Skip to main content

PIP: the performance review where the AI reviews you

Every AI tool reviews your code. PIP reviews the programmer.

Every developer keeps a work diary without knowing it: their git history. Every change is in there, with a timestamp and a little note that was supposed to explain it. PIP reads that record, locally, on your own machine, and issues a formal manager-style performance review. With evidence.

pip install on-a-pip

Then, inside any git repository:

on-a-pip

Or review somebody else's, straight from the internet:

on-a-pip psf/requests                      # GitHub shorthand
on-a-pip https://github.com/psf/requests    # any git URL
on-a-pip psf/requests --leaderboard        # rank the whole team

PIP clones the repo into a temp folder, reviews it, and deletes the clone when it is done (--keep keeps it, --shallow 500 fetches only recent history on huge repos). Reviewing a famous project is an education: the creator of Requests, 3,148 commits deep, grades out at a D with a Night Owl award. The record is the record.

  PERFORMANCE IMPROVEMENT PLAN  -  Dev Example
  pip-sample  |  8 commits reviewed  |  2025-10-14 to 2026-11-08
  --------------------------------------------------------------
  Communication    F    32.5  ######
  Reliability      C    64.4  ############
  Follow-through   B    84.3  ################
  Judgment         B    83.8  #################
  Consistency      D    43.0  ########
  --------------------------------------------------------------
  OVERALL          C    61.6
  Meets some expectations. There is a conversation to be had.

(That is a real review of a deliberately terrible sample history. Yours will be different. Possibly worse.)

What it grades

  • Communication. How many of your commit messages actually say anything. "stuff", "wip" and "asdf" are entered into the record, quoted back at you.
  • Reliability. Commits between 10pm and 5am. Commits shipped Friday after 4pm. Reverts. Chains of commits that start with "fix".
  • Follow-through. The TODOs still sitting in your code, and how old the oldest one is. A TODO that has been waiting 400 days has seniority.
  • Judgment. The spots labelled "temporary", "hack" or "workaround", and how long the oldest one has been load-bearing.
  • Consistency. Active days versus silent stretches, longest streak, longest disappearance.

Then it does what any good PIP does: it writes you a 30-day improvement plan, generated from your three weakest categories, and signs off with the only appeal process that has ever worked: improve the record.

Awards

Grades are the paperwork. Awards are the part people post. PIP reads your habits and hands out titles, each one earned by the record:

  • Night Owl - a quarter or more of your commits land between 10pm and 5am
  • Friday Gambler - repeated Friday-after-4pm shipping
  • Demolitionist - you delete more than you add, at scale
  • The Ghost - a 60+ day disappearance, then a casual return
  • Fossil Keeper - a TODO old enough to qualify for the pension plan
  • Surgeon, Streak Keeper, Open Book, Strong Silent Type, Chaos Agent, Weekend Warrior, Colourful Commentator
  • Marathon Day (15+ commits in one day), The Purge (a single commit deleting 1,000+ lines), Polyglot (real churn in 5+ languages), Old Faithful (3+ years on one codebase), Phoenix (a long disappearance followed by a genuine comeback)

It also names your nemesis file (the one with the most churn in your history) and your largest single commit, because management did the reading.

Choose your manager

on-a-pip --persona brutal       # the startup CTO
on-a-pip --persona supportive   # the kind lead
on-a-pip --persona passive      # the passive-aggressive middle manager
on-a-pip --persona corporate    # the default. HR is present.

The evidence never changes. The tone absolutely does.

