무료GitHub에 공개

optim-agent

Use when the user wants to optimize configurable system parameters against a measurable scalar objective, especially for model training, inference, quantitative strategies, reinforcement learning, scientific workflows, or other expensive black-box evaluations where reading the project can improve trial selection.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Optim-Agent/optim-agent

설치 방법 보기

포함된 파일(200)

SKILL.md5.0 KB
.agents/plugins/marketplace.json374 B
.claude-plugin/marketplace.json438 B
.github/dependabot.yml269 B
.github/ISSUE_TEMPLATE/bug_report.md279 B
.github/ISSUE_TEMPLATE/feature_request.md193 B
.github/pull_request_template.md199 B
.github/workflows/ci.yml1.2 KB
.github/workflows/docs.yml710 B
.gitignore746 B
.pre-commit-config.yaml262 B
benchmarks/manifest.json2.1 KB
benchmarks/README.md3.7 KB
CHANGELOG.md1.7 KB
CITATION.cff630 B
CODE_OF_CONDUCT.md1013 B
CONTRIBUTING.md1.6 KB
docs/assets/cifar10_curves_GPT-5.5-medium_s0.json9.9 KB
docs/assets/cifar10_curves_GPT-5.5-medium_s1.json9.8 KB
docs/assets/cifar10_curves_GPT-5.5-medium_s2.json9.8 KB
docs/assets/cifar10_curves_GPT-5.5-medium_s3.json9.9 KB
docs/assets/cifar10_curves_GPT-5.5-medium_s4.json9.8 KB
docs/assets/cifar10_curves_GPT-5.5-medium-no-context_s0.json10.5 KB
docs/assets/cifar10_curves_GPT-5.5-medium-no-context_s1.json10.5 KB
docs/assets/cifar10_curves_GPT-5.5-medium-no-context_s2.json10.4 KB
docs/assets/cifar10_curves_GPT-5.5-medium-no-context_s3.json10.5 KB
docs/assets/cifar10_curves_GPT-5.5-medium-no-context_s4.json10.7 KB
docs/assets/cifar10_curves_Random_s0.json10.8 KB
docs/assets/cifar10_curves_Random_s1.json10.8 KB
docs/assets/cifar10_curves_Random_s2.json10.8 KB
docs/assets/cifar10_curves_Random_s3.json10.7 KB
docs/assets/cifar10_curves_Random_s4.json10.9 KB
docs/assets/cifar10_curves_TPE_s0.json10.7 KB
docs/assets/cifar10_curves_TPE_s1.json10.8 KB
docs/assets/cifar10_curves_TPE_s2.json10.8 KB
docs/assets/cifar10_curves_TPE_s3.json10.7 KB
docs/assets/cifar10_curves_TPE_s4.json10.8 KB
docs/assets/classification_benchmarks.png102.4 KB
docs/assets/credit_card.png93.2 KB
docs/assets/credit_default_GP-BO_s0.json6.9 KB
docs/assets/credit_default_GP-BO_s1.json7.0 KB
docs/assets/credit_default_GP-BO_s2.json6.9 KB
docs/assets/credit_default_GP-BO_s3.json6.9 KB
docs/assets/credit_default_GP-BO_s4.json6.9 KB
docs/assets/credit_default_GPT-5.5_s0.json6.5 KB
docs/assets/credit_default_GPT-5.5_s1.json6.6 KB
docs/assets/credit_default_GPT-5.5_s2.json6.6 KB
docs/assets/credit_default_GPT-5.5_s3.json6.5 KB
docs/assets/credit_default_GPT-5.5_s4.json6.6 KB
docs/assets/credit_default_GPT-5.5-no-context_s0.json6.0 KB
docs/assets/credit_default_GPT-5.5-no-context_s1.json6.1 KB
docs/assets/credit_default_GPT-5.5-no-context_s2.json6.1 KB
docs/assets/credit_default_GPT-5.5-no-context_s3.json5.9 KB
docs/assets/credit_default_GPT-5.5-no-context_s4.json6.0 KB
docs/assets/credit_default_Random_s0.json6.7 KB
docs/assets/credit_default_Random_s1.json6.7 KB
docs/assets/credit_default_Random_s2.json6.7 KB
docs/assets/credit_default_Random_s3.json6.7 KB
docs/assets/credit_default_Random_s4.json6.7 KB
docs/assets/credit_default_TPE_s0.json6.7 KB
docs/assets/credit_default_TPE_s1.json6.7 KB
docs/assets/credit_default_TPE_s2.json6.7 KB
docs/assets/credit_default_TPE_s3.json6.7 KB
docs/assets/credit_default_TPE_s4.json6.7 KB
docs/assets/hard_benchmarks_free.png127.0 KB
