coldteadotai/pr-lens

_{vuejs/core#2532 · 11 files · +1,081/−670 · 3 lanes. <script setup> and the first version of ref sugar.}

_{rust-lang/rust#31954 · 26 files · +369/−16 · 4 lanes. The ? operator, a short way to write try! that you can chain.}

_{tokio-rs/tokio#1657 · 100 files · +7,408/−6,795 · 3 lanes. The work-stealing pool, rebuilt so the scheduler does less extra work.}

_{neovim/neovim#11336 · 15 files · +5,556/−1 · 3 lanes. The LSP client moves into Neovim itself.}

_{django/django#11209 · 38 files · +931/−42 · 5 lanes. An ASGI handler, and a request context that is safe to use with coroutines.}

_{webpack/webpack#10440 · 13 files · +567/−5 · 5 lanes. ContainerPlugin, which brought module federation with it.}

_{vllm-project/vllm#1348 · 6 files · +764/−139 · 3 lanes. PagedAttention V2 and its sequence-level parallelism.}

A lot happens before a pull request is opened. Your agent can use the same renderer to help you understand any codebase or check its own changes before there is a pull request. It also helps you keep a clear picture of your system as it grows.

npx skills add coldteadotai/pr-lens

Then ask your coding agent something like:

Diagram the change you just made with PR Lens and attach it to the pull request.

The agent reads the diff and writes the diagram file. It runs npx @coldtea/pr-lens-cli validate until the file passes every check. Then it draws the diagram and adds it to the pull request description with gh pr create --attach, so the diagram is there as soon as the pull request opens. If a diagram gets a name wrong, the skill teaches the agent to fix .github/pr-lens.yml instead of editing the files it made. See packages/agent-skill for more.

Want the agent to do the whole setup? Paste this:

Set up PR Lens for me. It draws code changes as moving diagrams of the system and how data flows through it.

1. Install the agent skill: `npx skills add coldteadotai/pr-lens`.

2. Help me install the GitHub App at https://github.com/apps/coldtea-pr-lens on every repository where I review pull requests. It posts one comment per pull request and updates that comment on every push. It does not need a model key from me.

3. If I'd rather run it from CI with my own model key, offer the Action instead: `.github/workflows/pr-lens.yml` using `coldteadotai/pr-lens/packages/action@v0`, with the key saved as a repository secret. It works with any service that speaks `/chat/completions`, such as OpenAI or Gemini.

4. Then test it: diagram the latest change in this repository and show me the SVGs or the canvas.

This posts the same comment from your own CI, drawn with your own model key. Save the key as a repository secret. The example below uses GEMINI_API_KEY because provider is Gemini by default. Then commit this file as .github/workflows/pr-lens.yml:

name: PR Lens

on:
  pull_request:

permissions:
  contents: write # to publish the rendered SVGs
  pull-requests: write # to post the comment

concurrency: # one run per pull request; a push supersedes the last
  group: pr-lens-${{ github.event.pull_request.number }}
  cancel-in-progress: true

jobs:
  lens:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0 # the diff is between two commits, so both must be here
      - uses: coldteadotai/pr-lens/packages/action@v0
        with:
          api-key: ${{ secrets.GEMINI_API_KEY }}

You can use any model. provider can be gemini (the default), openai, or openai-compatible with a base-url and model. So the same workflow works with OpenRouter, DeepSeek or your own server. The key reaches the CLI through the environment, never on a command line. The diff goes only to the provider you pick. This comment stays the same after it is posted, because an Action can't remember anything between runs. That is why the checkboxes only work in the App. See packages/action for providers, lenses, branding and the other inputs.

GitLab CI can post the same diagrams on a merge request. Add two masked variables: your model key, and a project access token with the api scope. The token is what posts the comment. Then add the component:

workflow:
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request_event"
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

include:
  - component: gitlab.com/coldteadotai/pr-lens/pr-lens@0.1.0

The workflow: rules block must be in your own .gitlab-ci.yml. GitLab only runs merge request pipelines when that file asks for them, and rules inside an included component don't count. You need the access token because CI_JOB_TOKEN can't post comments. On GitLab.com, project access tokens need the Premium or Ultimate plan. On the Free plan, a personal access token with the api scope works too, but the comment shows your name instead of a bot's.

