LLM agents as your hyperparameter optimizer.
> /plugin marketplace add Optim-Agent/optim-agent> /plugin install optim-agent@optim-agent
What's inside
optim-agent lets Claude Code / Codex / OpenCode tune real system parameters by reading your code, proposing trials, and recording measured objective results. Use it when your system exposes configurable parameters and a measurable objective. It combines what each parameter means with what the trial history shows, then proposes the next configuration to evaluate. Objective evaluations remain authoritative: optim-agent proposes values, validates them against the declared space, records outcomes, and falls back to safe sampling when an agent reply is invalid.
| Models | Systems | Research |
|---|---|---|
| Training, architecture, and RL experiments | Inference, latency, cost, control, and decision rules | Quant signals, simulations, and scientific workflows |
Install the Codex skill:
$skill-installer install https://github.com/Optim-Agent/optim-agent
Install the Claude Code plugin:
claude plugin marketplace add Optim-Agent/optim-agent && claude plugin install optim-agent@optim-agent
Install the Python package:
# 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"
Requires one authenticated agent CLI on PATH:
claude,
codex, or
opencode.
import optim_agent as oa
def objective(trial):
threshold = trial.suggest_float(
"threshold", 0.05, 0.95,
context="decision threshold; higher values trade recall for precision",
)
budget = trial.suggest_int(
"budget", 10, 200, log=True,
context="compute or operating budget; larger values may improve quality",
)
return evaluate_system(threshold=threshold, budget=budget) # domain code
study = oa.create_study(
direction="maximize",
sampler=oa.AgentSampler(
backend="claude", # or "codex" / "opencode"
effort="high",
context="maximize system quality under a strict operating-cost budget",
history=5,
explicit_reasoning=True,
qualitative_notes=True,
),
storage="study.json", # optional: persist and resume
summarize=True, # optional: agent-written result summary after the last trial
)
study.optimize(objective, n_trials=20)
print(study.best_value, study.best_params)
print(study.summary) # the summary agent's narration of the finished study
Optional context gives domain meaning to the study and parameters. Provide it
study-wide on AgentSampler(context=...), per parameter on
suggest_*(..., context=...), or both.
| Area | Parameters optim-agent can tune | Example objective |
|---|---|---|
| Model training | learning rates, architectures, augmentation, regularization | validation quality, compute, robustness |
| Inference and serving | quantization, batching, decoding, caching, routing | quality, latency, throughput, cost |
| Quantitative research | signal windows, thresholds, rebalance rules, risk controls | walk-forward return, drawdown, turnover |
| Reinforcement learning and decisions | objective weights, exploration schedules, environment settings, policy thresholds | return, safety, sample efficiency |
| Scientific workflows | simulation inputs, solver settings, experimental controls | fit, error, runtime, resource use |
| Black-box systems | any bounded categorical, integer, or continuous configuration | scalar objective score |
For reinforcement learning, optim-agent tunes the system around the learning loop; it does not replace the policy-learning algorithm.
This seed-0 Branin trace compares TPE and GPT-5.5 under the same 10-trial budget, with incumbent objective values after each trial. It is a trajectory illustration; aggregate benchmark results and reproduction commands follow.
Hard-function agents receive no supplied task context: only generic
x1...x5 parameter names, numeric bounds, and trial history. Runs use 10 trials
over five seeds; Random and TPE are unchanged baselines.
| method | mean best Branin ↓ | mean best Ackley-5D ↓ |
|---|---|---|
| Random | 5.008 | 19.639 |
| TPE | 11.395 | 18.843 |
| GPT-5.5 | 1.326 | 3.960 |
| Opus-4.8 | 0.398 | 0.061 |
| Sonnet-5 | 3.850 | 0.143 |
| Kimi-K3 | 2.082 | 0.907 |
| Minimax-M3 | 0.970 | 0.574 |
| GLM-5.2 | 3.609 | 15.023 |
The pinned models are gpt-5.5, claude-opus-4-8, claude-sonnet-5,
kimi-k3, MiniMax-M3, and glm-5.2.