More ways to be reviewed

on-a-pip ~/some/repo --md review.md --html card.html --svg card.svg --badge badge.svg
on-a-pip --author "your name"     # review one author (default: the most prolific)
on-a-pip --days 90                # just the last 90 days of history
on-a-pip --since 2026-01-01 --until 2026-03-31   # or a calendar window
on-a-pip --reference letter.md    # also write a letter of reference, based on the grade
on-a-pip --vs "Alice" "Bob"       # head to head: two authors, one repo, one winner
on-a-pip --scan ~/projects        # review the top author of every repo in a folder
on-a-pip --quarter                # the quarterly ritual: last 90 days, compared to your last review
on-a-pip --year 2026              # PIP Wrapped: your year in commits, as a shareable edition
on-a-pip --outdir ./my-review     # write every artifact at once: review, cards, badge, reference, JSON
on-a-pip --fail-below B           # exit code 3 if the grade is worse than B (a CI gate, serious or otherwise)
on-a-pip --leaderboard            # review every author in the repo and rank them
on-a-pip psf/requests               # or skip the local part entirely: any public repo by URL
on-a-pip --json                   # machine-readable review
  • The terminal summary above, now with awards, and a progress line comparing you to your last review (PIP remembers, locally).
  • A full formal review in Markdown: per-category evidence, management notes, additional findings, awards, the plan, the sign-off.
  • Shareable report cards in HTML and SVG (the SVG embeds straight into a README), plus a small grade badge for your profile.
  • A leaderboard mode for shared repos. Run it on your team. We accept no responsibility for the standup that follows.
  • A letter of reference (--reference), written by management in the tone your grade deserves. A grades get a glowing letter. F grades get a short one.
  • Career stats in every review: busiest day ever, monthly activity sparkline, languages ranked by churn. Identities are grouped by email, so one person under two names is one employee, and two people sharing a name stay two rows.
  • Head-to-head mode (--vs): two authors compared category by category, with a declared winner and consequences for the loser.
  • Folder scan (--scan): point PIP at a directory of projects and get one ranked board of every repo's top author.
  • Mailmap support: if the repo has a .mailmap, PIP honors it, so identity merging follows the project's own records.
  • PIP Wrapped (--year): the year-in-commits edition, with its own dark Wrapped card: busiest month, top award, your catchphrase.
  • Hall of fame stats: longest commit message (the novel), shortest (the haiku), and your catchphrase, the word your subjects cannot stop using.
  • Report cards chart your commits per month; --outdir writes the full artifact set in one run; --fail-below turns a grade into a CI gate.

Famous projects, formally reviewed: the PIP Gallery has real report cards and leaderboards for Requests, Flask, Express, and normalize.css, all generated by PIP itself.

Sample outputs live in examples/: a full sample review, sample report cards, a badge, and a real report card generated from the Decksmith repository.

Wear your grade

on-a-pip --readme

Run inside your repo, this writes pip-card.svg and pip-badge.svg and adds a marked My PIP review section to your README (the one at the bottom of this README is PIP reviewing this repo). Re-run it any time to refresh the grade.

To keep it current automatically, use the GitHub Action in your workflow:

name: pip-review
on:
  push:
    branches: [main]
permissions:
  contents: write
jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0
      - uses: littlestjames82-sys/pip-pip@main

Every push gets a fresh review, committed by the bot. Your README becomes a living performance record, which is either motivating or a threat, depending on your habits.

Private by construction

PIP runs git log, git grep and git blame on your machine and nothing else. No network, no account, no API key, no telemetry, pure Python standard library. Your history is embarrassing enough without uploading it somewhere.

Requirements

  • Python 3.9+
  • git, on your PATH

Honest notes

  • The package is called on-a-pip because PyPI rejected pip-pip as too similar to pip itself, which is fair: you do not get to out-pip pip. The tool was going to be pip install pip-pip. Now you install the news that you are on one.
  • Grades are deterministic: same history, same review. The formulas are plain arithmetic in one file, pip_pip.py, if you want to argue with them.
  • PIP reviews the most prolific author in the repo by default. Use --author in shared repos.

License

MIT. Built by Ghost Developer Studio.

My PIP review

My PIP grade

PIP report card

Grade: B (81/100) · 9 commits reviewed · Top award: Polyglot · Reviewed 2026-11-18

This section is maintained by PIP (on-a-pip --readme). The record speaks for itself.

Metadata

Release files for on-a-pip 0.7.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for on-a-pip 0.7.0
File Size Uploaded Provenance Actions
on_a_pip-0.7.0.tar.gz 34.5 kB With provenance Details for on_a_pip-0.7.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for on-a-pip 0.7.0
File Interpreter ABI Platform Actions
on_a_pip-0.7.0-py3-none-any.whl Python 3 none any Details for on_a_pip-0.7.0-py3-none-any.whl

Total release size: 63.2 kB

Release files / on_a_pip-0.7.0.tar.gz

Download URL on_a_pip-0.7.0.tar.gz
Size 34.5 kB
Tags Source
SHA-256 checksum
How to use checksums
6574d690e5afba2834b32bcaad12199131e5297a952b7626bc88d77dc225b60c
BLAKE2b-256 checksum
How to use checksums
cf9e3f7ff1b0a245ef9f9e50bf7bbb74cc2bf163ff79fcb9c396042427964017
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.

Transparency log

Release files / on_a_pip-0.7.0-py3-none-any.whl

Download URL on_a_pip-0.7.0-py3-none-any.whl
Size 28.7 kB
Tags Python 3
SHA-256 checksum
How to use checksums
23b4d5d9af96480939dc4c1c611922d800201680e7b2408c5c56f6508767d121
BLAKE2b-256 checksum
How to use checksums
5c4e342584f1312d27e22e6ea0feb37f7a66ab6f5ae5e5fe8edac03f72063e41
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.

Transparency log

Security for on-a-pip 0.7.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for on_a_pip-0.7.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page