Distilled SKILL.md beats Workflow Memory by 6.06 points — skills stabilize execution, not knowledge
rohanpaul_ai · x · 2026-09-08
In matched experiments, agents given the same past trajectories as a distilled SKILL.md outperformed Workflow Memory (which keeps more execution detail) by 6.06 percentage points — the gain came from better packaging of identical experience, not more experience.
Trajectory analysis clarifies the mechanism: 65.7% of successful skill cases worked through "procedural anchoring" — stabilizing what to do first, which tools to use, and what to verify — while only 4.5% worked by injecting missing knowledge. Skills mainly fix execution, not knowledge gaps.
This also explains the failure mode: skills hurt when applied in the wrong context or followed too rigidly. The takeaway: self-improving agents need better distillation and application of experience, not bigger memory libraries.
Related event: Papers Reveal When Agent Skill Retrieval Helps—and Hurts(4 posts)→
More from coding & agent
- Coding Agent Deleted a Production Database During Code Freeze: Anaconda's Governance Fix — anacondainc · 2026-10-27
- GPT-6.1 Terra Model Slug Spotted in APIs and Codex Changelog Before Launch — testingcatalog · 2026-10-27
- Claude Projects now lets you drag threads into the message box or @ mention them for precise context — dfeinition · 2026-10-27
- Jane Street's Yinsky shares tips for making codebases agent-friendly: types, property tests, formal methods — sh_reya · 2026-10-27
- Celesto launches CI runners nearly 90% cheaper than GitHub Actions, with agent sandboxes — aniketmaurya · 2026-10-27
- Runway launches App Studio: build no-code creative apps that generate image, video and audio — runwayml · 2026-10-27