review-pr command#

The my-own-accelerator package exposes:

review-pr to fetch information about a GitHub pull request and print a Markdown summary.
review-local to review one or multiple local files and print a Markdown summary.

Both commands support --copilot-review to append AI-generated feedback produced by GitHub Copilot.

Installation#

Install the package (preferably in a virtual environment) to make the command available:

pip install my-own-accelerator

After installation the review-pr and review-local commands are available on the path.

`review-local`#

See review-local command for the full documentation of the review-local command.

Synopsis#

    usage: python -m moa review-pr [-h] [--token TOKEN] [--api-url URL]
                                   [--user USERNAME] [--save] [--copilot-review]
                                   [--model MODEL] [--prompt PROMPT] [-v]
                                   owner repo pull_request
    
    Reviews a GitHub pull request and prints markdown.
    
    positional arguments:
      owner             GitHub repository owner
      repo              GitHub repository name
      pull_request      Pull request number
    
    options:
      -h, --help        show this help message and exit
      --token TOKEN     GitHub personal access token. Resolution order: flag →
                        GITHUB_TOKEN env var → project cached value (owner/repo in
                        /home/runner/.config/moa/review_pr.json) → cached value
                        (/home/runner/.config/moa/review_pr.json). See the
                        docs/cmds/github_token page for how to obtain one.
      --api-url URL     Base URL of the GitHub API. Resolution order: flag >
                        GITHUB_API_URL env var > cached value
                        (/home/runner/.config/moa/review_pr.json) >
                        https://api.github.com.
      --user USERNAME   GitHub username of the authenticated user. Resolution
                        order: flag > GITHUB_USER env var > cached value
                        (/home/runner/.config/moa/review_pr.json). When set and
                        owner is omitted, the username is used as the repository
                        owner.
      --save            Save the resolved --token, --api-url, and --user to
                        /home/runner/.config/moa/review_pr.json so they are used
                        automatically in future invocations. The file is created
                        with owner-only read permissions (0600).
      --copilot-review  Use GitHub Copilot (via GitHub Models API) to generate an
                        AI review of the PR. Requires a GitHub token with models
                        access.
      --model MODEL     AI model used for --copilot-review. When omitted, Copilot
                        chooses the model automatically and falls back to
                        openai/gpt-4.1 if needed. Any model available on the
                        GitHub Models API is accepted.
      --prompt PROMPT   Add a follow-up prompt to the Copilot review session. The
                        prompt is sent as a continuation of the same conversation
                        so the model has full context from the initial review. Can
                        be used multiple times to ask several follow-up questions.
                        Only meaningful when --copilot-review is also set.
      -v, --verbose     Print progress information to stderr.

Positional Arguments#

owner: GitHub user or organisation that owns the repository (e.g. xadupre).
repo: Name of the GitHub repository (e.g. my-own-accelerator).
pull_request: Integer number of the pull request to review (e.g. 42).

Optional Arguments#

--token TOKEN

GitHub personal access token used to authenticate API requests. Resolution order: explicit flag value → GITHUB_TOKEN environment variable (automatically set in GitHub Actions workflows) → project-specific value cached with --save (owner/repo key) → classic cached value with --save → unauthenticated. For private repositories or to avoid rate limiting, a token with at least repo:read scope is required.

See Obtaining a GitHub Personal Access Token for instructions on how to obtain a token.

Example:

review-pr --token ghp_xxxxxxxxxxxx owner repo 42

--api-url URL

Base URL of the GitHub API. Resolution order: explicit flag value → GITHUB_API_URL environment variable (automatically set in GitHub Actions workflows) → value cached with --save → https://api.github.com. Override this when working against a GitHub Enterprise instance:

review-pr --api-url https://github.example.com/api/v3 owner repo 42

--user USERNAME

Your GitHub username. Resolution order: explicit flag value → GITHUB_USER environment variable → value cached with --save. When set and owner is omitted from the command line, the username is used as the repository owner automatically.

Example – save your username once, then omit owner in future calls:

review-pr --user myname --token "$GITHUB_TOKEN" --save myname my-repo 1

review-pr my-repo 2

--save

Persist the resolved --token, --api-url, and --user values to ~/.config/moa/review_pr.json so they are used automatically in future invocations (without needing to pass the flags again). The file is created with owner-read-only permissions (0600).

Warning

This stores your token in plain text. Only use --save on a personal workstation that you control. In CI environments, rely on GITHUB_TOKEN instead.

