What is image_mcp AGENTS.md?

A set of coding-agent instructions for the karlcc/image_mcp repository, covering its commands, architecture, source files, and data flow. The project is an MCP server that sends image files to an OpenAI-compatible vision service.

Use it when developing, testing, linting, or configuring the image MCP server, including its standard and HTTP modes.

It explains how the repository is organized and which checks developers can run, including tests and type checks.

How karlcc/image_mcp configures Codex and OpenCode

Reuse

Borrowing it

Nothing to install: this file belongs to karlcc/image_mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/karlcc/image_mcp/main/AGENTS.md

Clone the repo

git clone --depth 1 https://github.com/karlcc/image_mcp

Made for: Codex, OpenCode.

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/karlcc/image_mcp/agents-md/github.svg)](https://agentmods.dev/instructions/karlcc/image_mcp/agents-md)

Your own site

<a href="https://agentmods.dev/instructions/karlcc/image_mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/karlcc/image_mcp/agents-md/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.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/karlcc/image_mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/karlcc/image_mcp/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 1,446 This file is loaded in full into every session.

When invoked 1,446 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin fork From a forked repository.

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.

Model	Per session	Once invoked
Fable 5.1	$0.01446	$0.01446
Opus 5.5	$0.00578	$0.00578
Sonnet 5.5	$0.00289	$0.00289
Haiku 4.5	$0.00145	$0.00145

Measured 2mo ago against content hash 285112d95907, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

image_mcp AGENTS.md 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.

AGENTS.md · 67 lines

How it starts

The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Commands

npm run build — Compile TypeScript to build/ (runs tsc then chmod index.js as executable)
npm run dev — Dev mode with tsx watch (hot reload)
npm test — Run full Jest suite
npm run test:watch — Jest in watch mode
npx jest tests/basic.test.ts — Run a single test file
npm run lint / npm run lint:fix — ESLint on src/
npm run type-check — tsc --noEmit
npm run test:smoke — Vision smoke tests (requires IMAGE_MCP_SMOKE=1 + live API config)
npm run preflight — Lint + type-check + smoke tests (pre-release gate)
npm run benchmark:models — Accuracy benchmark harness (scripts/benchmark-models.mjs, uses MCP SDK client). Requires ~/.config/image_mcp/model_candidates.json with a models array.

Architecture

MCP server that proxies image files to an OpenAI-compatible vision endpoint. Two transport modes: stdio (default) and HTTP/SSE (enabled via --http or MCP_USE_HTTP=true).

Source files

File	Role
`src/config.ts`	`ConfigManager` singleton. Parses CLI args (commander) + env vars + defaults via zod schema. Config precedence: CLI > env > defaults. Persistent config saved to `~/.config/image_mcp/config.json`. Supports `--reasoning-effort` / `OPENAI_REASONING_EFFORT` (values: none, minimal, low, medium, high, xhigh).
`src/image-processor.ts`	`ImageProcessor` static class. Normalizes input (`@`-prefix shorthand, `file://` stripping), detects input type (file path / HTTP URL / data URL / raw base64), reads and converts all inputs to base64 data URLs for the API. Validates MIME type and 10MB size limit.
`src/openai-client.ts`	`OpenAIClient` class. Axios-based client with exponential-backoff retry (1s→2s→4s→..., capped 30s). Supports streaming and non-streaming chat completions. Retries on network errors, 5xx, and 429.
`src/index.ts`	Server entrypoint. Registers 3 MCP tools and their handlers. `dispatchToolCall` switch routes by tool name. Stdio and HTTP/SSE transport setup. `extractMessageText()` handles thinking models (reads `reasoning_content`/`reasoning` when `content` is empty).
`src/vision-response.ts`	Vision guard system — `buildVisionGuardPrompt()` appends anti-hallucination instructions to user prompts; `assertVisionResponse()` validates API responses don't contain non-vision text patterns. `stripGrokAssetUrls()` removes hosted asset URLs that some gateways (e.g. grok2api) append to responses.
`src/vision-probe.ts`	Runtime probe to detect whether the configured model supports vision/image inputs.

Read the full file on GitHub · 67 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.

2mo ago First seen · 67 lines · 1,446 tokens per session scan A 285112d95907

Subscribe to this mod's changes

image_mcp AGENTS.md is an instructions file published in the GitHub repository karlcc/image_mcp (0 stars, last pushed 5mo ago), licensed MIT. It adds 1,446 tokens to every session, about $0.0058 per session on Opus 5.5. A static security scan graded it A with 0 findings. It comes from a forked repository.

image_mcp AGENTS.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
image_mcp AGENTS.mdkarlcc/image_mcp	Instructions file	0	5mo ago	MIT	A	1,446	Codex, OpenCode
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	9,719	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	yesterday	MIT	A	7,461	Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	yesterday	MIT	A	4,680	Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	5,516	GitHub Copilot
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	3,093	GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness	Instructions file	274k	2d ago	MIT	A	4,313	Codex, OpenCode

image_mcp: Instructions file for Codex