+ Executive Strategy for Self-Evolving Agent Skills. SkillOpt treats a compact + natural-language skill document as the trainable state of a frozen language + agent, then learns that document through rollouts, reflection, bounded edits, + and held-out validation gates. +
+ + + + + + ++ SkillOpt makes the skill document itself the optimization target. The + target model, backend, and harness stay fixed; the procedure that guides + evidence gathering, tool use, verification, and output formatting evolves. +
++ Instead of fine-tuning a model or hand-maintaining prompts, SkillOpt runs + the frozen agent on scored batches, asks a separate optimizer model to + propose structured edits, and accepts a candidate only when validation + performance improves. +
+The target model executes tasks with the current skill and records scored trajectories.
+The optimizer analyzes success and failure minibatches to find reusable procedures.
+Candidate add, delete, and replace operations are merged and ranked under a budget.
+The candidate skill is kept only if it improves held-out selection performance.
++ The loop deliberately mirrors a learning algorithm: rollout evidence acts + like a forward pass, reflection acts like a language-level backward pass, + and the textual learning rate bounds how far the skill can move. +
+Rollout batches capture messages, tool calls, verifier feedback, task metadata, and final scores.
+Failures and successes are reflected separately so edits correct recurring errors while preserving working behavior.
+An edit budget functions as a textual learning rate, preventing useful rules from being overwritten by broad rewrites.
+Rejected edits, slow update, and optimizer-side meta skill provide longer-horizon feedback without bloating deployment.
+
+ + The table reports main-result gains across target models and + execution harnesses, comparing no-skill execution with the final + SkillOpt skill on held-out test splits. +
+| Target model | +Harness | +SearchQA | +Sheet | +Office | +DocVQA | +LiveMath | +ALFWorld | +Avg gain | +
|---|---|---|---|---|---|---|---|---|
 GPT-5.5 |
+ Direct chat | ++9.6 | ++38.9 | ++39.0 | ++12.4 | ++29.3 | ++11.9 | ++23.5 | +
| GPT-5.4 |
+ Direct chat | ++6.2 | ++21.1 | ++12.8 | ++13.6 | ++7.2 | ++15.6 | ++12.8 | +
| GPT-5.4-mini |
+ Direct chat | ++4.3 | ++11.4 | ++26.7 | ++16.5 | ++4.8 | ++12.7 | ++12.7 | +
| GPT-5.4-nano |
+ Direct chat | ++19.0 | ++8.2 | ++33.7 | ++49.4 | ++4.0 | ++35.1 | ++24.9 | +
| GPT-5.2 |
+ Direct chat | ++11.2 | ++18.9 | ++21.5 | ++16.5 | ++15.2 | ++16.4 | ++16.6 | +
 Qwen3.5-4B |
+ Direct chat | ++3.1 | ++14.6 | ++15.2 | ++2.1 | ++29.6 | ++50.7 | ++19.2 | +
| Qwen3.6-35B-A3B |
+ Direct chat | ++7.6 | ++9.3 | ++1.2 | ++3.8 | ++10.4 | ++22.4 | ++9.1 | +
| GPT-5.5 |
+ Codex | ++5.5 | ++57.5 | ++12.8 | ++5.0 | ++28.0 | +N/A | ++21.8 | +
| GPT-5.5 |
+ Claude Code | ++4.0 | ++58.3 | ++13.9 | ++3.5 | ++13.3 | +N/A | ++18.6 | +
+ The paper isolates the optimizer components that keep skill learning stable: + enough evidence, bounded textual updates, rejected-edit feedback, slow + update, and optimizer-side memory. +
+| Component | +Setting | +SearchQA | +Spreadsheet | +LiveMath | +
|---|---|---|---|---|
| Learning rate | +lr=4 default | +87.1 | +77.5 | +61.3 | +
| Learning rate | +without lr | +84.6 | +75.7 | +57.3 | +
| Rejected buffer | +with buffer | +87.1 | +77.5 | +61.3 | +
| Rejected buffer | +without buffer | +85.5 | +72.9 | +58.9 | +
| Update memory | +meta skill + slow update | +87.1 | +77.5 | +61.3 | +
| Update memory | +without both | +86.3 | +55.0 | +59.7 | +
+ + This ALFWorld run uses GPT-5.4-mini as the frozen target model and + GPT-5.5 as the optimizer model. The plot tracks train rollout and + held-out selection scores; hover or focus a point to inspect the + skill edit proposed at that stage. +
+
+ SkillOpt exports a compact best_skill.md. The paper tests
+ whether that artifact transfers across model sizes, execution harnesses,
+ and nearby benchmarks without further target-side optimization.
+
GPT-5.4 LiveMath skill transferred to GPT-5.4-nano on LiveMathBench.
+Codex-trained SpreadsheetBench skill transferred into Claude Code.
+GPT-5.4-nano used as its own optimizer improved SpreadsheetBench over baseline.
+The target model consumes only the final skill, not optimizer memory.
++ The final paper link and BibTeX will be updated when the preprint is available. +
+Use this placeholder until the camera-ready citation is available.