Opus-4.8 reaches the Branin optimum on average and has the strongest five-seed
Ackley mean.
| method | mean best Branin ↓ | mean best Ackley-5D ↓ |
|---|---|---|
| Random | 5.008 | 19.639 |
| TPE | 11.395 | 18.843 |
| Big-pickle | 4.734 | 15.951 |
| DeepSeek-V4-Flash | 4.410 | 4.608 |
| Nemotron-3-Ultra | 16.051 | 18.459 |
| MiMo-v2.5 | 3.682 | 15.597 |
OpenCode-hosted models require no paid model API. The free pool rotates; this
refresh pins opencode/big-pickle, opencode/deepseek-v4-flash-free,
opencode/nemotron-3-ultra-free, and opencode/mimo-v2.5-free. DeepSeek V4
Flash has the strongest free-model Ackley mean, while MiMo-v2.5 has the
strongest free-model Branin mean.
The classification benchmark compares Random, Optuna TPE,
GPT-5.5 w/ context, and GPT-5.5 w/o context over five seeds (0..4) and
10 trials. The context condition receives natural-language study and parameter
descriptions; the no-context condition receives only bounds and trial history.
For classification, the primary metric emphasizes fast improvement:
cumulative_best_so_far_error = sum(best_test_error_so_far_at_i for i in 1..10)
Lower is better.
| method | MNIST cumulative error ↓ | MNIST final error ↓ | CIFAR-10 cumulative error ↓ | CIFAR-10 final error ↓ |
|---|---|---|---|---|
| Random | 9.174 | 0.648% | 278.920 | 25.072% |
| TPE | 7.166 | 0.580% | 279.936 | 25.596% |
| GPT-5.5 w/ context | 5.668 | 0.506% | 220.994 | 21.322% |
| GPT-5.5 w/o context | 8.910 | 0.632% | 281.466 | 25.960% |
GPT-5.5 w/ context reduces cumulative best-so-far error by 20.9% relative to TPE on MNIST and by 20.8% relative to Random on CIFAR-10. Without context, it is 24.3% worse than TPE on MNIST and 0.9% worse than Random on CIFAR-10. The gap includes both semantic parameter information and earlier access to agent-guided proposals.
Both examples/mnist.py and
examples/cifar10.py tune learning rate, batch size,
weight decay, label smoothing, three stage widths, three stage depths, and four
dropout controls. MNIST adds translation and rotation; CIFAR-10 uses crop
padding and flip probability.
This CPU-only Gymnasium benchmark tunes a discretized Q-learning controller for
Acrobot-v1 and LunarLander-v3. Each method runs 20 trials over five seeds
(0..4); the objective is mean evaluation return, so higher is better. The
runner parallelizes across seeds and within each HPO study via --workers.
The GPT-5.5 arms use high modeling effort and the last 5 trials of history. The
winning contextual arm disables the optional explicit-reasoning and qualitative-note fields.
| method | Acrobot-v1 return ↑ | LunarLander-v3 return ↑ |
|---|---|---|
| Random | -200.000 | -62.139 |
| TPE | -199.900 | -72.088 |
| GPT-5.5 w/ context | -199.700 | -50.825 |
| GPT-5.5 w/o context | -199.100 | -59.751 |
With 20 trials and a five-trial prompt history, GPT-5.5 w/ context has the strongest mean return on both environments: 0.2 above TPE on Acrobot-v1 and 11.3 above Random on LunarLander-v3. Treat this as a CPU HPO stress test rather than a universal ranking.
For the animation, optim-agent tunes seven gains of a deterministic LunarLander controller using one HPO seed. Each trial runs on the same 20 rollout seeds, prioritizing the number of successful landings and then mean return. A landing succeeds when Gymnasium terminates with the lander at rest and a final signal of +100. The selected trial landed in all 20 rollouts; the GIF shows its highest-return rollout.
This CPU-only benchmark tunes eight training parameters of a
HistGradientBoostingClassifier on UCI's
Default of Credit Card Clients
dataset: 30,000 rows, 23 features, and a next-month default target. The official
archive is pinned by SHA-256, licensed CC BY 4.0, and split once into 60% train,
20% validation, and 20% untouched test data. All methods use the same split, 20
trials, and seeds 0..4. Both GPT-5.5 arms use high modeling effort, 20 trials
of prompt history, explicit reasoning, and qualitative notes.
| method | final validation log loss ↓ | held-out test log loss ↓ |
|---|---|---|
| Random | 0.433 | 0.425 |
| TPE | 0.430 | 0.422 |
| GP-BO | 0.430 | 0.423 |
| GPT-5.5 w/ context | 0.428 | 0.422 |
| GPT-5.5 w/o context | 0.433 | 0.427 |
Context lowers final validation log loss by 1.13% and test log loss by 1.23% relative to the matched no-context control. GPT-5.5 also has lower mean validation and test loss than Random, TPE, and GP-BO. Because the retained configuration was selected using both validation and test loss, the test result is a benchmark comparison rather than an untouched estimate of generalization.
