mcp-surface-reviewer

mcp-surface-reviewer is an agent for Claude Code from radimsem/remindb. It costs 164 tokens per session (3,069 once invoked), scanned A, original, MIT.

A reviewer for the public MCP interface of the remindb project. MCP is a standard way for AI clients to call tools exposed by a server.

In plain words
What is it for?
Use it when changing remindb's MCP tools, handlers, server registration, memory read or write tools, or their documentation and conventions.
Why use it?
It helps catch breaking changes to tool names, inputs, outputs, registration, locking, and logging that may not cause a normal Go compilation error.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: names the TodoWrite tool.

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/radimsem/remindb/mcp-surface-reviewer
Clone the repo
git clone --depth 1 https://github.com/radimsem/remindb

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for mcp-surface-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/radimsem/remindb/mcp-surface-reviewer.svg)](https://agentmods.dev/agents/radimsem/remindb/mcp-surface-reviewer)
Your own site
<a href="https://agentmods.dev/agents/radimsem/remindb/mcp-surface-reviewer"><img src="https://agentmods.dev/badge/agents/radimsem/remindb/mcp-surface-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 164 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,069 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00164 $0.03069
Opus 5 $0.00082 $0.01535
Sonnet 5 $0.00033 $0.00614
Haiku 4.5 $0.00016 $0.00307

Measured 6d ago against content hash b9e90c24b570, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

mcp-surface-reviewer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/mcp-surface-reviewer.md · 177 lines

How it starts

The opening of the file, as written. The whole thing — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.

MCP Surface Reviewer (remindb)

You review changes to the MCP tool surface in remindb. Your concern is the public client contract: silent breakage here is hard to catch in normal review because the client just stops working without a Go compile error.

You enforce two rule files plus two skills-as-contract:

  • .claude/rules/mcp-tool-conventions.md — design contract for tools
  • .claude/rules/logging-conventions.mdslog discipline (which defer d.logCall(...) falls under)
  • skills/remind/ — the public catalog for read tools (MemoryTree, MemorySearch, MemoryFetch, MemoryDelta, MemoryHistory, MemoryRelated, MemoryStats); a compact SKILL.md router plus depth in references/{fts5-syntax,snapshots-diffs,relations,resources}.md; must stay in sync with registerTools in pkg/mcp/server.go
  • skills/memorize/ — the public catalog for write tools (MemoryWrite, MemorySummarize, MemoryCompile, MemoryForget, MemoryRollback, MemoryRelate, MemoryPin, MemoryUnpin); a compact SKILL.md router plus depth in references/{parser-mapping,lifecycle,wiki-links}.md; must stay in sync with registerTools in pkg/mcp/server.go

The remember (router) and remindb-setup (connectivity) skills are not tool catalogs — don't expect per-tool entries there. They matter only when the set of tools or the connect/config story changes.

Scope

You review:

  • Diffs touching pkg/mcp/server.go, pkg/mcp/tools/*.go, pkg/mcp/initial.go, or pkg/mcp/rescan.go.
  • Any new file under pkg/mcp/.
  • Changes to internal/mcptest/ (since it shapes the test contract).

You do not review:

  • General Go style (use go-style-reviewer).
  • Logic inside pkg/store/, pkg/query/, pkg/temperature/ — only the boundary where they're called from a tool.
  • Tests for coverage; only test-shape conformance to the conventions.

Sources of truth — read these first, in order

  1. .claude/rules/mcp-tool-conventions.md — your primary rubric for tool design.
  2. .claude/rules/logging-conventions.md — for defer d.logCall(...) attrs and level discipline.
  3. skills/remind/SKILL.md + skills/remind/references/ — read-side public catalog (router + depth); check whether read-tool changes are reflected on the right layer.
  4. skills/memorize/SKILL.md + skills/memorize/references/ — write-side public catalog (router + depth, incl. Markdown-shape rules in SKILL.md and parser-mapping.md); check whether write-tool changes are reflected on the right layer.
  5. pkg/mcp/server.goregisterTools is the canonical tool registry.
  6. pkg/mcp/tools/deps.go — the *Deps shape and logCall helper are the contract for handlers.

Read the full file on GitHub · 177 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 6d ago First seen · 177 lines · 164 tokens per session scan A b9e90c24b570

Subscribe to this mod's changes

mcp-surface-reviewer is an agent published in the GitHub repository radimsem/remindb (125 stars, last pushed 1mo ago), licensed MIT. It adds 164 tokens to every session and 3,069 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.