Image Generation MCP

Multi-provider image generation MCP server built on FastMCP. Generate images from Claude Desktop, Claude Code, or any MCP client using OpenAI, Google Gemini, Stable Diffusion (SD WebUI), or a zero-cost placeholder provider.

Documentation | PyPI | Docker

Features

• Multi-provider — OpenAI (gpt-image-1.5, gpt-image-1, dall-e-3), Google Gemini (gemini-2.5-flash-image, gemini-3.x previews), SD WebUI (Stable Diffusion / Forge / reForge), and a zero-cost placeholder for testing.
• Per-model style metadata — every model carries a style_profile (strengths, prompt grammar, lifecycle); list_providers includes a top-level warnings array for deprecated models. See Model Catalog.
• Keyword-based auto-selection — provider="auto" routes by prompt content (text/logo → OpenAI, photoreal/anime → SD WebUI, draft → placeholder).
• CDN-style image transforms — image://{id}/view?format=webp&width=512&crop_x=... resizes / re-encodes / crops on demand without re-generating.
• Hybrid background tasks — long-running SD generations run with task=True (poll for status); short OpenAI calls stream progress in the foreground.
• MCP Apps gallery + viewer — interactive UI surfaces (browse generated images, edit / crop / rotate) for clients that support app: resources.
• Production deployment — Docker (multi-arch), .deb/.rpm with hardened systemd, OIDC + bearer auth, persistent EventStore for HTTP session resumability.

What you can do with it

With this server mounted in an MCP client, you can ask:

• "Generate a coffee mug product photo on a worn oak table, 16:9, no text." Routes to gpt-image-1.5 for typography-aware photorealism.
• "Create three concept-art variations of a cyberpunk alley at dusk." Composes generate_image with provider="sd_webui" and a stylised checkpoint like dreamshaperXL.
• "Crop this image to a 1:1 square centred on the subject and resize to 512px." Uses image://{id}/view?width=512&height=512&crop_x=... resource transforms.
• "Show me my recent generations." Browses the gallery via the image://list resource and the MCP Apps gallery viewer.
• "Save this style as 'cyberpunk-night' so I can apply it to future requests." Uses the style library — markdown briefs the LLM interprets per-provider.

Installation

From PyPI

pip install image-generation-mcp

If you add optional extras via the PROJECT-EXTRAS-START / PROJECT-EXTRAS-END sentinels in pyproject.toml, document them below:

Extra	Includes	Use when
mcp	fastmcp[tasks]>=3.0,<4	Background-task support (task=True) — required for long SD generations.
openai	openai>=1.0	Enables the OpenAI provider.
google-genai	google-genai>=1.0	Enables the Gemini provider.
all	fastmcp[tasks] + openai + google-genai	Everything except SD WebUI (which is HTTP-only — no extra needed).

Example: pip install image-generation-mcp[all].

From source

git clone https://github.com/pvliesdonk/image-generation-mcp.git
cd image-generation-mcp
uv sync --all-extras --all-groups

Docker

docker pull ghcr.io/pvliesdonk/image-generation-mcp:latest

A compose.yml ships at the repo root as a starting point — copy .env.example to .env, edit, and docker compose up -d.

To attach a remote Python debugger (development only — the protocol is unauthenticated), see Remote debugging.

Linux packages (.deb / .rpm)

Download .deb or .rpm packages from the GitHub Releases page. Both install a hardened systemd unit; env configuration is sourced from /etc/image-generation-mcp/env (copy from the shipped /etc/image-generation-mcp/env.example).

Claude Desktop (.mcpb bundle)

Download the .mcpb bundle from the GitHub Releases page and double-click to install, or run:

mcpb install image-generation-mcp-<version>.mcpb

Claude Desktop prompts for required env vars via a GUI wizard — no manual JSON editing needed.

For manual Claude Desktop configuration and setup options, see Claude Desktop deployment.

Quick start

image-generation-mcp serve                                # stdio transport
image-generation-mcp serve --transport http --port 8000   # streamable HTTP

For library usage (embedding the domain logic without the MCP transport), import from the image_generation_mcp package directly — see the project's domain modules under src/image_generation_mcp/ for entry points.