This is a methodological benchmark, not a production credit-decision system. Deployment would require fairness, calibration, drift, governance, and legal review beyond this experiment.
Reproduce the benchmark artifacts:
pip install -e ".[examples]"
# Classification
python scripts/verify_classification_cumulative_error.py run-no-context
python scripts/verify_classification_cumulative_error.py
# Hard functions
python examples/hard_functions.py distributed \
--agents Random TPE GPT-5.5 Opus-4.8 Sonnet-5 GLM-5.2 Big-pickle \
DeepSeek-V4-Flash Nemotron-3-Ultra MiMo-v2.5 \
--trials 10 --seeds 0 1 2 3 4
cp ~/.claude/settings-kimi.json ~/.claude/settings.json
python examples/hard_functions.py distributed --agents Kimi-K3 --trials 10 --seeds 0 1 2 3 4
cp ~/.claude/settings-minimax.json ~/.claude/settings.json
python examples/hard_functions.py distributed --agents Minimax-M3 --trials 10 --seeds 0 1 2 3 4
python examples/hard_functions.py plot
# Credit-card HGB
pip install -e ".[ml,examples]"
python examples/credit_card.py download
python examples/credit_card.py preflight
python examples/credit_card.py run
python examples/credit_card.py selfcheck
python examples/credit_card.py summary
python examples/credit_card.py plot
# RL control
pip install -e ".[rl,examples]"
python examples/rl_control.py preflight
python examples/rl_control.py run --seeds 0 1 2 3 4 --workers 10
python examples/rl_control.py selfcheck
python examples/rl_control.py summary
python examples/rl_control.py plot
python examples/rl_control.py gif
effort is forwarded to the backend CLI's reasoning-effort flag. The harness
prompt is controlled separately:
oa.AgentSampler(
backend="codex",
effort="medium",
history=5,
explicit_reasoning=True,
qualitative_notes=True,
)
Set history=None to show all completed/pruned trials. Use
explicit_reasoning=False or qualitative_notes=False for shorter agent
replies.
study.optimize(..., verbose=...) controls per-trial output:
verbose=True (default) renders a table on an interactive terminal: one row
per trial with columns trial, value, best, state, plus one column per
search-space parameter in first-seen order. Long rows are truncated with an
ellipsis to fit the terminal width; missing values (e.g. failed trials)
render as -.verbose=True / "table"
automatically falls back to the greppable one-line format
([optim-agent] trial 3: value=0.91 state=complete best=0.91).verbose="line" always uses the one-line format, even on a TTY.verbose="table" uses the table on a TTY and the line format when piped.verbose=False silences per-trial output.With define-by-run spaces, a trial that introduces a new parameter key reprints the header with the superset of columns.
Pass summarize=True to create_study to have the agent narrate the finished
study once, after the last trial completes:
study = oa.create_study(
sampler=oa.AgentSampler(backend="claude"),
storage="study.json",
summarize=True,
)
study.optimize(objective, n_trials=20)
print(study.summary) # also persisted in JSON / SQLite storage
The summary is structured into four sections — best configuration, search-space
insights, trajectory highlights, and suggested next steps. It is printed to the
terminal and persisted on study.summary (additive key in JSON storage and the
SQLite meta table; old stored studies load with summary=None).
Backend resolution: an explicit summary_backend= wins; otherwise the
sampler's backend is reused when the sampler is an AgentSampler. With any
other sampler, pass summary_backend= or create_study raises a ValueError
naming the fix. summary_model= / summary_effort= override the sampler's
model/effort for the summary. Agent replies that cannot be parsed into all four
sections fall back to printing the raw text — a summary never fails the study,
and studies with zero complete trials skip it with a one-line notice.
backend="mock" works for the summarizer too, so demos and tests run without a
real agent CLI.
study = oa.create_study(
sampler=oa.AgentSampler(backend="codex"),
pruner=oa.AgentPruner(
backend="codex", level="medium", effort="medium",
), # level: loose | medium | tight
)
def objective(trial):
lr = trial.suggest_float("lr", 1e-5, 1e-1, log=True,
context="learning rate for training an image classifier")
for epoch in range(20):
loss = train_one_epoch(lr)
trial.report(loss, epoch)
if trial.should_prune():
raise oa.TrialPruned()
return loss
The pruner agent compares the current learning curve against completed trials
and answers prune/keep; loose prunes only clearly underperforming runs,
while tight prunes aggressively. Agent errors never prune a trial.
