pandas-experiment-management

pandas-experiment-management is a skill for Claude Code from hajibabaie/combinatorial-optimization-skills. It costs 140 tokens per session (9,174 once invoked), scanned A, original, MIT.

A guide to storing optimization experiment results in pandas tables, with one row for each run. It covers recording run details, writing CSV or Parquet files safely, combining results, and making summary tables.

In plain words
What is it for?
Use it to record algorithms, problem instances, settings, seeds, runtimes, and objectives, then compare results across instances and repeated runs.
Why use it?
It keeps large experiment campaigns organized and recoverable, including runs made in parallel or interrupted by a crash.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the combinatorial-optimization plugin — 76 skills shipped together

Good fit Use it to record algorithms, problem instances, settings, seeds, runtimes, and objectives, then compare results across instances and repeated runs.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add hajibabaie/combinatorial-optimization-skills --skill pandas-experiment-management
Clone the repo
git clone --depth 1 https://github.com/hajibabaie/combinatorial-optimization-skills

Made for: Claude Code.

Or install combinatorial-optimization, the plugin that ships this one along with the rest of its 76 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for pandas-experiment-management

README.md
[![agentmods](https://agentmods.dev/badge/skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management/github.svg)](https://agentmods.dev/skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management)
Your own site
<a href="https://agentmods.dev/skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management"><img src="https://agentmods.dev/badge/skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for pandas-experiment-management

Your own site · 80×15
<a href="https://agentmods.dev/skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management"><img src="https://agentmods.dev/badge/skills/hajibabaie/combinatorial-optimization-skills/pandas-experiment-management.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 140 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 9,174 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00140 $0.09174
Opus 5 $0.00070 $0.04587
Sonnet 5 $0.00028 $0.01835
Haiku 4.5 $0.00014 $0.00917

Measured 11d ago against content hash 93b9fa89a058, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

pandas-experiment-management scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

`subprocess.run(["git", "rev-parse", "--short", "HEAD"], capture_output=True, text=True)`, the
skills/pandas-experiment-management/SKILL.md · 800 lines

How it starts

The opening of the file, as written. The whole thing — 800 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Pandas Experiment Management

You are an expert in managing computational-experiment data for combinatorial optimization research. This skill covers tidy result tables (one row per run), run-metadata capture, atomic CSV/parquet writing, aggregation across instances and seeds, and pivot tables ready for papers. Use the pattern catalog below to build a results pipeline that survives crashes, parallel workers, and reviewer questions — and that turns thousands of raw runs into one table you can defend.

Initial Assessment

Establish these facts before recommending a results pipeline:

  • Campaign size. Count expected rows: instances × algorithms × configurations × seeds. A 3-algorithm, 30-instance, 10-seed study is 900 rows (CSV is fine); a tuning campaign with 500 configurations is 150,000 rows (parquet, partitioning).
  • Run cost. Seconds per run or hours per run? Expensive runs make crash-safe writing and resume logic mandatory, not optional.
  • Parallelism. Single process, multiprocessing pool, or cluster array jobs writing to a shared filesystem? This decides the write strategy (one file per run vs. one shared file).
  • What is recorded per run. Final objective only, or also the incumbent trace over time? Traces need their own table (long format), never list-valued cells in the runs table.
  • Optimization sense. Minimization or maximization? Mixed across problems? Store the raw objective plus a sense column; convert only at aggregation time.
  • Reference values. Are best-known solutions (BKS) available for gap computation, or is the reference the best value found inside the campaign itself?
  • Failure modes. Can runs time out, crash, or end infeasible? The schema needs a status column from day one; retrofitting it later contaminates every aggregate already computed.
  • Downstream consumers. Statistical tests, convergence plots, LaTeX tables for a paper, or all three? The tidy runs table must serve all of them without re-running experiments.
  • Storage stack. Is pyarrow available for parquet? Is the filesystem local or networked (atomicity of os.replace holds within one filesystem only)?
  • Existing data. Is there a legacy spreadsheet or ad-hoc CSV to migrate? Migrate once, into the schema below, and freeze the old files as read-only.

Read the full file on GitHub · 800 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 11d ago First seen · 800 lines · 140 tokens per session scan A 93b9fa89a058

Subscribe to this mod's changes

pandas-experiment-management is a skill published in the GitHub repository hajibabaie/combinatorial-optimization-skills (7 stars, last pushed 3mo ago), licensed MIT. It adds 140 tokens to every session and 9,174 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

phx-deps-audit

Audit Hex deps for supply-chain security risk — bidi chars, compile-time exec, maintainer changes, typosquats, CVEs. Use after mix deps.update, when checking if a package upgrade is safe, or reviewing mix.lock PR diffs.

oliver-kriska/claude-elixir-phoenix · 58 tokens

release

CONTRIBUTOR TOOL - Cut a plugin release: bump plugin.json version, finalize CHANGELOG, update README if needed, gate on make ci, commit, tag vX.Y.Z, and create the GitHub release. Use when shipping a new plugin version. NOT distributed.

oliver-kriska/claude-elixir-phoenix · 60 tokens

session-deep-dive

Deep qualitative analysis of high-signal sessions. Spawns subagents with v2 template, synthesizes patterns, compares against known findings. Use after /session-scan.

oliver-kriska/claude-elixir-phoenix · 40 tokens

catchup

Summarize and review what changed while you were away. Use after a weekend, vacation, or flight to check missed PRs, git commits, Linear tickets, and meetings — one prioritized brief, not a firehose.

oliver-kriska/claude-elixir-phoenix · 48 tokens

brainstorm

Brainstorm Elixir/Phoenix features — explore ideas, compare approaches, gather requirements. Use when vague idea, not sure how to approach, or want to discuss before plan.

oliver-kriska/claude-elixir-phoenix · 39 tokens

learn-from-fix

Capture Elixir/Ecto/LiveView lessons and Hex API rules. Use after corrections or when asked to document learning, record a lesson, prevent a fixed mistake, or remember package guidance with --library.

oliver-kriska/claude-elixir-phoenix · 46 tokens