AgentBrush
Summary
Coding agents can scaffold a landing page in minutes, but every generated image looks like stock-photo noise — wrong palette, wrong feel, nothing that matches the brand the client spent months defining. AgentBrush is a callable image service that gives those agents a fixed visual identity to work from, so the output looks deliberate rather than assembled.
AgentBrush sits behind an MCP-compatible interface and exposes five generation primitives: style presets, brand identity anchoring via uploaded colors and reference images, a two-stage draft-then-refine pipeline, mask-based inpainting, and multi-CLI routing across Claude Code, Cursor, and Windsurf. The brand identity layer is the load-bearing feature — without it, every agent call is a fresh roll of the dice on visual consistency. The two-model pipeline (fast draft, premium refine) gives you a cost lever that matters when image generation is inside a token-budgeted loop. Self-hosting is not available, so teams with air-gapped infrastructure or strict data residency requirements hit a wall immediately. There is no free tier, which means evaluation requires a paid commitment before you know whether the brand anchoring actually holds for your specific assets.
Bottom line: AgentBrush earns its place when a coding agent is building something that ships to real users and visual consistency is non-negotiable — but if your pipeline runs in a private cloud or your team needs to stress-test before paying, you will need a different path.
Pricing Plans
Flat RateLast verified 1 month ago- Price
- $6.99/mo
- Free Tier
- 50 tokens · ~7–10 realistic or ~50 draft
Free
50 tokens u00b7 ~7u201310 realistic or ~50 draft
- All 5 presets (realistic u00b7 illustration u00b7 pixel art u00b7 isometric u00b7 logo)
- Low & medium quality (high quality on Pro & Power)
- Per-project brand identity + token-based mask editor
- MCP for Claude Code u00b7 Cursor u00b7 Codex u00b7 Gemini
- No overage
Starter
100 tokens u00b7 ~15u201320 realistic or ~100 draft
- All 5 presets (realistic u00b7 illustration u00b7 pixel art u00b7 isometric u00b7 logo)
- Low & medium quality (high quality on Pro & Power)
- Per-project brand identity + token-based mask editor
- MCP for Claude Code u00b7 Cursor u00b7 Codex u00b7 Gemini
- No overage
Pro
600 tokens u00b7 ~100u2013120 realistic or ~600 draft
- Everything in Starter
- 6u00d7 the Starter token budget
- High quality unlocked, ~2.8u00d7 cheaper per token
- Room for a weekly delivery cadence
- No overage
Power
1300 tokens u00b7 ~210u2013260 realistic or ~1,300 draft
- Everything in Pro
- 13u00d7 the Starter token budget
- Pay-as-you-go overage ($0.04/token)
- Built for production workloads
View full pricing on agentbrush.dev →
Pricing may have changed since last verified. Check the official site for current plans.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- Brand identity anchoring via uploaded colors, fonts, and reference images, so every agent-generated asset matches the established visual system without re-prompting from scratch on each call.
- Two-model draft-then-refine pipeline, so teams running image generation inside a token-budgeted agent loop get photorealistic quality without paying premium-model costs on every generation.
- Mask-based inpainting exposed as an agent primitive, so iterating on a specific region — a product label, a background element — does not require regenerating the full image and risking consistency drift.
- Single subscription covers Claude Code, Cursor, Windsurf, and any MCP-compatible agent, so adding AgentBrush to an existing multi-tool pipeline does not require separate integrations or per-CLI licensing.
- Transparent-background PNG generation included as a supported output, so agents producing assets for web or game projects get cut-out-ready files without a post-processing step.
Cons
Sign in to edit- No self-hosted option exists, so any team operating under air-gapped infrastructure, private cloud mandates, or strict data residency requirements cannot use AgentBrush at all — they evaluate a self-hostable alternative before the first generation runs.
