# What changed in agent harnesses: 2026-10-25

Changes to the best-of-Agent-Harnesses list since the 2026-10-18 refresh. The list now ranks 187 agent harnesses, and the star counts below come from GitHub on 2026-10-25.

## Biggest star gains

| Harness | Category | Stars | Gain |
|---|---|---:|---:|
| [opencode](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/opencode/) | Coding agent products (IDEs, CLIs, full suites) | 221k | +5,228 |
| [superpowers](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/superpowers/) | Coding harness configs and SDKs | 305k | +3,403 |
| [addyosmani/agent-skills](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/agent-skills/) | Coding harness configs and SDKs | 110k | +2,333 |
| [Hermes](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/hermes-agent/) | Personal agent runtimes | 257k | +2,124 |
| [DeepSeek Harness](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/deepseek-harness/) | Coding agent products (IDEs, CLIs, full suites) | 261k | +2,036 |
| [Headroom](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/headroom/) | Progressive disclosure harnesses | 77.8k | +2,005 |
| [Anthropic Skills](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/skills/) | Coding harness configs and SDKs | 184k | +1,762 |
| [pi](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/pi/) | Coding agent products (IDEs, CLIs, full suites) | 118k | +1,658 |
| [context-mode](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/context-mode/) | Progressive disclosure harnesses | 32.2k | +1,535 |
| [claude-mem](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/claude-mem/) | Memory and state | 103k | +1,463 |

## Fastest growing

Projects that already had 500+ stars.

- [HarnessRouter](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/harnessrouter/): +10.9% (3.3k to 3.6k stars)
- [context-mode](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/context-mode/): +5.0% (30.7k to 32.2k stars)
- [Terminal-Bench](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/terminal-bench/): +4.9% (925 to 970 stars)
- [OpenJarvis](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/openjarvis/): +4.5% (11.5k to 12k stars)
- [ClawBench](https://ryanalberts.github.io/best-of-Agent-Harnesses/h/clawbench/): +3.9% (1.1k to 1.1k stars)

## Get this every week

Star and watch [the repo](https://github.com/RyanAlberts/best-of-Agent-Harnesses) (Watch, then Custom, then Releases) to get this digest each Sunday. Browse every harness on [the site](https://ryanalberts.github.io/best-of-Agent-Harnesses/). Agents can read the same data with the `whats_new` tool of the MCP server: `uvx agent-harnesses-mcp`.