New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Corl Experiments

by brycewang-stanford in brycewang-stanford/Awesome-Journal-Skills

A skill your agent uses when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence…

MITAuto-check passedData & Analytics

Install Corl Experiments

skills CLI
$ npx skills add brycewang-stanford/Awesome-Journal-Skills --skill corl-experiments -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install brycewang-stanford/Awesome-Journal-Skills corl-experiments --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/brycewang-stanford/Awesome-Journal-Skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/CoRL-Skills/skills/corl-experiments .claude/skills/corl-experiments && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
corl-experiments
GitHub stars
1.2k
Token cost
~1.7k tokens
SKILL.md length
768 words
Files
1
Skills in repo
2,387
Repo updated
First seen
Licence
MIT

At a glance

A skill your agent uses when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence…

  • Auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts
  • SKILL.md covers Match the evidence to the…, The two-layer randomness…, Sim, real, and the space between and Baseline fairness across…, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
  • Task-suite breadth

What it does

Corl Experiments is an agent skill from brycewang-stanford/Awesome-Journal-Skills. Use when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence, sim-to-real gap measurement, baseline fairness across BC/RL/VLA families, generalization splits, and statistics for success-rate claims.

Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Data & Analytics, covering Statistics. The repository describes itself as: Journal-specific Claude Code/Codex skill packs covering mainstream journals — AER, QJE, Nature, Cell, 管理世界, 经济研究 & 200+ more — your fast track to getting published. ｜ 覆盖主流期刊的… The licence is MIT.

When your agent uses it

  • Auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts
  • Task-suite breadth
  • Real-robot versus simulation evidence
  • Sim-to-real gap measurement

Example prompts

  • “/corl-experiments”

Requirements

  • Python 3

What it can do on your machine

Read from SKILL.md and the folder at commit 932eb23. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are python).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Corl Experiments loads about 1.7k tokens when it runs. Until then it costs about 82 tokens; SKILL.md has 768 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~82
When it runs · the whole SKILL.md, loaded when a task matches
~1.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from brycewang-stanford/Awesome-Journal-Skills at commit 932eb23, republished under its MIT licence (© brycewang-stanford). 768 words, ~1,744 tokens.

Download SKILL.mdSave it as .claude/skills/corl-experiments/SKILL.md (or your agent's skills folder).
name
corl-experiments
description
Use when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence, sim-to-real gap measurement, baseline fairness across BC/RL/VLA families, generalization splits, and statistics for success-rate claims.

CoRL Experiments

At CoRL the object under evaluation is a learned policy, which makes the evidence problem statistical twice over: training is stochastic (seeds, data order, initialization) and execution is stochastic (initial states, physics, sensor noise). An experimental design that controls only one of the two is the most common weakness this reviewer pool writes up.

Match the evidence to the claim, not the venue

Claim in the paperMinimum credible evidence shape
"Method X learns task family T"Multiple training seeds; per-task success over many scripted-reset episodes
"X outperforms baseline Y"Same data, same evaluation protocol, same tuning effort for both; dispersion reported
"X transfers sim-to-real"The same checkpoint evaluated in sim and on hardware; the gap reported as a number
"X generalizes to novel objects/scenes/instructions"Held-out splits defined before training; per-split breakdown, not a pooled average
"X scales with data"≥3 dataset sizes on the same axis; no two-point "trends"
"X runs in real time on the robot"Latency/frequency measured on the deployed compute, stated with hardware

The routing consequence: if none of your claims require the last four rows, ask whether the paper is CoRL-shaped at all (corl-topic-selection).

