is documented (9 of 20). No tool declares `readOnlyHint`, `destructiveHint` or `idempotentHint`, and pull request #181 adding them was opened on 8 October 2026. No retry guidance was found, and `lpush`, `xadd` and `publish` are not safe to repeat (3 of 20). Most tools need one or two arguments and defaults are stated in the descriptions. The server is a Python package only (10 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 2. Security & auth, 54 out of 100, up to 8.1 more on the total

Why it scored 54: The server holds one Redis username and password, which Redis ACL rules can scope to keys and command categories and an admin can revoke, or Entra ID credentials for Azure Managed Redis with token renewal. TLS certificate checks are required by default. The README's main example passes the password inside `--url` on the command line, and every caller shares the one user (22 of 30). The server has no read-only mode, no allowlist and no confirmation before `delete`, `json_del`, `xdel` or `xgroup_destroy`. The README gives an ACL example for a read-only user, and no tool runs an arbitrary command, so `FLUSHALL`, `CONFIG` and `EVAL` are out of reach (9 of 20). Stored values, pub/sub messages and docs search results reach the model as they are. No injection guidance was found in the README or the docs pages read (2 of 15). Logging goes to stderr at WARNING by default, so calls are not recorded unless the level is lowered. The connection names itself `redis-py(mcp-server_v0.5.1)`, which Redis shows in `CLIENT LIST` (6 of 15). Redis has a valid security.txt, a HackerOne disclosure programme, and SOC 2 Type 2 and ISO/IEC 27001 on its trust centre. CI runs Bandit and hardened runners, and Dependabot updates are merged within days. The repository has no SECURITY.md and no published advisories (15 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 3. Reliability, 62 out of 100, up to 7.6 more on the total

Why it scored 62: Scored as local software, since the owner runs it over stdio. Official PyPI package `redis-mcp-server` with Python 3.10 to 3.14 stated in the metadata, though the README badge says 3.14 and later (20). The CI workflow runs nightly and concluded in success on main on 9 October 2026, on a five-version Python matrix against a live Redis, with 347 test functions and an 80 per cent coverage floor (25). Fresh PyPI installs failed to start from 28 July to 5 August 2026 after `mcp` 2.0.0 removed the module the server imports. The fix merged on 31 July and reached PyPI five days later (issues #163 and #169). Issue #156, where the console command ignores `REDIS_HOST` and `REDIS_PORT`, has been open since 10 July 2026 with an acknowledgement and two unmerged fixes (13 of 25). Tagged releases with numbers in semantic form, but no changelog file, release notes that are lists of pull request titles, and a patch release (0.5.1) that changed the `unsubscribe` input and added six tools. Issue #137 asking for semver tags has no reply (4 of 15). Version 0.5.1 with the classifier `Development Status :: 4 - Beta` (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 4. Schema & documentation, 64 out of 100, up to 5.9 more on the total

Why it scored 64: Each tool is built from a typed Python signature through FastMCP, so inputs carry JSON Schema, read from the source since we do not run vendor software. `get_vector_from_hash` has no return type, several tools return either a value or an error string, and there are no output schemas (21 of 25). redis.io serves llms.txt and a Markdown form of each docs page, and the README is Markdown (10). Most descriptions are one line plus an argument list. `scan_keys`, `scan_all_keys`, `hybrid_search` and `xreadgroup` say when to use them and what to avoid, and the rest do not (11 of 20). Required fields and defaults are typed, but there are no enums (`distance_metric` and the `info` section are free strings) and the bounds on `block_ms`, `timeout_ms` and `max_messages` are checked in code, not declared in the schema (8 of 15). The README and docs pages give client configuration examples. Tool examples exist only inside a few descriptions, and no list of error responses was found (6 of 15). Tagged versions and GitHub release notes, with no changelog file. The 0.4.0 notes name the docs tool `search_documents` where the source has `search_redis_documents`, and Redis's docs page still links a Smithery install that the README dropped in 0.3.5 (8 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 5. Payments & pricing, 60 out of 100, up to 5 more on the total

Why it scored 60: Read with the self-hosted rule. The server is free under the MIT licence and needs no signup of its own, so 20 + 20 + 20. No x402, MPP or L402 (0). It needs a Redis database, which can be free Redis Open Source on the owner's machine. Redis Cloud prices were not checked. No hosted version of this server was found on the pages read.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 6. Maintenance & community, 57 out of 100, up to 3.8 more on the total

Why it scored 57: 0.5.1 was tagged on 4 August 2026 and published on GitHub and PyPI on 5 August, 65 days before the check (20). It is the only release in the 90 days before the check. The one before, 0.5.0, was on 16 March 2026 (0). 19 open items, 11 issues and 8 pull requests. Redis staff answered #156, #163 and #169 within weeks and merged the `mcp` pin the day it was reported, while #174, #137 and #31 have no reply and #45 on streamable HTTP has been open since August 2025 (14 of 25). Listed in the official MCP registry as `io.github.redis/mcp-redis`, a GitHub-verified namespace, with 0.5.1 published on 5 August 2026 and marked latest (15). Dependabot updates merged up to 2 October 2026, a lock file and a green nightly CI. The unbounded `mcp` dependency that broke installs in July is now pinned below 2 (8 of 10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 7. Transparency & trust, 76 out of 100, up to 2.1 more on the total

Made of editorial 65, provenance 87.