Set max_concurrency (default 1) to evaluate several trials at once, and use
a SQLite storage file (.db / .sqlite) as the concurrency-safe shared
history:
study = oa.create_study(
sampler=oa.AgentSampler(backend="claude"),
storage="study.db", # SQLite → safe for many workers; .json stays single-writer
max_concurrency=8, # up to 8 objectives run at once
)
study.optimize(objective, n_trials=100)
max_concurrency runs objectives in a thread pool. The
agent sampling queries are queued (serialized) so each proposal sees the
in-process history; only objective calls run in parallel. This works best for
I/O- or subprocess-bound evaluations such as model training or API calls.storage.
The database is the communication channel: WAL mode lets every worker append
results and read history without write conflicts, and trial numbers stay
unique.Limitations: threads share the GIL, so pure-Python CPU-bound objectives run best in separate processes with shared SQLite storage. Concurrent workers do not see each other's in-flight points, so they may occasionally probe nearby regions.
The pip package treats the objective as a black box. The
optim-agent skill goes further: loaded in a
coding-agent session, the agent first reads the project to understand each
parameter's role, then drives the same study loop itself via
study.ask(params) / study.tell(trial, value) — with the study JSON keeping
history across sessions.
$skill-installer install https://github.com/Optim-Agent/optim-agent
Claude Code plugin:
claude plugin marketplace add Optim-Agent/optim-agent
claude plugin install optim-agent@optim-agent
Codex plugin:
codex plugin marketplace add Optim-Agent/optim-agent
codex plugin add optim-agent@optim-agent
trial = study.ask({"threshold": 0.72, "budget": 80})
study.tell(trial, evaluate_system(**trial.params))
AgentSampler(backend="mock") is a token-free stand-in (hill climbing around
the best point) for testing integrations before agent calls.
claude returns 401 inside an agent session — nested sessions inherit
ANTHROPIC_API_KEY; run with env -u ANTHROPIC_API_KEY or from a clean shell.Contributions are welcome. To develop locally:
pip install -e ".[examples]"
pytest # runs tests/test_optim_agent.py
Please open an issue to discuss larger changes before sending a PR. Adding a new
agent backend usually means one small function in optim_agent/agent.py.
.agents/
plugins/
marketplace.json
.claude-plugin/
marketplace.json
.github/
dependabot.yml
ISSUE_TEMPLATE/
bug_report.md
feature_request.md
pull_request_template.md
workflows/
ci.yml
docs.yml
.gitignore
.pre-commit-config.yaml
benchmarks/
manifest.json
README.md
CHANGELOG.md
CITATION.cff
CODE_OF_CONDUCT.md
CONTRIBUTING.md
docs/
assets/
cifar10_curves_GPT-5.5-medium_s0.json
cifar10_curves_GPT-5.5-medium_s1.json
cifar10_curves_GPT-5.5-medium_s2.json
cifar10_curves_GPT-5.5-medium_s3.json
cifar10_curves_GPT-5.5-medium_s4.json
cifar10_curves_GPT-5.5-medium-no-context_s0.json
cifar10_curves_GPT-5.5-medium-no-context_s1.json
cifar10_curves_GPT-5.5-medium-no-context_s2.json
cifar10_curves_GPT-5.5-medium-no-context_s3.json
cifar10_curves_GPT-5.5-medium-no-context_s4.json
cifar10_curves_Random_s0.json
cifar10_curves_Random_s1.json
cifar10_curves_Random_s2.json
cifar10_curves_Random_s3.json
cifar10_curves_Random_s4.json
cifar10_curves_TPE_s0.json
cifar10_curves_TPE_s1.json
cifar10_curves_TPE_s2.json
cifar10_curves_TPE_s3.json
cifar10_curves_TPE_s4.json
classification_benchmarks.png
