{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_k9namas58gcr","handle":"mcp-gpt-image","url":"https://wellknown.network/agents/mcp-gpt-image","links":{"self":"https://wellknown.network/agents/mcp-gpt-image/record.json","html":"https://wellknown.network/agents/mcp-gpt-image","markdown":"https://wellknown.network/agents/mcp-gpt-image/record.md","api":"https://wellknown.network/api/v1/agents/mcp-gpt-image","status":"https://wellknown.network/api/v1/agents/mcp-gpt-image/status","claim":"https://wellknown.network/agents/mcp-gpt-image/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/mcp-gpt-image/claim.json","badge":"https://wellknown.network/agents/mcp-gpt-image/badge.svg","openapi":"https://wellknown.network/openapi.json","history":"https://wellknown.network/api/v1/agents/mcp-gpt-image/history","tools":"https://wellknown.network/api/v1/agents/mcp-gpt-image/tools"},"ard":{"identifier":"urn:air::server:mcp-gpt-image","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"mcp-gpt-image","summary":"MCP server for image generation, editing and multi-turn refinement with OpenAI gpt-image-2.5","description":"# 🎨 mcp-gpt-image\n\n**An image MCP server built for the second draft, not the first.**\n\nGetting one picture out of a model is the easy part. The afternoon goes on the twelve that\ncome after it: the sky is too bright, the logo drifted, the text on the sign came out wrong.\nMost image connectors stop at the first picture and leave you to redo the loop by hand.\n\nThree tools, OpenAI's\n[gpt-image-2.5](https://developers.openai.com/api/docs/guides/image-generation) in both of its\nvariants, and file paths instead of base64.\n\n## 🧱 The problem\n\nYou write a prompt, you get a picture, you want the sky darker. So you write the whole prompt\nagain with \"darker sky\" appended, because the model kept nothing from the first call. Every\ncorrection is a full rewrite, and the model has no idea what it just drew.\n\nWhen the connector hands images back as base64 it gets worse. The picture is now sitting in\nthe conversation: it cost tokens to arrive, it costs tokens again on every turn it stays\nthere, and feeding it into the next step means sending those same bytes back up. Two rounds\nof that on a 1536x1024 image and the context window is mostly pixels.\n\nThis server takes the other route on both counts. Images travel as **file paths**:\n`generate_image` writes a file and returns where it is, `edit_image` and `refine_image` take\npaths as input, so a picture moves from one step to the next without ever entering the\nconversation. And `refine_image` **holds the thread** with the model, so turn one is a full\ndescription and turn two is \"make the sky darker\". Nothing is re-uploaded, nothing is\nre-described.\n\n## 📦 Install\n\n```bash\nclaude mcp add gpt-image -e OPENAI_API_KEY=sk-... -- uvx mcp-gpt-image\n```\n\nOr in any MCP client's config:\n\n```json\n{\n  \"mcpServers\": {\n    \"gpt-image\": {\n      \"command\": \"uvx\",\n      \"args\": [\"mcp-gpt-image\"],\n      \"env\": {\n        \"OPENAI_API_KEY\": \"sk-...\",\n        \"GPT_IMAGE_OUTPUT_DIR\": \"/home/you/images\"\n      }\n    }\n  }\n}\n```\n\nNeeds Python 3.12 or newer, [uv](…","publisher":{"name":"Fabien Vauchelles","url":null},"homepage":"https://github.com/agelyhq/mcp-gpt-image","repository":"https://github.com/agelyhq/mcp-gpt-image/blob/main/CHANGELOG.md","version":"0.4.0","license":null,"protocols":["mcp"],"tags":["gpt-image-2.5","image-editing","image-generation","mcp","model-context-protocol","openai"],"pricing":null,"endpoints":[{"url":"pypi:mcp-gpt-image","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","repoUrl":"pypi","summary":"pypi","version":"pypi","description":"pypi","homepageUrl":"pypi","publisherName":"pypi"}},"derived":{"capabilities":[{"slug":"content.writing","name":"Writing & Editing","confidence":1,"provenance":"derived"},{"slug":"media.image-generation","name":"Image Generation","confidence":1,"provenance":"declared"},{"slug":"media.image-editing","name":"Image Editing","confidence":1,"provenance":"declared"}],"categories":["content","media"],"language":"en"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"mcp-gpt-image","registry":"pypi","observedAt":"2026-10-22T09:23:33.268Z","publishedAt":"2026-09-16T12:11:23.039080Z","latestVersion":"0.4.0"},"toolSurface":null,"endpointFacts":[]},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"mcp-gpt-image","url":"https://pypi.org/project/mcp-gpt-image/","firstSeenAt":"2026-09-10T01:21:27.080Z","fetchedAt":"2026-10-22T09:21:59.334Z","normalizedAt":"2026-10-22T09:21:59.334Z"}]},"firstSeenAt":"2026-09-10T01:21:27.080Z","updatedAt":"2026-10-22T09:23:33.268Z"}