The two-layer randomness protocol

  • Training layer: train ≥3 seeds (5 where budget allows) per method per configuration. Report the spread across seeds — a method whose best seed wins but whose median loses has not demonstrated superiority.
  • Evaluation layer: for each trained policy, evaluate over a fixed, scripted set of initial conditions — in simulation, dozens to hundreds of episodes per task is cheap and expected; on hardware, 10–25 trials per task per policy is typical practice, with the success criterion written down verbatim.
  • Never mix the layers in reporting: "80% success" must decompose into "mean over k seeds of per-seed success over n episodes," and the paper states k and n in the table caption, not only in the appendix.
python
# Evaluation bookkeeping: per-seed success with a binomial interval,
# then dispersion across seeds — the two layers stay separate.
import numpy as np
from scipy import stats

def summarize(results):           # results[seed] = list of 0/1 episode outcomes
    per_seed = {}
    for seed, eps in results.items():
        n, k = len(eps), int(np.sum(eps))
        lo, hi = stats.beta.ppf([0.025, 0.975], k + 1, n - k + 1)  # Jeffreys-ish CI
        per_seed[seed] = dict(rate=k / n, n=n, ci=(lo, hi))
    rates = [v["rate"] for v in per_seed.values()]
    return per_seed, dict(mean=np.mean(rates), sd=np.std(rates), seeds=len(rates))

Small-n hardware caveat: with 15 trials, a 73% vs 60% difference is not resolvable — either add trials, aggregate over tasks with a paired design, or soften the comparative language.

Sim, real, and the space between

  • Declare each experiment's regime in the table itself (sim / real / sim-trained real-evaluated). Reviewers at this venue actively hunt for regime laundering — headline numbers from sim standing in for a "real-world" abstract claim.
  • If the paper's story includes transfer, the sim-to-real delta is a result, not an embarrassment: evaluate the identical checkpoint in both regimes on matched task instances and print the gap. A measured 20-point drop with analysis outranks an unmeasured claim of robustness.
  • Describe the reality of hardware evaluation: reset procedure (human or scripted), object pose randomization method, stopping rule, and any trials excluded — exclusions disclosed with cause, never silently.
  • Simulation-only papers survive at CoRL when the sim is a recognized benchmark and the claims stay inside it; say why the simulator is adequate for the claim (contact fidelity, sensor models, prior validated transfer).
Show full SKILL.md (288 more words)Show less

Baseline fairness across method families

Robot-learning baselines span imitation (BC, diffusion policies), offline/online RL, and pretrained VLA models — families with wildly different data appetites:

  • Give every family its natural input: comparing your method-with-demos against an RL baseline denied demos measures data access, not algorithms. Either equal data or an explicit data-budget axis.
  • Tune baselines with the same effort budget you spent on your method and say so ("each method: 24 GPU-hours of search over its authors' recommended grid").
  • Pin baseline provenance: re-implementation vs official code vs released checkpoint — each is a different evidentiary object; name which one each row is.
  • Include one strong recent robot-learning baseline (not only classical control), because this reviewer pool benchmarks your table against the current PMLR volume, and one simple sanity baseline (scripted policy, nearest-neighbor over demos) to calibrate task difficulty.

Generalization splits that survive scrutiny

  • Freeze train/held-out splits (objects, scenes, language instructions, layouts) before training and publish the split lists in the supplementary.
  • Report per-axis breakdowns: "novel object, seen scene" and "seen object, novel scene" are different claims; a pooled number hides which one failed.
  • For language-conditioned policies, separate paraphrase-level from task-level novelty — reviewers with VLA experience will ask.

Ablations and the Limitations coupling

  • Ablate the components you advertise: each named contribution in the intro gets a row showing the system without it.
  • Feed the failures you observe into the mandatory Limitations section (corl-writing-style); a Limitations section that matches the failure cases in your video reads as credible, and one that contradicts them reads as concealment.

