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

Differential Audit

by pedrohcgs in pedrohcgs/claude-code-my-workflow

Compare two implementations of the same thing — a port (R↔Python↔Stata), a reimplementation, a replication package, a refactor, or a new version against the old — so that agreement means something.

MITAuto-check: notesResearch & Science

Install Differential Audit

skills CLI
$ npx skills add pedrohcgs/claude-code-my-workflow --skill differential-audit -a claude-code

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

GitHub CLI
$ gh skill install pedrohcgs/claude-code-my-workflow differential-audit --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/pedrohcgs/claude-code-my-workflow.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/differential-audit .claude/skills/differential-audit && 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
differential-audit
GitHub stars
1.7k
Token cost
~1.6k tokens
SKILL.md length
760 words
Files
1
Skills in repo
59
Repo updated
First seen
Licence
MIT

At a glance

Compare two implementations of the same thing — a port (R↔Python↔Stata), a reimplementation, a replication package, a refactor, or a new version against the old — so that agreement means something.

  • Works in 10 steps: State the claim and the reference → Freeze the inputs before inspecting… → Declare tolerance classes, and make them… → …
  • Cross-language parity
  • SKILL.md covers 1. State the claim and the…, 2. Freeze the inputs before…, 3. Declare tolerance classes,… and 4. Crosswalk and output…, plus 8 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Differential Audit is an agent skill from pedrohcgs/claude-code-my-workflow. Compare two implementations of the same thing — a port (R↔Python↔Stata), a reimplementation, a replication package, a refactor, or a new version against the old — so that agreement means something. Freeze inputs first, inventory every expected output, test the comparator itself, compare every channel (not just the headline number), and give each divergence a stable ID and a smallest witness. Use for cross-language parity, replication, upgrade/regression gates, or whenever "the numbers match" is about to license a…

Its SKILL.md is about 1.6k 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 Research & Science, covering Econometrics and empirical research. It works with Python. The repository describes itself as: A ready-to-fork Claude Code template for academics using LaTeX/Beamer + R. Multi-agent review, quality gates, adversarial QA, and replication protocols. The licence is MIT.

When your agent uses it

  • Cross-language parity
  • Upgrade/regression gates
  • Whenever the numbers match is about to license a claim

Example prompts

  • “the numbers match”
  • “/differential-audit”

Requirements

  • Pre-approved tools (allowed-tools): Read, Grep, Glob, Bash, Write

Workflow steps

10 steps, taken from the step headings in SKILL.md.

  1. State the claim and the reference
  2. Freeze the inputs before inspecting anything
  3. Declare tolerance classes, and make them binding
  4. Crosswalk and output inventory
  5. Build cases that isolate mechanisms
  6. Test the comparator itself
  7. Compare every declared channel
  8. Give every divergence an ID and a smallest witness
  9. Adversarial expansion by someone else
  10. Close with separate reviews and an honest scope statement

What it can do on your machine

Read from SKILL.md and the folder at commit ae72617. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Read
    • Grep
    • Glob
    • Bash
    • Write

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    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

Differential Audit loads about 1.6k tokens when it runs. Until then it costs about 136 tokens; SKILL.md has 760 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~136
When it runs · the whole SKILL.md, loaded when a task matches
~1.6k

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: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Read, Grep, Glob, Bash, Write

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 pedrohcgs/claude-code-my-workflow at commit ae72617, republished under its MIT licence (© pedrohcgs). 760 words, ~1,554 tokens.