- No free tier means the brand anchoring quality — the feature the whole service is built around — cannot be validated against real brand assets before a paid commitment. Teams that need proof before budget sign-off either pay to test or move on.
- The service is invoked by agents rather than running autonomously, so if your pipeline requires the image layer to make its own planning or retry decisions without an external caller, AgentBrush does not cover that gap — you are adding orchestration logic on top.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- Web, Claude Code, Cursor, Codex CLI, Gemini CLI
- API Available
- Yes
- Self-Hosted
- No
- Last Updated
- 2026-06-18T13:41:56.164Z
Best For
Who it's for
- Developers using Claude Code or Cursor
- Projects requiring brand-consistent AI images
- Iterative image refinement via mask editing
- Token-budgeted production image generation
What it does well
- Generate on-brand marketing and social creatives
- Create consistent product photography from reference images
- Produce 2D game assets like sprites and tilesets
- Edit specific image regions without full re-generation
- Generate transparent-background PNGs for projects
Integrations
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Recommended skills for this tool
Auto-curated by the AIDiveForge recommendation matrix. These skills are predicted to enhance this tool based on category, capability, and domain signals.
-
Diffusion Prompt Library enhance 32%
A curated, tagged prompt library for diffusion models with negative-prompt presets and CFG/sampler defaults.
Why: category partial · caps 0/0 · domain design
-
Thumbnail A/B Generator enhance 32%
Generate 5 thumbnail variants for a single video using a brand template and a topic prompt — ready for YouTube multi-variant testing.
Why: category partial · caps 0/0 · domain design
-
Product-Shot Background Swap enhance 32%
Replace a product photo's background using a reference palette while preserving the product's reflections and shadows.
Why: category partial · caps 0/0 · domain design
Frequently Asked Questions
- Is AgentBrush free?
- AgentBrush has a permanent free tier alongside paid upgrades (paid plans from $6.99/mo). You can keep using a baseline version indefinitely without paying.
- Is AgentBrush open source?
- No — AgentBrush is a closed-source tool. Source code is not publicly available.
- Does AgentBrush have an API?
- Yes. AgentBrush exposes a developer API. See the official documentation at https://agentbrush.dev for details.
- What platforms does AgentBrush support?
- AgentBrush is available on: Web, Claude Code, Cursor, Codex CLI, Gemini CLI.
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Best AgentBrush alternatives →
Curated lists that include this category
AgentBrush is a paid image generation service designed to be called by AI coding agents rather than operated by humans directly. The core workflow: an agent invokes AgentBrush via an MCP-compatible connection, selects a style preset (realistic photo, flat illustration, pixel art, isometric, or logo), and the service applies a brand identity layer — built from uploaded colors, fonts, and reference images — to every generation. A two-stage pipeline drafts with a fast model and refines with a premium one, so the agent can get photorealistic output without burning the full token budget on every call. Mask editing lets the agent isolate and repaint a specific region without regenerating the whole image, which matters when you are iterating on a product shot or a UI element rather than starting from scratch.
The brand identity system is what separates AgentBrush from calling a generic image API directly. Instead of encoding brand constraints into a prompt every time — which drifts — you upload the reference material once and every subsequent generation inherits it. The vendor describes this as giving agents ‘persistent visual taste,’ and the architecture supports that claim: the brand context travels with the subscription, not with the prompt.
AgentBrush fits workflows where a coding agent is generating production assets — marketing creatives, product photography, game sprites, transparent-background PNGs — and where visual inconsistency would require a human design review to catch. It does not fit teams that need on-premises image generation, those under data residency constraints, or those building pipelines that must be validated at zero cost before committing. There is no self-hosted option and no free tier, so both of those teams are looking elsewhere from day one.
On the integration side, the vendor states compatibility with Claude Code, Cursor, Windsurf, and any MCP-compatible agent under a single subscription. The API is available, so teams building custom agent pipelines outside those named CLIs can still wire AgentBrush in without waiting for a first-party integration.