Design-review checklist

text
[ ] Every abstract-level claim mapped to a table/figure with regime declared
[ ] k seeds x n episodes stated per cell; two randomness layers separated
[ ] Hardware protocol written: resets, success criterion, stopping rule, exclusions
[ ] Same-checkpoint sim/real pairing for any transfer claim; gap printed
[ ] Baselines: fair data, disclosed tuning, pinned provenance, one recent + one simple
[ ] Splits frozen pre-training; per-axis generalization breakdown
[ ] Compute + data volumes reported (GPU-hours, demo counts, env steps)

Evidence norms here are community culture rather than a posted rulebook — they move each year with the field. Calibrate against the newest PMLR volume (v305 = CoRL 2025) and the current reviewer instructions at corl.org.

© brycewang-stanford, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in CoRL-Skills/skills/corl-experiments of brycewang-stanford/Awesome-Journal-Skills.

Open the folder on GitHubat commit 932eb23

Compare with similar skills

Corl Experiments next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Corl Experiments compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Corl Experiments this skillbrycewang-stanford/Awesome-Journal-Skills1.2k—~1.7kAutomated safety check: PassMIT
Sandbox Benchvercel/next.js143k—~4.1kAutomated safety check: PassMIT
Statistical Analysisspacering-net/codeg3.9k3 repos~5kAutomated safety check: PassMIT
StatsmodelszLanqing/codex-claude-academic-skills4.7k15 repos~4.9kAutomated safety check: PassBSD-3-Clause
AI Daily DigestvigorX777/ai-daily-digest1.6k—~1.3kAutomated safety check: PassNone
Statistical Powerspacering-net/codeg3.9k1 repos~3.6kAutomated safety check: NotesMIT

Similar skills

  • Sandbox Bench

    vercel/next.js

    Official

    Benchmark React or Next.js changes on Vercel Sandbox VMs with paired A/B statistics: react PR/commit vs base, or Next.js PR/commit vs base, measured end-to-end through the bench/render-pipeline app…

    143k GitHub stars~4.1k tokensUpdated today
    Data & AnalyticsAuto-check passed

  • Statistical Analysis

    spacering-net/codeg

    Guided statistical analysis for research data - test selection, assumption checking, effect sizes, power analysis, Bayesian alternatives, and APA-formatted reporting.

    3.9k GitHub starsUsed in 3 repos~5k tokens
    Data & AnalyticsAuto-check passed

  • Statsmodels

    zLanqing/codex-claude-academic-skills

    Statistical models library for Python. An agent skill from zLanqing/codex-claude-academic-skills.

    4.7k GitHub starsUsed in 15 repos~4.9k tokens
    Data & AnalyticsAuto-check passed

  • AI Daily Digest

    vigorX777/ai-daily-digest

    Fetches RSS feeds from 90 top Hacker News blogs (curated by Karpathy), uses AI to score and filter articles, and generates a daily digest in Markdown with Chinese-translated titles, category…

    1.6k GitHub stars~1.3k tokensUpdated 7 mo ago
    Data & AnalyticsAuto-check passed

  • Statistical Power

    spacering-net/codeg

    Sample-size and statistical power calculations for planning studies.

    3.9k GitHub starsUsed in 1 repo~3.6k tokens
    Data & AnalyticsAuto-check: notes

  • Agent Session Monitor

    higress-group/higress

    Real-time agent conversation monitoring - monitors Higress access logs, aggregates conversations by session, tracks token usage.

    9.5k GitHub stars~3.3k tokensUpdated today
    Data & AnalyticsAuto-check passed

More from brycewang-stanford/Awesome-Journal-Skills

All 2,387 skills in this repo

  • Aaag Data Analysis

    brycewang-stanford/Awesome-Journal-Skills

    A skill your agent uses when running and reporting the analysis for an Annals of the American Association of Geographers manuscript — spatial statistics and modeling, remote-sensing accuracy, or…

    1.2k GitHub stars~1.3k tokensUpdated 12 days ago
    Auto-check passed

  • Aaag Literature Positioning

    brycewang-stanford/Awesome-Journal-Skills

    A skill your agent uses when positioning an Annals of the American Association of Geographers manuscript in the literature — engaging geographic scholarship across the relevant area and the…