docs/assets/hard_benchmarks_tier.png148.7 KB
docs/assets/hard_curves_Big-pickle_s0.json2.4 KB
docs/assets/hard_curves_Big-pickle_s1.json2.4 KB
docs/assets/hard_curves_Big-pickle_s2.json2.4 KB
docs/assets/hard_curves_Big-pickle_s3.json2.4 KB
docs/assets/hard_curves_Big-pickle_s4.json2.4 KB
docs/assets/hard_curves_DeepSeek-V4-Flash_s0.json2.4 KB
docs/assets/hard_curves_DeepSeek-V4-Flash_s1.json2.4 KB
docs/assets/hard_curves_DeepSeek-V4-Flash_s2.json2.5 KB
docs/assets/hard_curves_DeepSeek-V4-Flash_s3.json2.4 KB
docs/assets/hard_curves_DeepSeek-V4-Flash_s4.json2.5 KB
docs/assets/hard_curves_GLM-5.2_s0.json2.4 KB
docs/assets/hard_curves_GLM-5.2_s1.json2.4 KB
docs/assets/hard_curves_GLM-5.2_s2.json2.4 KB
docs/assets/hard_curves_GLM-5.2_s3.json2.5 KB
docs/assets/hard_curves_GLM-5.2_s4.json2.4 KB
docs/assets/hard_curves_GP-BO_s0.json2.8 KB
docs/assets/hard_curves_GP-BO_s1.json2.8 KB
docs/assets/hard_curves_GP-BO_s2.json2.8 KB
docs/assets/hard_curves_GP-BO_s3.json2.8 KB
docs/assets/hard_curves_GP-BO_s4.json2.8 KB
docs/assets/hard_curves_GPT-5.5_s0.json2.5 KB
docs/assets/hard_curves_GPT-5.5_s1.json2.5 KB
docs/assets/hard_curves_GPT-5.5_s2.json2.4 KB
docs/assets/hard_curves_GPT-5.5_s3.json2.5 KB
docs/assets/hard_curves_GPT-5.5_s4.json2.5 KB
docs/assets/hard_curves_Kimi-K3_s0.json2.4 KB
docs/assets/hard_curves_Kimi-K3_s1.json2.5 KB
docs/assets/hard_curves_Kimi-K3_s2.json2.4 KB
docs/assets/hard_curves_Kimi-K3_s3.json2.5 KB
docs/assets/hard_curves_Kimi-K3_s4.json2.5 KB
docs/assets/hard_curves_MiMo-v2.5_s0.json2.4 KB
docs/assets/hard_curves_MiMo-v2.5_s1.json2.4 KB
docs/assets/hard_curves_MiMo-v2.5_s2.json2.4 KB
docs/assets/hard_curves_MiMo-v2.5_s3.json2.4 KB
docs/assets/hard_curves_MiMo-v2.5_s4.json2.5 KB
docs/assets/hard_curves_Minimax-M3_s0.json2.4 KB
docs/assets/hard_curves_Minimax-M3_s1.json2.4 KB
docs/assets/hard_curves_Minimax-M3_s2.json2.4 KB
docs/assets/hard_curves_Minimax-M3_s3.json2.4 KB
docs/assets/hard_curves_Minimax-M3_s4.json2.4 KB
docs/assets/hard_curves_Nemotron-3-Ultra_s0.json2.4 KB
docs/assets/hard_curves_Nemotron-3-Ultra_s1.json2.4 KB
docs/assets/hard_curves_Nemotron-3-Ultra_s2.json2.5 KB
docs/assets/hard_curves_Nemotron-3-Ultra_s3.json2.5 KB
docs/assets/hard_curves_Nemotron-3-Ultra_s4.json2.5 KB
docs/assets/hard_curves_Opus-4.8_s0.json2.5 KB
docs/assets/hard_curves_Opus-4.8_s1.json2.5 KB
docs/assets/hard_curves_Opus-4.8_s2.json2.5 KB
docs/assets/hard_curves_Opus-4.8_s3.json2.5 KB
docs/assets/hard_curves_Opus-4.8_s4.json2.4 KB
docs/assets/hard_curves_Random_s0.json3.1 KB
docs/assets/hard_curves_Random_s1.json3.1 KB
docs/assets/hard_curves_Random_s2.json3.0 KB
docs/assets/hard_curves_Random_s3.json3.0 KB
docs/assets/hard_curves_Random_s4.json3.1 KB
docs/assets/hard_curves_Sonnet-5_s0.json2.4 KB
docs/assets/hard_curves_Sonnet-5_s1.json2.4 KB
docs/assets/hard_curves_Sonnet-5_s2.json2.4 KB
docs/assets/hard_curves_Sonnet-5_s3.json2.4 KB
docs/assets/hard_curves_Sonnet-5_s4.json2.4 KB
docs/assets/hard_curves_TPE_s0.json3.1 KB
docs/assets/hard_curves_TPE_s1.json3.0 KB
docs/assets/hard_curves_TPE_s2.json3.1 KB
docs/assets/hard_curves_TPE_s3.json3.1 KB
docs/assets/hard_curves_TPE_s4.json3.0 KB
docs/assets/lunarlander_landing.json1.4 KB
docs/assets/lunarlander_policy.gif462.7 KB
docs/assets/mnist_curves_GPT-5.5-medium_s0.json10.0 KB
docs/assets/mnist_curves_GPT-5.5-medium_s1.json10.0 KB
docs/assets/mnist_curves_GPT-5.5-medium_s2.json10.0 KB
