A · leads
deepseek-harness
97.4
282,805 stars · MIT
AI Agents
deepseek-harness leads by 4.4 · narrow
A · leads
97.4
282,805 stars · MIT
B
93.0
117,459 stars · MIT
deepseek-harness versus taste-skill is a AI Agents comparison assembled from public repository metrics, the license on file, and whether each product can be self-hosted. deepseek-harness is written mainly in TypeScript; taste-skill is written mainly in JavaScript. The composite score is 97.4 for deepseek-harness and 93.0 for taste-skill.
deepseek-harness has 282,805 GitHub stars to taste-skill's 117,459, a gap of 165,346 (2.41×). Forks are 35,041 for deepseek-harness and 7,858 for taste-skill. On the log scale used inside this pair, adoption is 100 versus 93, and the fork community is 100 versus 86.
Push history: deepseek-harness was last pushed on 13 November 2026 (4 days ago), and taste-skill was last pushed on 08 November 2026 (8 days ago). Open issues stand at 0 for deepseek-harness (0.000 per star) and 10 for taste-skill (0.000 per star). Issue-health scores are 96 and 96; activity scores are 92 and 92.
Both deepseek-harness and taste-skill are listed as MIT. A permissive grant. A company can usually ship it inside a commercial product and keep its own application code private, provided it keeps the copyright notice. The license score is therefore level at 96 versus 96, and the decision moves to adoption, activity, and hosting. This is a procurement signal, not legal advice.
In AI Agents, both deepseek-harness (282,805 stars) and taste-skill (117,459 stars) are marked self-hostable. Deploy scores are 100 and 100. Running either one means you own upgrades, backups, and the license duties that follow a modified copy. Pricing labels on file: deepseek-harness is open source, taste-skill is open source.
deepseek-harness is ahead on every scored dimension that is not level. Level dimensions: recent activity, issue health, license flexibility, self-hosting. The widest gap is fork community at 14 points (100 vs 86), and the tightest is self-hosting at 0 points (100 vs 100).
deepseek-harness finishes ahead at 97.4 against taste-skill at 93.0, a 4.4 point narrow margin under the published weights.
Scores are computed inside this pair only. Method and weights.
Bars are scaled within the row. Higher is stronger. Open issues are a raw count, so that row has no bar.
| Signal | deepseek-harness | taste-skill |
|---|---|---|
| GitHub stars | 282,805 | 117,459 |
| Forks | 35,041 | 7,858 |
| Open issues | 0 | 10 |
| Adoption | 100 | 93 |
| Activity | 92 | 92 |
| Community | 100 | 86 |
| Issue health | 96 | 96 |
| License | 96 | 96 |
| Self-hosting | 100 | 100 |
| Composite | 97.4 | 93.0 |
DeepSeek Harness: Everything is a Plugin.
Agent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
deepseek-harness has more GitHub stars: 282,805 versus 117,459 for taste-skill. The gap is 165,346 stars.
deepseek-harness was pushed more recently: 13 November 2026 versus 08 November 2026 for taste-skill. That is 4 days apart.
Yes. This catalog marks both deepseek-harness and taste-skill as self-hostable. Confirm each project's deploy docs before you plan a migration.
deepseek-harness is listed as MIT, which this catalog treats as a permissive grant. A company can usually ship it inside a commercial product and keep its own application code private, provided it keeps the copyright notice. taste-skill is listed as MIT, which this catalog treats as a permissive grant. A company can usually ship it inside a commercial product and keep its own application code private, provided it keeps the copyright notice. Read both grants before you embed either codebase in a hosted product. This is not legal advice.
deepseek-harness scores 97.4 and taste-skill scores 93.0. This page calls that a narrow lead for deepseek-harness.