Why it scored 76: MIT (30). Local software that connects to the configured Redis. `search_redis_documents` sends each question to redis.io, which the README mentions through `MCP_DOCS_SEARCH_URL` without saying what is sent or kept. The Redis Privacy Policy, last updated 29 November 2024, does not address this software, and no telemetry code was found in the 0.5.1 source (18 of 30). No deprecation policy. The README warns that the main branch may contain breaking changes, and the `unsubscribe` change in 0.5.1 came with no notice (4 of 20). No usage statistics on the evidence of the source. The docs search call can be redirected or switched off with an empty `MCP_DOCS_SEARCH_URL`, which the README does not present as an opt-out (13 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Status page: not found (0 of 10)

## Deductions

Each comes off the total. A fixed and documented problem counts for less at the next check.

- -2: release 0.5.1, tagged 4 August 2026 as a patch release, changed the `unsubscribe` tool's input from `channel` to `subscription_id` and the return type of `subscribe` (pull request #133). The release notes give only the pull request title, 'Add stateful Redis pub/sub subscriptions with message polling'. The earlier tools opened a new pub/sub object on each call and so could not deliver messages, which is why the deduction is reduced (https://github.com/redis/mcp-redis/releases/tag/0.5.1).

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- The lead was right on the vendor, the repository and the docs page. It did not say the server is stdio only and marked beta
- unchecked: the PyPI project page and JSON API, which robots.txt disallows. The 0.5.1 date rests on the git tag (4 August 2026), the GitHub release and the MCP registry entry (both 5 August)
- unchecked: tool definitions as a client receives them from `tools/list`. We did not run the server, so input schemas and the absence of annotations are read from the source
- unchecked: the `mcp/redis` image on Docker Hub, which the README calls official. Who builds it and from which version was not read
- unchecked: what the docs search endpoint at `redis.io/convai/api/docs/search` logs or keeps. We sent it no request
- unchecked: Redis Cloud pricing and its free database, which matter only when the server points at Redis Cloud
- unchecked: the HackerOne programme page and whether mcp-redis is in scope
- unchecked: whether the redis organisation has a default SECURITY.md outside this repository. None is in the repository itself
- The trust centre answered a request for its robots.txt with the trust page itself, so it has no robots file. The compliance names were read from that one response
- No privacy or terms document addresses this software, so `provenance.privacy` and `provenance.terms` are both left out

## Weaknesses

- 53 tools load at once, with no toolsets and no read-only subset. Pull request #161 for a tool allowlist has been open since 28 July 2026
- No tool declares `readOnlyHint` or `destructiveHint`, and nothing asks for confirmation before `delete`, `json_del` or `xgroup_destroy`. Pull request #181 was opened on 8 October 2026
- Fresh PyPI installs failed to start from 28 July to 5 August 2026, because 0.5.0 had no upper bound on the `mcp` package (issues #163 and #169)
- Started through the `redis-mcp-server` command without flags, the server ignores `REDIS_HOST` and `REDIS_PORT` and connects to 127.0.0.1:6379. Issue #156 has been open since 10 July 2026
- No changelog file, no SECURITY.md and no prompt-injection guidance. Patch release 0.5.1 changed the `unsubscribe` input from `channel` to `subscription_id` without a note

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Pass the connection with `--url` or explicit flags. The `redis-mcp-server` command overrides `REDIS_HOST`, `REDIS_PORT`, `REDIS_DB`, `REDIS_SSL` and `REDIS_CLUSTER_MODE` with its own defaults when flags are absent
- Use `scan_keys` with the returned cursor until it is 0. `scan_all_keys`, `hgetall`, `smembers` and `json_get` return everything they match with no limit
- Read failures from the text. Tools return strings such as `Error ...` or an `error` key in place of an MCP error result
- Keep the `subscription_id` from `subscribe` or `psubscribe` and pass it to `read_messages` and `unsubscribe`. Idle subscriptions close after 300 seconds
- Connect as a Redis ACL user limited to the keys and command categories needed, since the server has no read-only switch and `delete` runs without confirmation

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.