Download SKILL.mdSave it as .claude/skills/differential-audit/SKILL.md (or your agent's skills folder).
name
differential-audit
description
Compare two implementations of the same thing — a port (R↔Python↔Stata), a reimplementation, a replication package, a refactor, or a new version against the old — so that agreement means something. Freeze inputs first, inventory every expected output, test the comparator itself, compare every channel (not just the headline number), and give each divergence a stable ID and a smallest witness. Use for cross-language parity, replication, upgrade/regression gates, or whenever "the numbers match" is about to license a claim.
allowed-tools
Read, Grep, Glob, Bash, Write
metadata.protocol
implementation-fidelity

Make agreement mean something

Two implementations agreeing proves they satisfy a prespecified contract. It does not prove either is correct, and it never validates the method or its assumptions. Both can be wrong in the same way — especially when one was written by reading the other. Design the comparison so that agreement is informative and disagreement is legible.

Rule: freeze before you compare; test the comparator before you trust it.

1. State the claim and the reference

Write down: what is being compared, which side is the reference, and what agreement would and would not establish. "Matches the R package" is a conformance claim, not a correctness claim. Say so explicitly, so nobody later reads parity as validation.

2. Freeze the inputs before inspecting anything

Record and fix: data versions or hashes, code and package versions, random seeds or realized sample splits, options and defaults, the outputs to be compared, and the acceptance thresholds. Freezing after a first look invites tolerance drift toward whatever the run produced.

Do not compare defaults across systems as if only the language changed. Map the choices explicitly — a "default" is a substantive modeling decision that usually differs between implementations.

3. Declare tolerance classes, and make them binding

Do not carry a single fuzzy epsilon. Classify each output:

  • EXACT — names, ordering, sample masks, counts, statuses, return/error codes, warning classes, option defaults. Byte-equal after documented normalization.
  • Scalar numeric — deterministic estimates, standard errors, p-values, critical values. State absolute and relative tolerances and the justification.
  • Matrix/vector — covariance matrices, influence summaries, weight vectors, plot data.
  • Stochastic — must meet a prespecified error-rate criterion with uncertainty reported.

A looser tolerance may be used only through a recorded approved divergence with a reason. Silent widening is the most common way a parity gate stops testing anything.

4. Crosswalk and output inventory

Write the mapping between the two implementations, plus an inventory of every expected output with expected row/cell counts. Every declared object must have a live comparison or an explicit out-of-scope reason. Without an inventory, both sides can silently omit the same result and the comparison reports success.

5. Build cases that isolate mechanisms

Not just the happy path:

  • analytically solvable or known-truth cases;
  • relevant data problems (missing, unbalanced, near-collinear, extreme-but-valid weights, shuffled row order, ties);
  • compound cases combining several problems;
  • published examples;
  • randomized valid designs across the supported surface, not a handful of fixtures.

Fixed fixtures are necessary but not sufficient — they test what the author already thought of.

6. Test the comparator itself

Before trusting a green result, feed the comparison a wrong value, a missing result, a misaligned row, and an empty result. It must fail, not skip. A comparator that silently passes over what it cannot reconcile turns every subsequent green into noise. (See vaccinate.)

Show full SKILL.md (303 more words)Show less

7. Compare every declared channel

Not only the headline coefficient: estimates, uncertainty measures, sample counts, labels and ordering, diagnostics, warnings, and failure statuses. Divergent warnings and differing error behavior are real defects — they change what a user does next.

8. Give every divergence an ID and a smallest witness

For each difference: a stable identifier, the smallest reproducible case, and a classification — defect / intentional difference / limitation of the reference / unresolved. Unresolved stays red; it is not averaged away or waived. A fix must turn its witness green and survive a rerun of the full audit, so a local patch does not break something else.

9. Adversarial expansion by someone else

Fixtures written by the implementer test the implementer's mental model. Have an independent reviewer add designs not shared in advance, and preserve any that reveal bugs or materially increase coverage as permanent fixtures. This is the cheapest defense against a suite that passes because it was written to pass.

10. Close with separate reviews and an honest scope statement

End with distinct scientific and implementation sign-off, recording: checks run, open findings, accepted differences, explicit non-claims, and who approved release. State plainly that the audit establishes conformance to the frozen contract — not the truth of the method.