Server info

The server registers a built-in get_server_info tool (via fastmcp_pvl_core.register_server_info_tool) so operators can confirm the deployed version with a single MCP call. The default response carries server_name, server_version, and core_version. Servers that talk to a remote upstream wire upstream version reporting inside the DOMAIN-UPSTREAM-START / DOMAIN-UPSTREAM-END sentinel in src/image_generation_mcp/server.py — see CLAUDE.md for the wiring pattern.

Configuration

Core environment variables shared across all fastmcp-pvl-core-based services:

Variable	Default	Description
FASTMCP_LOG_LEVEL	INFO	Log level for FastMCP internals and app loggers (DEBUG / INFO / WARNING / ERROR). The -v CLI flag overrides to DEBUG.
FASTMCP_ENABLE_RICH_LOGGING	true	Set to false for plain / structured JSON log output.
IMAGE_GENERATION_MCP_KV_STORE_URL	file:///data/state	Persistent-state backend URL for pvl-core subsystems — file:///path (survives restarts), memory:// (dev/ephemeral).

Domain-specific variables go below under Domain configuration.

Authorization (opt-in)

This server inherits opt-in per-subject authorization from fastmcp-pvl-core. The default posture is off — every authenticated caller can use every tool, resource, and prompt. Turn it on by pointing IMAGE_GENERATION_MCP_ACL_PATH at a TOML ACL file; the middleware is installed only when the path is set, and individual tools opt in by declaring meta={"required_scope": "<scope>"} at registration. A tool without required_scope is unrestricted regardless of caller.

Wire it in by uncommenting the acl_path field in src/image_generation_mcp/config.py and the AuthorizationMiddleware stanza in src/image_generation_mcp/server.py — both ship as commented stubs in the scaffold.

ACL TOML schema

[subjects]
"user:alice@example.com" = ["read", "write"]
"user:admin@example.com" = ["*"]              # wildcard — any required scope passes
"service:ci-bot"         = ["read"]
"local"                  = ["*"]              # stdio mode subject

• Subject strings are opaque. The <kind>:<id> convention is documentation only; the library treats each subject as a literal string.
• * is the only library-treated special scope — it grants every required scope. Subject-side wildcards (* as an ACL key) are rejected at load time.
• Scope vocabulary is domain-defined. Per-project or per-folder gating is encoded into the scope string itself (e.g. read:project-foo, write:vault/personal); fastmcp-pvl-core treats every scope except * as opaque.

Subject ↔ bearer-token alignment

The subject string used as a value in the bearer-tokens TOML (IMAGE_GENERATION_MCP_BEARER_TOKENS_FILE) is the same string used as a key in the ACL TOML. Same string, opposite roles — keep the two files consistent when adding or removing a principal. See Mapped bearer tokens in the authentication guide for the bearer-tokens TOML schema.

In single-token mode (IMAGE_GENERATION_MCP_BEARER_TOKEN) every authenticated caller shares one subject — the library's default (currently "bearer-anon"), override with IMAGE_GENERATION_MCP_BEARER_DEFAULT_SUBJECT; reference that string as the ACL key. In stdio mode the subject is the literal "local".

Load semantics

The ACL file is loaded once at server startup. Restart the server to pick up changes; live reload is not part of the initial implementation. load_acl fails fast with ConfigurationError on every malformed condition, so a typo in the ACL file aborts startup rather than silently denying requests.

Privacy default

Denied requests are logged at WARNING with the subject string for audit attribution. The wire-side error payload omits the subject by default to limit cross-user information disclosure. For internal-only servers where the subject is safe to surface to clients, construct the middleware with AuthorizationMiddleware(..., expose_subject_in_error=True).

See also

• fastmcp-pvl-core README — Authorization — full design, the check_authorization per-call helper, and per-token subject mapping.
• Authorization submodule spec — design rationale and deviations table.

Post-scaffold checklist

After copier copy and gh repo create --push:

1. Fill in the DOMAIN blocks in this README (Features, What you can do with it, Domain configuration, Key design decisions) and in CLAUDE.md.
2. Configure GitHub secrets — see below.
3. Install dev + docs tooling: uv sync --all-extras --all-groups.
4. Install pre-commit hooks: uv run pre-commit install.
5. Run the gate locally: uv run pytest -x -q && uv run ruff check --fix . && uv run ruff format . && uv run mypy src/ tests/.
6. Push the first commit — CI should be green.

