harbor: evaluate agents in sandboxed environments
harborframework.com Source★ 302Apache-2.0 DiscordAI Models & Agents / Prompts, Evals & LLM Observability
- Monthly visits
- 142K
- Down14.4%MoMOct 2026
- Public revenue
- Undisclosed
- No public revenue evidence yet
- Entry price
- Free
- No monthly subscription pricing
- Launched
- 12mo ago
- 2025-11-06 · Est.
Traffic
Marketing moves
Launches, search & content, referrals, promotions- We ported 369 of the formally stated problems into @harborframework environments so they can be run for evaluation or RL training with OpenEnv. Each environment gives the agent a Lean 4 theorem with sorry in place of the proof. The reward
@adithya_s_k · 2026-10-09 · 48 likes · 4 reposts · 3,481 views
Boxd announces support across Harbor, AG2, Sheprd, Tardigrade, and BastProduct Activity
Boxd announced that five additional projects—Harbor, AG2, Sheprd, Tardigrade, and Bast—now work with Boxd machines.
Nov 12 · 3d agoX Post
At the time: 142K visits/mo1 concurrent moveHarbor co-hosts RL / data / SkyRL meetup with Anyscale and MercorProduct Activity
Anyscale announced that it is hosting an RL, data, and SkyRL meetup in San Francisco with Mercor and Harbor. The event agenda covers topics including SkyRL at Mercor, updates in SkyRL, Ray Sandboxes, and Harbor.
Nov 3 · 12d agoX Post
At the time: 142K visits/mo1 concurrent moveNew product launchLaunch
Nov 6, 2025 · 12mo agoWebsite
Open Source & Agent Ecosystem
- GitHub Stars
- ★ 302
- Forks
- 197
- License
- Apache-2.0
- Primary language
- Go
- Ecosystem & integrations:
- Claude CodeCodexClaude
- Exposed tool capabilities:
- Model EvaluationAgent MonitoringEvaluationOptimizationTrajectory analysisTask QA
Traffic sources
Traffic breakdown
Channels
- Direct74%
- Organic Search13%
- Referrals8%
- Social3%
- Other1%
Countries
- United States44%
- China23%
- India12%
- Singapore5%
- Germany3%
Search
Domain authority —Domain registered 2025-11-06
| Keyword | Monthly volume |
|---|---|
| harbor | 75.3K |
| harbor framework | 2,250 |
| harbor evals | 890 |
| harbor rl agent | 200 |
| harbor rl docs | 200 |
Traffic data: Vaneform
Pricing
Pricing page not identified yet, or no public pricing listed.
Pages & changes
Key pages
Weekly snapshotsChanges
No pricing, plan, or positioning changes detected since tracking started.
Design
Dark Developer (Light canvas, airy density) featuring centered headline + cta layout and code block / terminal / cli visuals.
First-screen preview · click to zoom
First-screen pitch & conversion flow
“harbor: evaluate agents in sandboxed environments”
Centered Headline + CTA hero structure paired with download app / install extension CTA and no explicit trust bar.
Conversion path: Centered Headline + CTA + Code Block / Terminal / CLI → Download App / Install Extension
- Category Positioning Anchor
- Code & Terminal
- Video Demo
Palette · click to copy
- Typography
- 8.0
- Color & Contrast
- 7.0
- Layout & Spacing
- 8.0
- Craft & Polish
- 7.0
- Distinctiveness
- 7.0
Structure
- Fonts
- Google Sans Code / Google Sans Flex
- Hero layout
- Centered Headline + CTA
- Primary visual
- Code Block / Terminal / CLI
- Conversion CTA
- Download App / Install Extension
- Social proof
- No Explicit Trust Bar
- Navigation
- Full-Width Top Bar
- Theme & density
- Light · Airy
- Corners & depth
- Small Radius · Hairline borders
Preview full AI design brief (includes :root CSS variables & layout rules)Style tokens only; never copy brand names, logos, copy, or images
# Design Brief: Dark Developer (with Minimal Monochrome)
Deep dark surfaces, monospace typography, terminal/code snippets, and high-precision borders.
## Design Tokens
```css
:root {
--background: #ffffff;
--surface: #f1f1f1;
--foreground: #111111;
--muted-foreground: #777777;
--accent: #111111;
--radius: 4px;
--font-heading: "Google Sans Code", ui-sans-serif, system-ui, sans-serif;
--font-body: "Google Sans Flex", ui-sans-serif, system-ui, sans-serif;
--font-mono: "Google Sans Code", ui-sans-serif, system-ui, sans-serif;
}
```
Color area ratio: Background #FFFFFF (~80%) · Surface #F1F1F1 (~7%) · Text #111111 (~6%) · Muted Text #777777 (~4%) · Accent #111111 (~3%). Background and surface dominate the canvas; reserve the accent color for primary buttons and active states.
## Layout & Conversion Structure
- Hero layout: Centered Headline + CTA
- Navigation: Full-Width Top Bar
- Theme & density: Light · Airy
- Border radius & elevation: Small Radius; no drop shadows—separate layers with hairline borders and whitespace
- Primary visual medium: Code Block / Terminal / CLI
- Conversion CTA: Download App / Install Extension
- Headline hook: Category Positioning Anchor
- Hero components: Code & Terminal, Video Demo
- Spatial layout notes: Centered Headline + CTA hero structure paired with download app / install extension CTA and no explicit trust bar.
- Visual summary: Dark Developer (Light canvas, airy density) featuring centered headline + cta layout and code block / terminal / cli visuals.
## Avoid
- Warm pastel backgrounds
- Playful cartoon illustrations
- Oversized pill buttons everywhere
Implement the UI following the design tokens and layout specifications above: use only these color variables, and reserve the accent color strictly for primary actions and active states. Adopt only this visual style—never copy any reference product’s brand name, logo, copy, or imagery.
Questions & evidence
- How much website traffic does Harbor get?
- Vaneform estimates 142K website visits in 2026-10. Visits can include repeat visits; this estimate measures web traffic separately from paying customers or revenue.Source
- What is Harbor and what platforms does it support?
- Harbor is catalogued under Prompts, Evals & LLM Observability: harbor: evaluate agents in sandboxed environments. Supported platforms: CLI / SDK.
- Is Harbor open source?
- Harbor has a public repository (https://github.com/laude-institute/harbor) under the Apache-2.0 license, recording 302 GitHub stars.Source
Related products in this category
These products share a catalog category. Compare customer tasks, features and deployment requirements before treating them as alternatives.
| Product | Entry price | Visits | Revenue |
|---|---|---|---|
| LMSYS Chatbot ArenaGiantExperience the frontier | — | 30M | Undisclosed |
| Prompt CowboyTurn lazy prompts into great ones | $8/mo | 1.5M | Undisclosed |
| LiveBench 2.0A challenging, contamination-free long-horizon AI benchmark. | — | 824.6K | Undisclosed |
| LangfuseOpen Source Agent Evals &Observability | $29/mo | 722.9K | Undisclosed |
| Codex雷达站查看雷达站正在改什么:工单分组内全部工单的阶段、进展与最近变化。 | — | 605.1K | Undisclosed |
| Pydantic LogfirePydantic Build agents you’ll actually ship to production | $49/mo | 454.4K | Undisclosed |