Minimum checklist

  1. Name the reference; state what agreement would and would not establish.
  2. Freeze versions, hashes, seeds, options, outputs, tolerances.
  3. Declare binding tolerance classes; record any approved divergence.
  4. Crosswalk + expected-output inventory with counts.
  5. Known-truth, dirty, compound, and randomized cases.
  6. Seed comparator faults — it must fail, not skip.
  7. Compare all channels, including warnings and failures.
  8. Stable ID + smallest witness + classification per divergence; unresolved stays red.
  9. Independent reviewer adds unseen designs.
  10. Separate sign-offs; state the non-claims.

Cross-references

© pedrohcgs, 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 .claude/skills/differential-audit of pedrohcgs/claude-code-my-workflow.

Open the folder on GitHubat commit ae72617

Compare with similar skills

Differential Audit 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.

Differential Audit compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Differential Audit this skillpedrohcgs/claude-code-my-workflow1.7k—~1.6kAutomated safety check: NotesMIT
Stata C Pluginsdylantmoore/stata-skill2911 repos~5.8kAutomated safety check: PassCustom licence
Fin Data Acquisitioncsmar432/finai-research109—~2kAutomated safety check: PassMIT
Empirical Research MethodsCitrus-bit/Anaxa120—~1.9kAutomated safety check: PassCC-BY-SA-4.0
Causal Inference Mixtapebrycewang-stanford/Auto-Empirical-Research-Skills4.5k—~1.5kAutomated safety check: PassCustom licence
Aer Statspaibrycewang-stanford/Auto-Empirical-Research-Skills4.5k—~3kAutomated safety check: PassCustom licence

Similar skills

  • Stata C Plugins

    dylantmoore/stata-skill

    Develop high-performance C/C++ plugins for Stata using the stplugin.h SDK.

    291 GitHub starsUsed in 1 repo~5.8k tokens
    Research & ScienceAuto-check passed

  • Fin Data Acquisition

    csmar432/finai-research

    根据REFINEDDESIGN.md中的变量定义，自动获取所需数据并生成可执行的回归分析脚本（Python/Stata）。

    109 GitHub stars~2k tokensUpdated 3 days ago
    Research & ScienceAuto-check passed

  • Empirical Research Methods

    Citrus-bit/Anaxa

    A skill your agent uses for empirical social-science research, applied economics, public policy, education, finance, management, sociology, psychology, epidemiology, or public-health data studies.

    120 GitHub stars~1.9k tokensUpdated 1 mo ago
    Research & ScienceAuto-check passed

  • Causal Inference Mixtape

    brycewang-stanford/Auto-Empirical-Research-Skills

    This skill should be used when the user asks to "implement a DiD regression", "write a causal inference pipeline", "set up an event study", "implement instrumental variables", "run a regression…

    4.5k GitHub stars~1.5k tokensUpdated 4 days ago
    Research & ScienceAuto-check passed

  • Aer Statspai

    brycewang-stanford/Auto-Empirical-Research-Skills

    A skill your agent uses when aer-identification has fixed the design, after methodology choice and before aer-robustness or aer-tables-figures, to run an AER-track analysis with StatsPAI — the…

    4.5k GitHub stars~3k tokensUpdated 4 days ago
    Research & ScienceAuto-check passed

  • Audit Replication

    brycewang-stanford/Auto-Empirical-Research-Skills

    Validate the replication package for the sewage-house-prices project.

    4.5k GitHub stars~984 tokensUpdated 4 days ago
    Research & ScienceAuto-check: notes

More from pedrohcgs/claude-code-my-workflow

All 59 skills in this repo

  • Devils Advocate

    pedrohcgs/claude-code-my-workflow

    Adversarial 5-7 question challenge to a deck's pedagogical choices — ordering, prerequisites, cognitive load, motivation.

    1.7k GitHub starsUsed in 2 repos~641 tokens
    Auto-check passed

  • Vaccinate

    pedrohcgs/claude-code-my-workflow

    Qualify a check before it is allowed to clear anything — prove it can detect the failure it is meant to catch.