docs/assets/mnist_curves_GPT-5.5-medium_s3.json10.0 KB
docs/assets/mnist_curves_GPT-5.5-medium_s4.json10.0 KB
docs/assets/mnist_curves_GPT-5.5-medium-no-context_s0.json10.8 KB
docs/assets/mnist_curves_GPT-5.5-medium-no-context_s1.json10.8 KB
docs/assets/mnist_curves_GPT-5.5-medium-no-context_s2.json10.7 KB
docs/assets/mnist_curves_GPT-5.5-medium-no-context_s3.json10.7 KB
docs/assets/mnist_curves_GPT-5.5-medium-no-context_s4.json10.8 KB
docs/assets/mnist_curves_Random_s0.json11.0 KB
docs/assets/mnist_curves_Random_s1.json11.1 KB
docs/assets/mnist_curves_Random_s2.json11.1 KB
docs/assets/mnist_curves_Random_s3.json11.0 KB
docs/assets/mnist_curves_Random_s4.json11.0 KB
docs/assets/mnist_curves_TPE_s0.json11.0 KB
docs/assets/mnist_curves_TPE_s1.json11.0 KB
docs/assets/mnist_curves_TPE_s2.json10.9 KB
docs/assets/mnist_curves_TPE_s3.json11.1 KB
docs/assets/mnist_curves_TPE_s4.json11.0 KB
docs/assets/optim-agent-logo-dark.svg4.5 KB
docs/assets/optim-agent-logo-light.svg4.5 KB
docs/assets/optim-agent-mark.svg633 B
docs/assets/optim-agent-overview.png90.6 KB
docs/assets/optimization_trajectory.gif493.7 KB
docs/assets/rl_control_GPT-5.5_s0.json6.9 KB
docs/assets/rl_control_GPT-5.5_s1.json6.9 KB
docs/assets/rl_control_GPT-5.5_s2.json6.9 KB
docs/assets/rl_control_GPT-5.5_s3.json6.9 KB
docs/assets/rl_control_GPT-5.5_s4.json7.0 KB
docs/assets/rl_control_GPT-5.5-no-context_s0.json4.9 KB
docs/assets/rl_control_GPT-5.5-no-context_s1.json5.0 KB
docs/assets/rl_control_GPT-5.5-no-context_s2.json4.8 KB
docs/assets/rl_control_GPT-5.5-no-context_s3.json4.9 KB
docs/assets/rl_control_GPT-5.5-no-context_s4.json4.9 KB
docs/assets/rl_control_Random_s0.json5.7 KB
docs/assets/rl_control_Random_s1.json5.7 KB
docs/assets/rl_control_Random_s2.json5.7 KB
docs/assets/rl_control_Random_s3.json5.8 KB
docs/assets/rl_control_Random_s4.json5.7 KB
docs/assets/rl_control_TPE_s0.json5.7 KB
docs/assets/rl_control_TPE_s1.json5.7 KB
docs/assets/rl_control_TPE_s2.json5.7 KB
docs/assets/rl_control_TPE_s3.json5.8 KB
docs/assets/rl_control_TPE_s4.json5.7 KB
docs/assets/rl_control.png77.8 KB
docs/assets/social-preview.png93.6 KB
docs/i18n/README_DE.md22.3 KB
docs/i18n/README_ES.md22.4 KB
docs/i18n/README_FR.md22.9 KB
docs/i18n/README_JA.md23.6 KB
docs/i18n/README_KO.md22.5 KB
docs/i18n/README_PT.md22.2 KB
docs/i18n/README_RU.md27.1 KB
docs/i18n/README_ZH.md20.8 KB
docs/index.html37.1 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry/ABORTED_RUN_MANIFEST.40b13e0d.json4.9 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry/EXECUTION_MANIFEST.json5.1 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry/WORKER_BLOCKED.TASK-001.json520 B
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry/WORKER_BLOCKED.TASK-001.reporting.py4.2 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry2/EXECUTION_MANIFEST.json5.6 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry3/EXECUTION_MANIFEST.json6.3 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging-retry4/EXECUTION_MANIFEST.json5.9 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging/PLAN_v1_reviewer_comments.md3.0 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging/PLAN_v1.md9.6 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging/PLAN_v2_reviewer_comments.md2.8 KB
docs/optim-plans/2026-07-27-summary-agent-and-table-logging/PLAN_v2.md12.4 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