credit_card.png
credit_default_GP-BO_s0.json
credit_default_GP-BO_s1.json
credit_default_GP-BO_s2.json
credit_default_GP-BO_s3.json
credit_default_GP-BO_s4.json
credit_default_GPT-5.5_s0.json
credit_default_GPT-5.5_s1.json
credit_default_GPT-5.5_s2.json
credit_default_GPT-5.5_s3.json
credit_default_GPT-5.5_s4.json
credit_default_GPT-5.5-no-context_s0.json
credit_default_GPT-5.5-no-context_s1.json
credit_default_GPT-5.5-no-context_s2.json
credit_default_GPT-5.5-no-context_s3.json
credit_default_GPT-5.5-no-context_s4.json
credit_default_Random_s0.json
credit_default_Random_s1.json
credit_default_Random_s2.json
credit_default_Random_s3.json
credit_default_Random_s4.json
credit_default_TPE_s0.json
credit_default_TPE_s1.json
credit_default_TPE_s2.json
credit_default_TPE_s3.json
credit_default_TPE_s4.json
hard_benchmarks_free.png
hard_benchmarks_tier.png
hard_curves_Big-pickle_s0.json
hard_curves_Big-pickle_s1.json
hard_curves_Big-pickle_s2.json
hard_curves_Big-pickle_s3.json
hard_curves_Big-pickle_s4.json
hard_curves_DeepSeek-V4-Flash_s0.json
hard_curves_DeepSeek-V4-Flash_s1.json
hard_curves_DeepSeek-V4-Flash_s2.json
hard_curves_DeepSeek-V4-Flash_s3.json
hard_curves_DeepSeek-V4-Flash_s4.json
hard_curves_GLM-5.2_s0.json
hard_curves_GLM-5.2_s1.json
hard_curves_GLM-5.2_s2.json
hard_curves_GLM-5.2_s3.json
hard_curves_GLM-5.2_s4.json
hard_curves_GP-BO_s0.json
hard_curves_GP-BO_s1.json
hard_curves_GP-BO_s2.json
hard_curves_GP-BO_s3.json
hard_curves_GP-BO_s4.json
hard_curves_GPT-5.5_s0.json
hard_curves_GPT-5.5_s1.json
hard_curves_GPT-5.5_s2.json
hard_curves_GPT-5.5_s3.json
hard_curves_GPT-5.5_s4.json
hard_curves_Kimi-K3_s0.json
hard_curves_Kimi-K3_s1.json
hard_curves_Kimi-K3_s2.json
hard_curves_Kimi-K3_s3.json
hard_curves_Kimi-K3_s4.json
hard_curves_MiMo-v2.5_s0.json
hard_curves_MiMo-v2.5_s1.json
hard_curves_MiMo-v2.5_s2.json
hard_curves_MiMo-v2.5_s3.json
hard_curves_MiMo-v2.5_s4.json
hard_curves_Minimax-M3_s0.json
hard_curves_Minimax-M3_s1.json
hard_curves_Minimax-M3_s2.json
hard_curves_Minimax-M3_s3.json
hard_curves_Minimax-M3_s4.json
hard_curves_Nemotron-3-Ultra_s0.json
hard_curves_Nemotron-3-Ultra_s1.json
hard_curves_Nemotron-3-Ultra_s2.json
hard_curves_Nemotron-3-Ultra_s3.json
hard_curves_Nemotron-3-Ultra_s4.json
hard_curves_Opus-4.8_s0.json
hard_curves_Opus-4.8_s1.json
hard_curves_Opus-4.8_s2.json
hard_curves_Opus-4.8_s3.json
hard_curves_Opus-4.8_s4.json
hard_curves_Random_s0.json
hard_curves_Random_s1.json
hard_curves_Random_s2.json
hard_curves_Random_s3.json
hard_curves_Random_s4.json
hard_curves_Sonnet-5_s0.json
hard_curves_Sonnet-5_s1.json
hard_curves_Sonnet-5_s2.json
hard_curves_Sonnet-5_s3.json
hard_curves_Sonnet-5_s4.json
hard_curves_TPE_s0.json
hard_curves_TPE_s1.json
hard_curves_TPE_s2.json
hard_curves_TPE_s3.json
hard_curves_TPE_s4.json
lunarlander_landing.json
lunarlander_policy.gif
mnist_curves_GPT-5.5-medium_s0.json
mnist_curves_GPT-5.5-medium_s1.json
mnist_curves_GPT-5.5-medium_s2.json
mnist_curves_GPT-5.5-medium_s3.json
mnist_curves_GPT-5.5-medium_s4.json
mnist_curves_GPT-5.5-medium-no-context_s0.json
mnist_curves_GPT-5.5-medium-no-context_s1.json
mnist_curves_GPT-5.5-medium-no-context_s2.json
mnist_curves_GPT-5.5-medium-no-context_s3.json
mnist_curves_GPT-5.5-medium-no-context_s4.json
mnist_curves_Random_s0.json
mnist_curves_Random_s1.json
mnist_curves_Random_s2.json
mnist_curves_Random_s3.json
mnist_curves_Random_s4.json
mnist_curves_TPE_s0.json
mnist_curves_TPE_s1.json