    1.7k GitHub stars~2.1k tokensUpdated 12 days ago
    Auto-check: notes

  • Compile Latex

    pedrohcgs/claude-code-my-workflow

    Compile a Beamer LaTeX slide deck with XeLaTeX (3 passes + bibtex).

    1.7k GitHub starsUsed in 1 repo~492 tokens
    Auto-check: notes

  • Context Status

    pedrohcgs/claude-code-my-workflow

    Show current context status and session health. An agent skill from pedrohcgs/claude-code-my-workflow.

    1.7k GitHub starsUsed in 1 repo~613 tokens
    Auto-check: notes

  • Capture Environment

    pedrohcgs/claude-code-my-workflow

    Snapshot the computational environment for a replication package — detects the analysis stack (R / Stata / Python) and emits the right lockfiles (renv.lock + sessionInfo.txt, requirements.txt /…

    1.7k GitHub stars~2.8k tokensUpdated 12 days ago
    Auto-check: notes

  • Checkpoint

    pedrohcgs/claude-code-my-workflow

    Save a structured state snapshot before stopping or handing off.

    1.7k GitHub stars~2.8k tokensUpdated 12 days ago
    Auto-check: notes

Works with

Categories

Questions about Differential Audit

What does Differential Audit do?

Compare two implementations of the same thing — a port (R↔Python↔Stata), a reimplementation, a replication package, a refactor, or a new version against the old — so that agreement means something. Differential Audit is an agent skill from pedrohcgs/claude-code-my-workflow. Compare two implementations of the same thing — a port (R↔Python↔Stata), a reimplementation, a replication package, a refactor, or a new version against the old — so that agreement means something.

When should I use Differential Audit?

Differential Audit fits situations like: cross-language parity; upgrade/regression gates; whenever the numbers match is about to license a claim.

How do I install Differential Audit in Claude Code?

Run `npx skills add pedrohcgs/claude-code-my-workflow --skill differential-audit -a claude-code`. Or copy the skill folder (.claude/skills/differential-audit in pedrohcgs/claude-code-my-workflow) into .claude/skills/differential-audit in your project. Claude Code loads it when a task matches its description.

How do I install Differential Audit in Codex?

Run `npx skills add pedrohcgs/claude-code-my-workflow --skill differential-audit -a codex`. Or copy the skill folder (.claude/skills/differential-audit in pedrohcgs/claude-code-my-workflow) into .agents/skills/differential-audit in your project. Codex loads it when a task matches its description.

Can I use Differential Audit 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 pedrohcgs/claude-code-my-workflow --skill differential-audit -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/differential-audit, .gemini/skills/differential-audit, .github/skills/differential-audit and .opencode/skills/differential-audit in your project.

What does Differential Audit need to run?

SKILL.md names no scripts, command-line tools or credentials: Differential Audit is instructions for the agent only. Its frontmatter pre-approves these tools: Read, Grep, Glob, Bash, Write.

Does Differential Audit 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 Differential Audit safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. Review the folder before installing.

What licence does Differential Audit use?

Differential Audit 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 Differential Audit use?

About 1.6k tokens (SKILL.md is roughly 6.2k 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 Differential Audit?

Skills that share tags, products or a category with Differential Audit: Stata C Plugins (dylantmoore/stata-skill, 291 stars), Fin Data Acquisition (csmar432/finai-research, 109 stars), Empirical Research Methods (Citrus-bit/Anaxa, 120 stars) and Causal Inference Mixtape (brycewang-stanford/Auto-Empirical-Research-Skills, 4.5k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Differential Audit?

pedrohcgs (a GitHub user) maintains it in pedrohcgs/claude-code-my-workflow, which has 1,653 GitHub stars. The repository holds 59 skills in this directory. The repository was last updated on September 27, 2026.

Source: pedrohcgs/claude-code-my-workflow on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.