optim-agent

Act as the sampler inside any coding-agent session: Claude Code, Codex, OpenCode/OpenClaw, or another agent that can read project files and run shell commands. Read the project to understand parameter meaning and interactions, propose one configuration, run the real evaluator, and record the result through optim-agent's ask/tell API. Let the measured objective, not the agent's intuition, decide what works.

Load the workflow

Use this file as the operating guide for the active coding agent. In Codex, it can be installed directly from GitHub:

$skill-installer install https://github.com/Optim-Agent/optim-agent

In Claude Code, OpenCode/OpenClaw, or another coding-agent environment, place this repository or SKILL.md in the agent-visible workspace and ask the agent to follow the optim-agent workflow. The workflow does not depend on Codex-only APIs; it needs file access, shell access, and Python.

Ensure the Python package is importable. Choose one source; do not install both:

# Stable release from PyPI
python -m pip install optim-agent

# Latest source from GitHub
python -m pip install "optim-agent @ git+https://github.com/Optim-Agent/optim-agent.git"

For a reproducible GitHub install, append @<tag-or-commit> after .git.

Workflow

Understand the system. Read the evaluation entry point and every file that defines the target parameters. Record each parameter's type, legal range, semantics, interactions, and operational constraints.
Define the experiment. Confirm the scalar objective, minimize or maximize, trial budget, evaluation command, runtime/cost limit, and fixed workload or seed. For multiple metrics or hard constraints, agree on one scalar feasibility or penalty rule before running trials.
Establish a baseline. Evaluate the current/default configuration with the same command and environment used for every later trial.