GitHub secrets

CI workflows reference three repository secrets. Configure them via Settings → Secrets and variables → Actions or with gh secret set:

Secret	Used by	How to generate
RELEASE_TOKEN	release.yml, copier-update.yml	Fine-grained PAT at https://github.com/settings/personal-access-tokens/new with contents: write and pull_requests: write (the copier-update cron opens PRs). Scoped to this repo.
CODECOV_TOKEN	ci.yml	https://codecov.io — sign in with GitHub, add the repo, copy the upload token from the repo settings page.
CLAUDE_CODE_OAUTH_TOKEN	claude.yml, claude-code-review.yml	Run claude setup-token locally and paste the result.

gh secret set RELEASE_TOKEN
gh secret set CODECOV_TOKEN
gh secret set CLAUDE_CODE_OAUTH_TOKEN

GITHUB_TOKEN is auto-provided — no action needed.

Local development

The PR gate (matches CI):

uv run pytest -x -q                                  # tests
uv run ruff check --fix . && uv run ruff format .    # lint + format
uv run mypy src/ tests/                              # type-check

Pre-commit runs a subset of the gate on each commit; see .pre-commit-config.yaml for details, or CLAUDE.md for the full Hard PR Acceptance Gates.

Troubleshooting

Moving a scaffolded project

uv sync creates .venv/bin/* scripts with absolute shebangs pointing at the venv Python. If you move the repo after scaffolding (mv /old/path /new/path), uv run pytest fails with ModuleNotFoundError: No module named 'fastmcp' because the stale shebang resolves to a different interpreter than the venv's site-packages.

Fix:

rm -rf .venv
uv sync --all-extras --all-groups

uv run python -m pytest also works as a one-shot workaround (bypasses the stale entry-script shim).

uv.lock refresh after copier update

When copier update introduces new dependencies (e.g. a new extra added to pyproject.toml.jinja), CI runs uv sync --frozen which fails against a stale lockfile. Run uv lock locally and commit the refreshed uv.lock alongside accepting the copier-update PR.

Links

• Documentation
• llms.txt
• FastMCP
• fastmcp-pvl-core

Domain configuration

All domain environment variables use the IMAGE_GENERATION_MCP_ prefix.

Core

Variable	Default	Required	Description
IMAGE_GENERATION_MCP_SCRATCH_DIR	~/.image-generation-mcp/images/	No	Directory for saved generated images.
IMAGE_GENERATION_MCP_READ_ONLY	true	No	Hide write-tagged tools (generate_image). Set to false to enable generation.
IMAGE_GENERATION_MCP_DEFAULT_PROVIDER	auto	No	Default provider: auto, openai, gemini, sd_webui, placeholder.

Providers

Variable	Default	Required	Description
IMAGE_GENERATION_MCP_OPENAI_API_KEY	—	No	OpenAI API key; enables OpenAI provider when set.
IMAGE_GENERATION_MCP_GOOGLE_API_KEY	—	No	Google API key with Gemini access; enables Gemini provider when set.
IMAGE_GENERATION_MCP_SD_WEBUI_HOST	—	No	SD WebUI URL (e.g. http://localhost:7860); enables SD WebUI provider when set. Deprecated alias: A1111_HOST.
IMAGE_GENERATION_MCP_SD_WEBUI_MODEL	—	No	SD WebUI checkpoint name for preset detection and override. Deprecated alias: A1111_MODEL.

Authentication

Variable	Default	Required	Description
IMAGE_GENERATION_MCP_BEARER_TOKEN	—	No	Static bearer token; enables bearer auth when set.
IMAGE_GENERATION_MCP_BASE_URL	—	No	Public base URL for OIDC and MCP File Exchange downloads (e.g. https://mcp.example.com).
IMAGE_GENERATION_MCP_OIDC_CONFIG_URL	—	No	OIDC discovery endpoint URL.
IMAGE_GENERATION_MCP_OIDC_CLIENT_ID	—	No	OIDC client ID.
IMAGE_GENERATION_MCP_OIDC_CLIENT_SECRET	—	No	OIDC client secret.
IMAGE_GENERATION_MCP_OIDC_JWT_SIGNING_KEY	ephemeral	Yes on Linux/Docker	JWT signing key.
IMAGE_GENERATION_MCP_OIDC_AUDIENCE	—	No	Expected JWT audience claim.
IMAGE_GENERATION_MCP_OIDC_REQUIRED_SCOPES	openid	No	Comma-separated required scopes.
IMAGE_GENERATION_MCP_OIDC_VERIFY_ACCESS_TOKEN	false	No	Verify access token as JWT instead of id token.