+
+ @misc{skillopt2026,
+ title = {SkillOpt: Executive Strategy for Self-Evolving Agent Skills},
+ author = {SkillOpt Authors},
+ year = {2026},
+ note = {Preprint forthcoming},
+ url = {https://microsoft.github.io/SkillOpt/}
+}+ Executive Strategy for Self-Evolving Agent Skills. SkillOpt treats a compact + natural-language skill document as the trainable state of a frozen language + agent, then learns that document through rollouts, reflection, bounded edits, + and held-out validation gates. +
+ + + + + + ++ SkillOpt makes the skill document itself the optimization target. The + target model, backend, and harness stay fixed; the procedure that guides + evidence gathering, tool use, verification, and output formatting evolves. +
++ Instead of fine-tuning a model or hand-maintaining prompts, SkillOpt runs + the frozen agent on scored batches, asks a separate optimizer model to + propose structured edits, and accepts a candidate only when validation + performance improves. +
+The target model executes tasks with the current skill and records scored trajectories.
+The optimizer analyzes success and failure minibatches to find reusable procedures.
+Candidate add, delete, and replace operations are merged and ranked under a budget.
+The candidate skill is kept only if it improves held-out selection performance.
++ The loop deliberately mirrors a learning algorithm: rollout evidence acts + like a forward pass, reflection acts like a language-level backward pass, + and the textual learning rate bounds how far the skill can move. +
+Rollout batches capture messages, tool calls, verifier feedback, task metadata, and final scores.
+Failures and successes are reflected separately so edits correct recurring errors while preserving working behavior.
+An edit budget functions as a textual learning rate, preventing useful rules from being overwritten by broad rewrites.
+Rejected edits, slow update, and optimizer-side meta skill provide longer-horizon feedback without bloating deployment.
+
+ + The table reports main-result gains across target models and + execution harnesses, comparing no-skill execution with the final + SkillOpt skill on held-out test splits. +
+| Target model | +Harness | +SearchQA | +Sheet | +Office | +DocVQA | +LiveMath | +ALFWorld | +Avg gain | +
|---|---|---|---|---|---|---|---|---|
| GPT-5.5 |
+ Direct chat | ++9.6 | ++38.9 | ++39.0 | ++12.4 | ++29.3 | ++11.9 | ++23.5 | +
| GPT-5.4 |
+ Direct chat | ++6.2 | ++21.1 | ++12.8 | ++13.6 | ++7.2 | ++15.6 | ++12.8 | +
| GPT-5.4-mini |
+ Direct chat | ++4.3 | ++11.4 | ++26.7 | ++16.5 | ++4.8 | ++12.7 | ++12.7 | +
| GPT-5.4-nano |
+ Direct chat | ++19.0 | ++8.2 | ++33.7 | ++49.4 | ++4.0 | ++35.1 | ++24.9 | +
| GPT-5.2 |
+ Direct chat | ++11.2 | ++18.9 | ++21.5 | ++16.5 | ++15.2 | ++16.4 | ++16.6 | +
| Qwen3.5-4B |
+ Direct chat | ++3.1 | ++14.6 | ++15.2 | ++2.1 | ++29.6 | ++50.7 | ++19.2 | +
| Qwen3.6-35B-A3B |
+ Direct chat | ++7.6 | ++9.3 | ++1.2 | ++3.8 | ++10.4 | ++22.4 | ++9.1 | +
| GPT-5.5 |
+ Codex | ++5.5 | ++57.5 | ++12.8 | ++5.0 | ++28.0 | +N/A | ++21.8 | +
| GPT-5.5 |
+ Claude Code | ++4.0 | ++58.3 | ++13.9 | ++3.5 | ++13.3 | +N/A | ++18.6 | +
+ The paper isolates the optimizer components that keep skill learning stable: + enough evidence, bounded textual updates, rejected-edit feedback, slow + update, and optimizer-side memory. +
+| Component | +Setting | +SearchQA | +Spreadsheet | +LiveMath | +
|---|---|---|---|---|
| Learning rate | +lr=4 default | +87.1 | +77.5 | +61.3 | +
| Learning rate | +without lr | +84.6 | +75.7 | +57.3 | +
| Rejected buffer | +with buffer | +87.1 | +77.5 | +61.3 | +
| Rejected buffer | +without buffer | +85.5 | +72.9 | +58.9 | +
| Update memory | +meta skill + slow update | +87.1 | +77.5 | +61.3 | +
| Update memory | +without both | +86.3 | +55.0 | +59.7 | +
+ + This ALFWorld run uses GPT-5.4-mini as the frozen target model and + GPT-5.5 as the optimizer model. The plot tracks train rollout and + held-out selection scores; hover or focus a point to inspect the + skill edit proposed at that stage. +
+
+ SkillOpt exports a compact best_skill.md. The paper tests
+ whether that artifact transfers across model sizes, execution harnesses,
+ and nearby benchmarks without further target-side optimization.
+
GPT-5.4 LiveMath skill transferred to GPT-5.4-nano on LiveMathBench.
+Codex-trained SpreadsheetBench skill transferred into Claude Code.
+GPT-5.4-nano used as its own optimizer improved SpreadsheetBench over baseline.
+The target model consumes only the final skill, not optimizer memory.
++ The final paper link and BibTeX will be updated when the preprint is available. +
+Use this placeholder until the camera-ready citation is available.
+@misc{skillopt2026,
+ title = {SkillOpt: Executive Strategy for Self-Evolving Agent Skills},
+ author = {SkillOpt Authors},
+ year = {2026},
+ note = {Preprint forthcoming},
+ url = {https://microsoft.github.io/SkillOpt/}
+}