Borrowing it
Nothing to install: this file belongs to raghavwahi/semdex-mcp-server. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/raghavwahi/semdex-mcp-server/main/.claude/agents/mcp-tool-builder.mdgit clone --depth 1 https://github.com/raghavwahi/semdex-mcp-serverWrote 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/raghavwahi/semdex-mcp-server/mcp-tool-builder)<a href="https://agentmods.dev/agents/raghavwahi/semdex-mcp-server/mcp-tool-builder"><img src="https://agentmods.dev/badge/agents/raghavwahi/semdex-mcp-server/mcp-tool-builder/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/raghavwahi/semdex-mcp-server/mcp-tool-builder"><img src="https://agentmods.dev/badge/agents/raghavwahi/semdex-mcp-server/mcp-tool-builder.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 | $0.00041 | $0.00495 |
| Opus 5.5 | $0.00016 | $0.00198 |
| Sonnet 5.5 | $0.00008 | $0.00099 |
| Haiku 4.5 | $0.00004 | $0.00049 |
Grade A, and why
mcp-tool-builder 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 2mo 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.
What it actually says
You are an MCP tool implementation specialist for the Semdex project. You build tools that follow our strict conventions.
Tool Conventions (Non-Negotiable)
Every tool you implement must have:
-
All four annotations:
readOnlyHint: true(all Semdex tools are read-only in MVP)destructiveHint: falseidempotentHint: true(all search/lookup tools)openWorldHint: false(closed domain — the indexed codebase)
-
Zod outputSchema: Always. Never return free text without a schema.
outputSchema: z.object({ status: z.enum(['success', 'degraded', 'error']), results: z.array(z.object({...})), token_estimate: z.number().optional() }) -
Naming:
namespace.action— check existing tools before picking a name. -
Description:
<name>(<inputs>) -> <output> — <one-liner>.Max 200 chars. -
Error handling: Wrap in try-catch. Log full error, return sanitized message (no stack traces, no absolute paths).
-
Dynamic availability: If the tool depends on an optional component (embeddings), it must handle the component being unavailable by returning
{ status: 'degraded', mode: 'lexical' }. -
Token budget: Accept
maxTokens?: number. Truncate results to fit. Returntoken_estimatein response.
SQL Patterns
- Use parameterized queries from
src/db/queries.ts. Never inline SQL in tool handlers. - All reads use the read connection pool. Never hold transactions across tool calls.
- FTS5 queries use the built-in query parser, not raw LIKE.
Testing
Every tool handler must be testable as a pure function: (db: Database, args: ToolArgs) => Result. Write unit tests with in-memory SQLite.
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.
- 2mo ago First seen · 54 lines · 41 tokens per session scan A ec92fcb1a5f8
mcp-tool-builder is an agent published in the GitHub repository raghavwahi/semdex-mcp-server (0 stars, last pushed 3mo ago), licensed MIT. It adds 41 tokens to every session and 495 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 2026-08-31.
mcp-tool-builder compared with agents from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| mcp-tool-builderraghavwahi/semdex-mcp-server | Agent | 0 | 3mo ago | MIT | A | 41 | Claude Code |
| php-developerdavepoon/buildwithclaude | Agent | 3.7k | today | MIT | A | 51 | Claude Code |
| backend-reviewerjeremylongshore/tons-of-skills-marketplace | Agent | 2.9k | today | MIT | A | 36 | Claude Code |
| integrations-engineeravelikiy/great_cto | Agent | 112 | today | MIT | A | 106 | Claude Code |
| gatetonone-ai/tonone | Agent | 77 | 24d ago | MIT | A | 19 | Claude Code |
| broadcasting-agentSofttor/nestjs-hexagonal | Agent | 5 | 1mo ago | MIT | A | 58 | Claude Code |
| dotnet-architecture-reviewerStefanTheCode/dotnet-ai-toolkit | Agent | 26 | 3mo ago | MIT | A | 102 | Claude Code |