The SVGs are uploaded to the project as attachments. That way everyone who reads the merge request can see them, even people who can't open files in the repository. On a private project, there is a downside. Anyone who has the attachment link can see the diagram, even if they are not a member. The link can't be guessed, but it is the only thing that protects the image. If a project turns on Require authentication to view media files, the images won't load. Readers can still open the diagrams from the canvas link in the comment. See packages/gitlab-component for more.

On Bitbucket Pipelines, add the pipe under pull-requests::

clone:
  depth: full # the diff needs history back to the merge base

pipelines:
  pull-requests:
    "**":
      - step:
          name: PR Lens
          script:
            - pipe: docker://ghcr.io/coldteadotai/pr-lens-pipe:0.1.0
              variables:
                GEMINI_API_KEY: $GEMINI_API_KEY
                PR_LENS_TOKEN: $PR_LENS_TOKEN

A pipe only sees the variables its step gives it, so the step passes in two secured variables by name. The first is your model key. The second is a repository access token with the pullrequest:write and repository:write scopes. The repository:write scope lets the pipe save the diagrams to the repository's Downloads. Every Bitbucket plan has repository access tokens. Tokens for a whole workspace need Premium.

Bitbucket shows comments as plain Markdown. So the comment has no sections to open and close, and no light and dark pair. It shows the headline, the numbers, one diagram per lens, and then the zoomed-in views in order. Bitbucket never runs a pipeline for a pull request opened from a fork, so the pipe can't draw those pull requests, no matter how you set it up. If your repository gets a lot of help from forks, use the hosted app. See packages/bitbucket-pipe for more.

The CLI does everything the other options do, one step at a time, on your computer. Only analyze talks to a model. It reads the key from the environment, never from a flag:

export GEMINI_API_KEY=…    # the default provider; OPENAI_API_KEY with --provider openai

# Diff in, graph document out — measured against the merge base, not the branch tip.
npx @coldtea/pr-lens-cli analyze --base origin/main

# The document as light and dark SVGs, plus the manifest a comment is built from.
# Each drawing lands in its own directory under .pr-lens/, named after its title.
npx @coldtea/pr-lens-cli render .pr-lens/graph.json

# The pull request comment as markdown, on stdout. Posting is your business.
npx @coldtea/pr-lens-cli comment --graph .pr-lens/<drawing>/drawn.graph.json --manifest .pr-lens/<drawing>/manifest.json \
  --asset-base-url https://raw.githubusercontent.com/owner/repo/pr-lens/42

# Any PR Lens document, checked against the contract — every problem, not just the first.
npx @coldtea/pr-lens-cli validate .pr-lens/graph.json .github/pr-lens.yml

# After the merge: the pull-request document as a stored map of the system, worth committing.
npx @coldtea/pr-lens-cli export .pr-lens/graph.json -o .github/pr-lens.map.json

Everything goes in .pr-lens/. The first time the CLI writes there, it adds that folder to your .gitignore. Think of it as scratch space, because the CLI can rebuild those files from the diff at any time. The only file worth committing is the map that export writes. Use --out if you want the files somewhere else.

comment --target gitlab or --target bitbucket writes a comment that GitLab or Bitbucket can show. analyze finds out which one you use from your git remote. To use Ollama, DeepSeek, OpenRouter, or anything else that speaks /chat/completions, pass --provider openai-compatible --base-url <url>. See packages/cli for every command, the fix file, and the error codes a script can check.

The diagrams don't need a pull request. Draw them on your computer and look at the change before anyone else does:

npx @coldtea/pr-lens-cli analyze --base origin/main
npx @coldtea/pr-lens-cli render .pr-lens/graph.json
open .pr-lens/*-dark-*.svg    # macOS; the SVGs are self-contained, any browser reads them

This also helps when you review an agent's work. While you read the diff, the agent that wrote it can draw it. With the skill installed, say "render this change with PR Lens and open the SVGs". You get the diagram next to the diff, minutes before the pull request shows the same picture.