Evos

We codified decades of
operational expertise.
Then we open sourced it.

Evos are the first to translate and codify decades of operational domain expertise into agent-ready skills.

01

Every industry runs on expertise that took decades to build.

Humans have always acquired and stored the most valuable expertise in their fields. Yet, it's been uncaptured by technology — locked in the heads of the people who do the work.

Compatible with

Claude
OpenClaw
OpenAI
GitHub
VS Code
Cursor
Gemini
ClawHub
SkillsMP
Amp
Roo Code
OpenCode
Letta
Mistral
Databricks
Hacker News
Claude
OpenClaw
OpenAI
GitHub
VS Code
Cursor
Gemini
ClawHub
SkillsMP
Amp
Roo Code
OpenCode
Letta
Mistral
Databricks
Hacker News

Benchmarked performance

Every skill, eval-tested and benchmarked against real operational scenarios.

Logistics·30 eval scenarios

Handles shipment delays, damages, losses, refused deliveries, and carrier disputes. Applies Carmack Amendment filing logic, financial impact assessment, and carrier-specific escalation protocols.

With Evos skill95%
Base model (no domain skill)85.2%

+9.8pp improvement over baseline across 30 scenarios

Capabilities

Carmack Amendment filingFinancial impact assessmentCarrier-specific escalationDispute resolution routing

The library

8 skills. Open source. Growing.

Each skill is a self-contained module of codified operational expertise — ready to plug into any agent framework.

01
Logistics30+ evals

Logistics Exception Management

Handles shipment delays, damages, losses, refused deliveries, and carrier disputes. Applies Carmack Amendment filing logic, financial impact assessment, and carrier-specific escalation protocols.

02
Logistics22+ evals

Carrier Relationship Management

Manages carrier portfolios — rate negotiation, performance scorecarding, tender acceptance, routing guide construction, and RFP evaluation across asset carriers, brokers, and niche specialists.

03
Logistics28+ evals

Customs & Trade Compliance

Navigates HS tariff classification, FTA utilisation, restricted party screening, and documentation requirements across US CBP, EU customs union, UK post-Brexit, and Middle East COC regimes.

04
Retail24+ evals

Inventory & Demand Planning

Runs forecasting, safety stock calculation, reorder logic, and promotional lift estimation. Handles seasonal transitions, ABC/XYZ segmentation, and vendor MOQ constraints.

Why it's different

AI models are smart. But without expertise, they guess.

Domain skills close the gap between general intelligence and operational expertise.

01

Built from real operators

Not generated, not scraped. Every skill is built from structured conversations with professionals who've spent 10–20 years in the field.

02

Eval-verified, not vibes-tested

201 automated evaluation scenarios. Every skill is benchmarked against base models on real operational tasks before release.

03

Framework-agnostic

Works across AI models, coding agents, and dedicated platforms — Claude, OpenAI, Gemini, Cursor, OpenClaw, and more. Drop in a skill, get domain expertise.

04

Open source, always

MIT licensed. Use it, fork it, extend it. The operational expertise that powers traditional industries shouldn't be locked behind enterprise contracts.

Quick start

Three commands to domain expertise.

git clone https://github.com/ai-evos/agent-skills.git
cp -r agent-skills/capabilities/logistics-exception-management ~/.claude/skills/
Coming soon

Contribute your expertise.
Get paid for its use.

Domain experts will be able to speak with Evos via WhatsApp — codifying their years of operational knowledge into capabilities, that agents can use. Earning from each use. Future proof your decades of expertise.

01

Talk to Evos — answer scenario-based questions, and tell us about how you actually work

02

We translate your expertise into a tested, agent-usable skill

03

You earn every time your expertise is utilised by an agent

Evos

Evos Discovery Agent

Processing...

Evos

Scenario: A carrier no-shows on a dedicated lane during peak season. Walk me through your first 30 minutes.

01 / 08

The bigger picture

Operational expertise, unlocked for machines.

The future of work is autonomous. For that to happen in the physical economy, agents need decades of domain expertise. Evos translates it, then builds the systems that run the work.

See the product
8Open source capabilities

Covering logistics, retail, manufacturing, and energy — with more being codified.

201Eval scenarios

Every capability benchmarked against base models on real operational tasks.

16+Compatible platforms

From Claude and OpenAI to Cursor, OpenClaw, and dedicated agent frameworks.

For developers

Start building with
domain expertise today.

Open source. MIT licensed. Drop a skill into your agent in three lines of code.

View on GitHub

For enterprises

Looking to turn your
ops autonomous?

Evos assesses your operations, codifies your team's expertise, and deploys autonomous systems in 24 hours.

Learn more