Initialize or resume. Keep artifacts in the repository's ignored .optim-agent-runs/ directory:

if git rev-parse --git-dir >/dev/null 2>&1 && ! git check-ignore -q .optim-agent-runs/; then
  printf '/.optim-agent-runs/\n' >> "$(git rev-parse --git-path info/exclude)"
fi

from pathlib import Path
import optim_agent as oa

run_dir = Path(".optim-agent-runs")
run_dir.mkdir(exist_ok=True)
study = oa.create_study(
    direction="minimize",
    storage=run_dir / "skill-study.json",
    seed=0,
)
print([(t.params, t.value, t.state) for t in study.trials])

Run one informed trial. Choose parameters from code understanding and all completed history, then use explicit ask/tell:
```
params = {"threshold": 0.72, "budget": 80}
trial = study.ask(params)
try:
    value = evaluate_system(**trial.params)
except Exception:
    study.tell(trial, state="failed")
    raise
else:
    study.tell(trial, value)
```
For a deliberately stopped trial, report the latest valid intermediate metric first, then call study.tell(trial, state="pruned").
Select the next point. Avoid accidental repeats, explore broadly before exploiting, respect bounds and constraints, and treat failed regions as evidence. If the evaluator is noisy, repeat promising configurations under the same workload before declaring a winner.
Stop and report. Stop at the approved budget or stopping condition. Report the baseline, best value and parameters, trial count, failed/pruned trials, convergence trend, and exact reproduction command.

Recovery

JSON storage records a trial when study.tell runs. Before launching an expensive external evaluation, save its parameters, command, and output path in a per-trial directory under .optim-agent-runs/. After interruption, inspect that output before rerunning: if a valid result exists, recreate the same point with study.ask(params) and record it; otherwise rerun it deliberately.

Use SQLite storage (skill-study.db) only when the user explicitly wants multiple processes. Sequential trials are the default because each proposal should use the complete prior history.

Rules

Use ask/tell in skill mode; do not delegate proposal selection to AgentSampler when the session agent is meant to read and reason over code.
Keep evaluation inputs and outputs isolated from production configuration.
Never fabricate, infer, or manually improve an objective value.
Record crashes as failed; record intentional early stops as pruned.
Preserve the study and trial artifacts so the result is auditable and resumable.
Do not tune secrets, credentials, or unbounded parameters.

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

같은 저장소의 스킬

optim-agent

무료

Use when optimizing configurable system parameters against a measurable scalar objective.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Optim-Agent/optim-agent8002026년 8월 15일 업데이트

Optim-Agent의 스킬 모두 보기

이 스킬의 문제 신고

설치

명령으로 설치

npx skills add https://github.com/Optim-Agent/optim-agent --skill optim-agent

터미널에 붙여 넣고 실행합니다(Node.js 필요). Claude Code, Codex 등 설치할 에이전트를 고를 수 있습니다.

에이전트에게 부탁하기

다음 Agent Skill을 설치해 주세요.
https://github.com/Optim-Agent/optim-agent

먼저 SKILL.md와 첨부 파일의 내용을 확인해 수상한 처리(외부 전송, 인증 정보 읽기 등)가 없는지 확인한 뒤 Claude Code는 ~/.claude/skills/optim-agent/, Codex는 ~/.agents/skills/optim-agent/ 에 배치해 주세요.

이 문장을 그대로 Claude Code나 Codex에 보내면 내용을 확인한 뒤 설치해 줍니다.

GitHub에서 소스 보기

저장소: Optim-Agent/optim-agent
스타: 800
포크: 46
카테고리: 리서치·학습
라이선스: MIT
원문 언어: 영어
최종 업데이트: 2026년 8월 15일

스킬은 에이전트를 위한 지시와 스크립트의 묶음입니다. cccskills는 내용의 안전성을 보증하지 않습니다. 설치하기 전에 내용을 확인하세요.