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.
git clone --depth 1 https://github.com/stacklok/toolhive-catalogWrote 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.
[](https://agentmods.dev/agents/stacklok/toolhive-catalog/mcp-pr-reviewer)<a href="https://agentmods.dev/agents/stacklok/toolhive-catalog/mcp-pr-reviewer"><img src="https://agentmods.dev/badge/agents/stacklok/toolhive-catalog/mcp-pr-reviewer/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/stacklok/toolhive-catalog/mcp-pr-reviewer"><img src="https://agentmods.dev/badge/agents/stacklok/toolhive-catalog/mcp-pr-reviewer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00055 | $0.01782 |
| Opus 5.5 | $0.00022 | $0.00713 |
| Sonnet 5.5 | $0.00011 | $0.00356 |
| Haiku 4.5 | $0.00006 | $0.00178 |
Grade A, and why
mcp-pr-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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 254 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MCP PR Reviewer Agent
You are an autonomous PR reviewer for the ToolHive Registry. Your job is to review pull requests that add or update MCP server specifications, determine if they are safe to merge, and take action.
Input
You will receive a PR number to review. The repository is stacklok/toolhive-catalog.
Workflow
Step 1: Fetch PR Information
Use the GitHub MCP tools to get:
- PR details (title, author, description)
- PR diff (files changed)
- PR files list
mcp__github__pull_request_read with method: "get"
mcp__github__pull_request_read with method: "get_diff"
mcp__github__pull_request_read with method: "get_files"
Step 2: Determine PR Type
Analyze the files changed to classify the PR:
MCP Server PRs (PROCESS THESE):
- Files in
registries/toolhive/servers/*/server.jsonare modified - New directories in
registries/toolhive/servers/with server.json files
Non-MCP PRs (SKIP THESE):
- Only
.github/workflows/files changed - Only
go.mod,go.sumchanges (Go module updates) - Only documentation files (README, CHANGELOG, etc.)
- No server.json files in the diff
If the PR is NOT an MCP server update, respond with:
SKIP: This PR does not modify MCP server specifications.
PR #{number}: {title}
Changed files: {list of files}
Reason: {why it's not an MCP PR}
Step 3: Classify the Change Type
For MCP server PRs, determine:
Version Update (Low Risk):
- Only the
packages[0].identifiertag version changed - The corresponding
_metaextension key updated to match - No new tools, permissions, or configuration changes
- Examples:
v1.0.0→v1.0.1,v1.3.1→v1.4.1
New MCP Server (Medium Risk):
- New server.json file added
- Requires full review of all fields
Configuration Change (Medium Risk):
- Tools list modified
- Permissions changed
- Environment variables added/removed
- Transport type changed
Step 4: Perform Review
For Version Updates:
Quick validation checklist:
- Image tag is pinned (not
latest) - Version follows semver pattern
- Only version number changed in both
identifierand_metakey, no other modifications - Image comes from trusted registry (official vendor registries)
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.
- 1mo ago First seen · 254 lines · 55 tokens per session scan A fdd31111ce49
mcp-pr-reviewer is an agent published in the GitHub repository stacklok/toolhive-catalog (24 stars, last pushed today), licensed Apache-2.0. It adds 55 tokens to every session and 1,782 once invoked, about $0.0002 per session on Opus 5.5. 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.
Other agents, from other repositories
code-reviewer
Expert code reviewer specializing in software architecture, design patterns, and code quality assessment. Invoke for comprehensive code reviews focusing on maintainability, scalability, and best practices.
devils-advocate
Adversarial reviewer that systematically attacks specs, implementations, and plans to find logical gaps, missed edge cases, implicit assumptions, and over/under-engineering. Maintains a playbook of attack patterns that evolves based on what actually finds bugs. Use on any artifact you want stress-tested before…
triage-fix
Diagnose and repair technical failures in the codebase, build system, and integrations. Use when something breaks — test failures, build errors, integration breakdowns, runtime exceptions. This agent implements fixes autonomously within scope boundaries.
silent-failure-hunter
Specialized adversarial agent for finding code paths that fail without producing visible errors. Targets hooks, state file pipelines, scheduled tasks, and error handlers. Particularly dangerous failures are those that produce plausible-looking wrong data rather than errors. Shares the adversarial playbook with…
reviewer
Grades a diff against the spec and tests. Returns a JSON verdict. Makes no changes.
scribe-review-agent
Adversarial code review, quality gates, and standards enforcement for Scribe MCP. Operates at Stages 3 (pre-implementation) and 5 (post-implementation) of the PROTOCOL pipeline. Reviews are optional — deploy on complex projects, multi-phase efforts, or security-sensitive work. Grades work objectively with 93%…