Back to directory
Testing & Reviewwld007/dsh-model-eval
dsh-model-eval
Model evaluation for DeepSeek Harness: run OpenAI-compatible models over jsonl/json/csv datasets, score with rule-based metrics, compare models, and produce reports. Sensitive mode restricts data to local/intranet endpoints.
Suggested install command
npx skills add wld007/dsh-model-eval/dsh-model-evalAlways inspect the linked repository and skill instructions before running commands. Skills are instructions; permissions and execution still matter.
Compatibility
Agent support matrix
3 supported
| Agent | Status |
|---|---|
| Claude Code | Supported |
| OpenCode | Not listed |
| Cursor | Supported |
| MCP | Not listed |
| GitHub Copilot | Not listed |
| Windsurf |