Example – authenticate once and save:

review-pr --token "$GITHUB_TOKEN" --save xadupre my-own-accelerator 1

Subsequent invocations on the same machine no longer need --token:

review-pr xadupre my-own-accelerator 2

--copilot-review

After fetching the pull request data, send the summary to the GitHub Models API (powered by GitHub Copilot) to obtain an AI-generated code review. The review is appended to the Markdown output as a ## Copilot Review section. A valid GitHub token is required (see --token):

review-pr --copilot-review xadupre my-own-accelerator 1

--prompt PROMPT

Add a follow-up prompt to the Copilot review session. The prompt is sent as a continuation of the same conversation so the model has full context from the initial review. The flag can be repeated to ask several follow-up questions in sequence. Only meaningful when --copilot-review is also set:

review-pr --copilot-review \
    --prompt "What are the security implications?" \
    --prompt "Are there any performance concerns?" \
    xadupre my-own-accelerator 1

Each follow-up response is appended to the ## Copilot Review section, preceded by a separator and the prompt text.

--model MODEL

AI model to use when --copilot-review is set. Accepts any model identifier available on the GitHub Models API. When omitted, Copilot chooses the model automatically and falls back to openai/gpt-4.1 if no default model is available:

review-pr --copilot-review --model openai/gpt-4o xadupre my-own-accelerator 1

-v, --verbose

Print progress information to standard error while the command is running. Verbose output also shows which token source is used (--token, GITHUB_TOKEN, project-specific cache, classic cache, or unauthenticated) and the token type (explicit, fine-grained, classic, or none). For tokens supplied via --token or GITHUB_TOKEN, the type is inferred from the token prefix (ghp_* → classic, github_pat_* → fine-grained).

-h, --help

Print a short help message and exit.

Examples#

Review a public pull request without authentication:

review-pr xadupre my-own-accelerator 1

Review a pull request and save the output to a file:

review-pr xadupre my-own-accelerator 1 > review.md

Authenticate with a personal access token (recommended for private repositories or to increase the API rate limit):

review-pr --token "$GITHUB_TOKEN" xadupre my-own-accelerator 1

Include an AI-powered Copilot review:

review-pr --copilot-review --token "$GITHUB_TOKEN" xadupre my-own-accelerator 1

Review a pull request on a GitHub Enterprise server:

review-pr \
    --api-url https://github.example.com/api/v3 \
    --token "$GHE_TOKEN" \
    myorg myrepo 7

Output Format#

The command prints a Markdown document to standard output. The document contains three sections (plus an optional Copilot Review section when --copilot-review is used):

Summary – title, state, author, URL, number of changed files, and total additions/deletions.
Description – the body text of the pull request.
Changed Files – list of every file touched by the pull request together with the number of added and deleted lines.
Copilot Review (optional) – AI-generated review produced by the GitHub Models API when --copilot-review is passed.

Example output (values are illustrative):

# Pull Request Review

## Summary
- **Title:** Fix typo in README
- **State:** open
- **Author:** octocat
- **URL:** https://github.com/owner/repo/pull/42
- **Files changed:** 2
- **Additions/Deletions:** +10 / -3

## Description
Fixes a typo in the README introduction paragraph.

## Changed Files
- `README.md` (+10/-3)
- `docs/index.rst` (+1/-0)

## Copilot Review

The change looks straightforward. Consider adding a test to ensure
the README renders correctly in CI.

Exit Codes#

0: The pull request was retrieved and the Markdown summary was printed successfully.
1: An error occurred (network failure, invalid PR number, or authentication problem). A human-readable message is printed to standard error.

Python API#

The command can also be invoked programmatically:

from moa.commands.review_pr import review_pull_request

# Basic summary
markdown = review_pull_request(
    owner="xadupre",
    repo="my-own-accelerator",
    pull_request=1,
    token="ghp_xxxxxxxxxxxx",   # optional
    api_url="https://api.github.com",  # optional
)
print(markdown)

# With Copilot review
markdown = review_pull_request(
    owner="xadupre",
    repo="my-own-accelerator",
    pull_request=1,
    token="ghp_xxxxxxxxxxxx",
    copilot_review=True,
    model="openai/gpt-4o",  # optional override
)
print(markdown)

# With Copilot review and follow-up prompts (session)
markdown = review_pull_request(
    owner="xadupre",
    repo="my-own-accelerator",
    pull_request=1,
    token="ghp_xxxxxxxxxxxx",
    copilot_review=True,
    extra_prompts=[
        "What are the security implications?",
        "Are there any performance concerns?",
    ],
)
print(markdown)

See the moa.commands.review_pr API reference for full details.

review-pr command#

Installation#

review-local#

Synopsis#

Positional Arguments#

Optional Arguments#

Examples#

Output Format#

Exit Codes#

Python API#

`review-local`#