    1.2k GitHub stars~1.3k tokensUpdated 12 days ago
    Auto-check passed

  • Aaag Rebuttal

    brycewang-stanford/Awesome-Journal-Skills

    A skill your agent uses when responding to an Annals of the American Association of Geographers decision letter (major/minor revision) — building a point-by-point response to the subject editor and…

    1.2k GitHub stars~1.4k tokensUpdated 12 days ago
    Auto-check passed

  • Aaag Research Design

    brycewang-stanford/Awesome-Journal-Skills

    A skill your agent uses when defending the research design of an Annals of the American Association of Geographers manuscript — spatial/quantitative analysis and GIScience, remote-sensing and…

    1.2k GitHub stars~1.4k tokensUpdated 12 days ago
    Auto-check passed

  • Aaag Review Process

    brycewang-stanford/Awesome-Journal-Skills

    A skill your agent uses when you need to understand how the Annals of the American Association of Geographers evaluates a manuscript — double-anonymous review routed through a subject editor by…

    1.2k GitHub stars~1.3k tokensUpdated 12 days ago
    Auto-check passed

  • Aaag Submission

    brycewang-stanford/Awesome-Journal-Skills

    A skill your agent uses when running the final pre-submission preflight for the Annals of the American Association of Geographers via ScholarOne Manuscripts — area/article-type selection…

    1.2k GitHub stars~1.6k tokensUpdated 12 days ago
    Auto-check passed

Categories

Questions about Corl Experiments

What does Corl Experiments do?

A skill your agent uses when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence…. Corl Experiments is an agent skill from brycewang-stanford/Awesome-Journal-Skills. Use when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence, sim-to-real gap measurement, baseline fairness across BC/RL/VLA families, generalization splits, and statistics for success-rate claims.

When should I use Corl Experiments?

Corl Experiments fits situations like: auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts; task-suite breadth; real-robot versus simulation evidence; sim-to-real gap measurement.

How do I install Corl Experiments in Claude Code?

Run `npx skills add brycewang-stanford/Awesome-Journal-Skills --skill corl-experiments -a claude-code`. Or copy the skill folder (CoRL-Skills/skills/corl-experiments in brycewang-stanford/Awesome-Journal-Skills) into .claude/skills/corl-experiments in your project. Claude Code loads it when a task matches its description.

How do I install Corl Experiments in Codex?

Run `npx skills add brycewang-stanford/Awesome-Journal-Skills --skill corl-experiments -a codex`. Or copy the skill folder (CoRL-Skills/skills/corl-experiments in brycewang-stanford/Awesome-Journal-Skills) into .agents/skills/corl-experiments in your project. Codex loads it when a task matches its description.

Can I use Corl Experiments in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add brycewang-stanford/Awesome-Journal-Skills --skill corl-experiments -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/corl-experiments, .gemini/skills/corl-experiments, .github/skills/corl-experiments and .opencode/skills/corl-experiments in your project.

What does Corl Experiments need to run?

SKILL.md names no scripts, command-line tools or credentials: Corl Experiments is instructions for the agent only. Our summary lists: Python 3.

Does Corl Experiments access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Corl Experiments safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Corl Experiments use?

Corl Experiments is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Corl Experiments use?

About 1.7k tokens (SKILL.md is roughly 7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Corl Experiments?

Skills that share tags, products or a category with Corl Experiments: Sandbox Bench (vercel/next.js, 143k stars), Statistical Analysis (spacering-net/codeg, 3.9k stars), Statsmodels (zLanqing/codex-claude-academic-skills, 4.7k stars) and AI Daily Digest (vigorX777/ai-daily-digest, 1.6k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Corl Experiments?

brycewang-stanford (a GitHub user) maintains it in brycewang-stanford/Awesome-Journal-Skills, which has 1,228 GitHub stars. The repository holds 2,387 skills in this directory. The repository was last updated on September 27, 2026.

Source: brycewang-stanford/Awesome-Journal-Skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.