Cost control & performance

Variable	Default	Required	Description
IMAGE_GENERATION_MCP_PAID_PROVIDERS	openai,gemini	No	Comma-separated paid provider names. Triggers elicitation confirmation on capable clients. Set to empty to disable.
IMAGE_GENERATION_MCP_TRANSFORM_CACHE_SIZE	64	No	Max cached transforms. Set to 0 to disable caching.

File Exchange (MCP downloads)

Variable	Default	Required	Description
IMAGE_GENERATION_MCP_FILE_EXCHANGE_ENABLED	true on http/sse, false on stdio	No	Master switch for the file-exchange producer. Set false to suppress all file_ref publishing.
IMAGE_GENERATION_MCP_FILE_EXCHANGE_TTL	3600	No	Default and maximum TTL (seconds) for published files and download URLs. create_download_link's ttl_seconds is clamped to this.
IMAGE_GENERATION_MCP_FILE_EXCHANGE_CONSUME	true	Recommended false	Master switch for the consumer side. This server is producer-only; set false to silence the upstream "consume on, no consumer_sink wired" startup warning.

Server identity

Variable	Default	Required	Description
IMAGE_GENERATION_MCP_SERVER_NAME	image-generation-mcp	No	Server name shown to MCP clients.
IMAGE_GENERATION_MCP_INSTRUCTIONS	(dynamic)	No	System instructions for LLM context.
IMAGE_GENERATION_MCP_HTTP_PATH	/mcp	No	HTTP endpoint mount path.
IMAGE_GENERATION_MCP_APP_DOMAIN	(auto)	No	MCP Apps widget sandbox domain. Auto-computed from BASE_URL for Claude; override for other hosts.

Domain-config fields are composed inside src/image_generation_mcp/config.py between the CONFIG-FIELDS-START / CONFIG-FIELDS-END sentinels; env reads go through fastmcp_pvl_core.env(_ENV_PREFIX, "SUFFIX", default) so naming stays consistent.

For the full MCP tool / resource / prompt surface and per-provider setup notes, see the documentation site.

Key design decisions

• Multi-provider with capability discovery, not feature flags. Each provider's discover_capabilities() reports its actual supported aspect ratios / qualities / formats / negative-prompt support at startup; routing logic asks the capability surface, not a hard-coded enum. New providers slot in by implementing the protocol — no router edits needed. (See docs/decisions/0001-…, 0002-…, 0007-….)
• Per-model style_profile metadata, surfaced via list_providers. Closed-list providers (OpenAI, Gemini, placeholder) use exact-key lookup; SD WebUI uses a regex-ordered pattern table. Profiles include lifecycle flags (current / legacy / deprecated) and feed an auto-built top-level warnings array. (See docs/decisions/0009-….)
• Hybrid background tasks. Short calls (OpenAI ~5 s) stream progress in-line; long calls (SD WebUI 30-180 s) run as background tasks with check_generation_status polling — clients pick the mode via task=True. (See docs/decisions/0005-….)
• Image asset model: content-addressed registry + sidecar JSON metadata + on-demand transforms. Generated images keep their full-resolution original; image://{id}/view?format=webp&width=512&crop_x=… resources do format conversion / resize / crop on demand without re-generating. Transforms are cached. (See docs/decisions/0006-….)
• Style library. User-saved markdown briefs (with YAML frontmatter for tags / aspect ratio / quality) that the LLM interprets per-provider — not copy-pasted verbatim. Distinct from per-model style_profile: style library is the brief; style_profile describes the model. (See docs/decisions/0008-… and 0009-… for disambiguation.)
• Composes fastmcp_pvl_core.ServerConfig, never inherits. Domain config goes between CONFIG-FIELDS-START / CONFIG-FIELDS-END sentinels; env reads route through fastmcp_pvl_core.env(...) to keep prefix naming consistent.