code-review

A code-review checklist for dataprof, a Rust and Python library that examines data files and reports findings without changing them. It focuses on bugs caused by inconsistent behavior across different input paths.

In plain words
What is it for?
Use it when reviewing changes to dataprof engines, parsers, metrics, type inference, or quality scores. It helps check that the same data produces the same results everywhere.
Why use it?
It helps reviewers look beyond the changed lines, where a fix can appear correct but leave related code using different rules or numbers.

Skill for Claude CodeCodex

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.

agentmods
npx agentmods add skills/andreabozzo/dataprof/code-review
Any agent
npx skills add AndreaBozzo/dataprof --skill code-review
Clone the repo
git clone --depth 1 https://github.com/AndreaBozzo/dataprof

Made for: Claude Code, Codex.

Per session 84 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,912 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00084 $0.01912
Opus 5 $0.00042 $0.00956
Sonnet 5 $0.00017 $0.00382
Haiku 4.5 $0.00008 $0.00191

Measured 2d ago against content hash 746ee6d599a1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

code-review scanned grade A with 0 findings 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 2d 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.github/skills/code-review/SKILL.md · 153 lines

How it starts

The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Reviewing dataprof

dataprof is a local, deterministic data profiler: a Rust workspace with PyO3 Python bindings. It reports on data and never transforms, cleans, or moves it. The release surface is Rust library crates plus Python wheels; there is no CLI binary. Standing repository conventions live in AGENTS.md — this file is only about what to look for when reviewing a change.

The bug class that costs most here is rarely a local mistake inside the diff. It is a change that is correct where it was made and inconsistent with something that was not touched: a sibling input path, a second predicate for the same concept, or a number quoted in documentation. Read the diff, then go look at the places it should have changed too.

1. The same input must produce the same numbers everywhere

Profile numbers must be identical regardless of which engine or input path produced them. A diff that changes one path and leaves its siblings alone is the most common real defect, and it passes every test that only exercises the path it touched.

When a diff touches any of these, check the others for the same behaviour:

path file
CSV parser (shared by file and reader entry points) crates/dataprof-csv/src/lib.rs
JSON / JSONL parser crates/dataprof-json/src/lib.rs
Parquet crates/dataprof-parquet/src/arrow_profiler.rs, record_batch_analyzer.rs
incremental streaming engine crates/dataprof-engines/src/streaming/incremental.rs
async streaming reader crates/dataprof-engines/src/streaming/async_reader.rs
memory-mapped reader crates/dataprof-engines/src/streaming/memmap.rs
columnar (Arrow) engine crates/dataprof-engines/src/columnar/mod.rs
partial analysis (infer_schema, analyze_structure, quick_row_count) crates/dataprof-partial/src/lib.rs
database connectors crates/dataprof-db/
shared profile construction crates/dataprof-runtime/src/profile_builder.rs

Flag it when a fix lands in one of these and the same input would still be answered differently by another. Check both raw attributes and the serialized report, since a diff can bring one into line and leave the other out.

Read the full file on GitHub · 153 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. 2d ago First seen · 153 lines · 84 tokens per session scan A 746ee6d599a1

Subscribe to this mod's changes

code-review is a skill published in the GitHub repository AndreaBozzo/dataprof (20 stars, last pushed 2d ago), licensed MIT. It adds 84 tokens to every session and 1,912 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.