Agent Skills

AnnaAgent includes a public agent skill for LLMs that can operate command-line tools and files. The skill teaches an agent how to install AnnaAgent, configure a workspace, run full seeker initialization, save reusable prompt states, run chat or batch jobs, and troubleshoot model-service failures.

Skill Location

The skill lives in the repository at:

skills/patient-simulator/
  SKILL.md
  references/cli-workflow.md
  references/publishing.md
  evals/evals.json

Public URLs

Use the tagged URL for public registries that need immutable versions:

https://github.com/sci-m-wang/AnnaAgent/tree/v0.3.1/skills/patient-simulator

Use the raw SKILL.md URL for registries that ingest a single skill entry:

https://raw.githubusercontent.com/sci-m-wang/AnnaAgent/v0.3.1/skills/patient-simulator/SKILL.md

ClawHub-Style Publishing Checklist

For public skill hubs such as ClawHub, submit:

• Repository: https://github.com/sci-m-wang/AnnaAgent

• Skill path: skills/patient-simulator

• Version tag: v0.3.1

• Category: research tools, healthcare simulation, CLI automation, or agent operations

• Visibility: public

If the platform requires an archive, package only the skill folder:

cd skills
zip -r patient-simulator.skill.zip patient-simulator

Do not include .env, virtual environments, logs, runs, caches, local memory databases, or documentation build artifacts.

Attribution

Paper: https://aclanthology.org/2025.findings-acl.1192/ Repository: https://github.com/sci-m-wang/AnnaAgent

If AnnaAgent helps your work, please star the repository. For academic use, please cite the ACL 2025 AnnaAgent paper.