# microsoft/SkillOpt

SkillOpt is a text-space optimizer that trains reusable natural-language skills for frozen LLM agents through trajectory-driven edits, validation-gated updates, and deployable best_skill.md artifacts.

Repository: https://github.com/microsoft/SkillOpt
Canonical: https://ross.abutalabs.com/products/skillopt
Homepage: https://aka.ms/skillopt
Language: Python
License: MIT
License Family: permissive
Topics: agent-skills, self-evolving-agents
Last push: 2026-08-26T15:42:40+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 91, longevity 8
- inputs: {"age_days": 117, "days_push": 7, "days_rel": 62, "gap_med": 29, "n_releases_24m": 2}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 16387, forks 1539 (observed 2026-08-28T04:11:15.123677+00:00)

## What it is
SkillOpt is a Python library from Microsoft that trains reusable natural-language skills for frozen LLM agents using trajectory-driven edits, validation-gated updates, and deployable best_skill.md artifacts. It applies neural-network-style training loops (epochs, batch sizes, learning rates) to text-space optimization without modifying model weights.

## Use cases
- optimize agent skills without fine-tuning model weights
- train reusable natural-language skills for LLM agents
- improve agent performance through trajectory-driven edits
- run offline nightly self-evolution of agent skills
- deploy optimized agent skills as markdown artifacts
- apply validation-gated updates to agent prompts

## When to choose
- you want to improve LLM agent performance without touching model weights
- you need reusable, deployable skill artifacts like best_skill.md
- you want a training-loop-style optimizer for text-space agent skills
- you use Python 3.10+ and want an MIT-licensed, pip-installable skill optimizer

## When to avoid
- you need actual weight fine-tuning or model training
- your agents don't use LLMs or natural-language skills
- you need a framework-agnostic solution outside the Python ecosystem

## Facets
- artifact type: library
- maturity: active
- function: agent-framework, prompt-engineering, llm-training, machine-learning
- domain: artificial-intelligence, large-language-models, developer-tools
- platform: python, cli, cross-platform
- tags: agent-skills, self-evolving-agents, text-space-optimization, prompt-optimization, skill-learning, frozen-llm, trajectory-optimization, validation-gates, ai-agents

## Member repositories
- microsoft/SkillOpt (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:15.123677+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:05:07.966396+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/SkillOpt (fetched 2026-08-28T04:11:15.123677+00:00, sha e0606d629156)
  - registry_pypi: https://pypi.org/pypi/skillopt/json (fetched 2026-08-29T08:02:26.042073+00:00, sha a98678cddf5d)
- Data as of 2026-08-30T08:39:29.467469+00:00.