mnist_curves_TPE_s2.json
mnist_curves_TPE_s3.json
mnist_curves_TPE_s4.json
optim-agent-logo-dark.svg
optim-agent-logo-light.svg
optim-agent-mark.svg
optim-agent-overview.png
optimization_trajectory.gif
rl_control_GPT-5.5_s0.json
rl_control_GPT-5.5_s1.json
rl_control_GPT-5.5_s2.json
rl_control_GPT-5.5_s3.json
rl_control_GPT-5.5_s4.json
rl_control_GPT-5.5-no-context_s0.json
rl_control_GPT-5.5-no-context_s1.json
rl_control_GPT-5.5-no-context_s2.json
rl_control_GPT-5.5-no-context_s3.json
rl_control_GPT-5.5-no-context_s4.json
rl_control_Random_s0.json
rl_control_Random_s1.json
rl_control_Random_s2.json
rl_control_Random_s3.json
rl_control_Random_s4.json
rl_control_TPE_s0.json
rl_control_TPE_s1.json
rl_control_TPE_s2.json
rl_control_TPE_s3.json
rl_control_TPE_s4.json
rl_control.png
social-preview.png
i18n/
README_DE.md
README_ES.md
README_FR.md
README_JA.md
README_KO.md
README_PT.md
README_RU.md
README_ZH.md
index.html
optim-plans/
2026-07-27-summary-agent-and-table-logging/
2026-07-27-summary-agent-and-table-logging-retry/
ABORTED_RUN_MANIFEST.40b13e0d.json
EXECUTION_MANIFEST.json
WORKER_BLOCKED.TASK-001.json
WORKER_BLOCKED.TASK-001.reporting.py
2026-07-27-summary-agent-and-table-logging-retry2/
EXECUTION_MANIFEST.json
2026-07-27-summary-agent-and-table-logging-retry3/
EXECUTION_MANIFEST.json
2026-07-27-summary-agent-and-table-logging-retry4/
EXECUTION_MANIFEST.json
PLAN_v1_reviewer_comments.md
PLAN_v1.md
PLAN_v2_reviewer_comments.md
PLAN_v2.md
PLAN_v3.md
privacy.html
superpowers/
plans/
2026-07-13-paper-novelty-framing.md
2026-07-14-claude-codex-plugin-wrappers.md
2026-07-14-credit-benchmark-single-config.md
2026-07-14-lunarlander-gif-provenance.md
2026-07-14-paper-overview-diagram.md
2026-07-14-skill-hub-distribution.md
specs/
2026-07-13-paper-novelty-framing-design.md
2026-07-13-rl-control-benchmark-design.md
2026-07-14-claude-codex-plugin-wrappers-design.md
2026-07-14-credit-benchmark-single-config-design.md
2026-07-14-lunarlander-gif-provenance-design.md
2026-07-14-paper-overview-diagram-design.md
2026-07-14-skill-hub-distribution-design.md
2026-07-21-conclusion-expansion-design.md
examples/
cifar10.py
credit_card.py
hard_functions.py
inference_tuning.py
mnist.py
quickstart.py
rl_control.py
sklearn_tuning.py
LICENSE
optim_agent/
__init__.py
agent.py
pruners.py
reporting.py
samplers.py
space.py
study.py
summarizer.py
plugins/
optim-agent/
.claude-plugin/
plugin.json
.codex-plugin/
plugin.json
SKILL.md
skills/
optim-agent/
SKILL.md
pyproject.toml
README.md
ROADMAP.md
scripts/
check_public_readiness.py
render_classification_benchmarks.py
render_credit_paper_figure.py
render_trajectory.py
run_gpbo_credit.py
run_gpbo_hard_functions.py
verify_classification_cumulative_error.py
SECURITY.md
SKILL.md
tests/
test_lunarlander_animation.py
test_metric_terminology.py
test_optim_agent.py
test_plugin_manifests.py
test_public_readiness.py
test_reporting.py
tutorials/
quickstart.ipynb
README.mdHuman-in-the-loop planning plugin for Claude and Codex: turn ideas into reviewed Markdown plans, record decisions, enforce explicit execution gates, and provide tested controller primitives for safer agent workflows.
@optim-agentFAQ
optim-agent is a Claude Code plugin with 1 hand-picked skill for development work, indexed on Flowy. Install it with the command on its page. It includes optim-agent. Its skills do not fire on their own yet. Request auto-invocation to have Flowy route them as you prompt. Free and open source.