Open Contributor Program

Contribute a Routine to the SkillSimm Community

Routines are structured workflow templates that run on the SkillSimm harness. Build one, pass the quality gate, and reach every enterprise that simulates your domain.

What is a routine?

A routine is three files that tell the harness how to run a simulation.

Workflow YAML

Step DAG definition

Auto steps (AI-handled) and HITL gates (human decisions), with dependencies, LLM instructions, risk levels, and estimated token budgets per step.

Format: YAML or JSON

Evaluation Rubric

Reference decisions + scoring

Each HITL step gets a reference_decisions block: optimal_decision, rationale, and a scoring_map assigning 0–100 to each decision option.

Format: JSON embedded in template

Sample Data

Test scenario (optional)

A representative case for harness validation — the scenario the simulated participants will encounter during the quality gate run.

Format: markdown or text file

75

>75% decision quality score required for featured placement

Your routine must pass a harness validation run with simulated participants before being listed on the marketplace. The harness scores each HITL decision against your eval rubric's scoring_map. This ensures every featured routine is production-ready — not just theoretically valid.

What contributors earn

No transactional marketplace fees. Direct creator-to-enterprise support.

Featured placement

Your routine is listed on the marketplace with your name and GitHub profile. Featured routines appear at the top of the gallery.

Sponsor donations

Enterprises who deploy your routine in production can sponsor you directly through GitHub Sponsors — no platform cut, direct creator support.

Verified Creator badge

Displayed on your routine cards and GitHub profile after your first routine passes the quality gate. Builds your reputation in the community.

How to submit

Five steps from idea to featured routine.

01

Fork github.com/skillsimm/routines on GitHub

02

Copy the routine template: routine.yaml + eval_rubric.json

03

Fill in your steps — reference claim_adjuster_escalation.json as a real example

04

Open a pull request — the harness runs a validation simulation automatically

05

Pass >75% quality gate → merged and featured on the marketplace