dx-pr-review-all

You list active Azure DevOps pull requests, present them for selection, then review each one using /dx-pr-review — which supports posting comments, proposing fix patches, and following up on author responses.

Flow

digraph pr_reviews {
    "Parse input" [shape=box];
    "Input type?" [shape=diamond];
    "Delegate to dx-pr-review" [shape=doublecircle];
    "Load MCP tools + resolve repo" [shape=box];
    "List active PRs" [shape=box];
    "PRs found?" [shape=diamond];
    "Present PR list" [shape=box];
    "User selects PRs" [shape=box];
    "Prepare repo access" [shape=box];
    "Pre-fetch branches" [shape=box];
    "Fetch existing threads" [shape=box];
    "Spawn review agents" [shape=box];
    "Process next PR result" [shape=box];
    "Present findings + ask action" [shape=box];
    "User choice?" [shape=diamond];
    "Generate fixes in worktree" [shape=box];
    "Present patches + ask approval" [shape=box];
    "Post comments to ADO" [shape=box];
    "Set vote" [shape=box];
    "Save session" [shape=box];
    "Cleanup per-PR" [shape=box];
    "More PRs?" [shape=diamond];
    "Print aggregate summary + cleanup" [shape=doublecircle];
    "No active PRs" [shape=doublecircle];

    "Parse input" -> "Input type?";
    "Input type?" -> "Delegate to dx-pr-review" [label="PR URL"];
    "Input type?" -> "Load MCP tools + resolve repo" [label="repo URL or empty"];
    "Load MCP tools + resolve repo" -> "List active PRs";
    "List active PRs" -> "PRs found?";
    "PRs found?" -> "Present PR list" [label="yes"];
    "PRs found?" -> "No active PRs" [label="no"];
    "Present PR list" -> "User selects PRs";
    "User selects PRs" -> "Prepare repo access";
    "Prepare repo access" -> "Pre-fetch branches";
    "Pre-fetch branches" -> "Fetch existing threads";
    "Fetch existing threads" -> "Spawn review agents";
    "Spawn review agents" -> "Process next PR result";
    "Process next PR result" -> "Present findings + ask action";
    "Present findings + ask action" -> "User choice?";
    "User choice?" -> "Generate fixes in worktree" [label="post with patches"];
    "User choice?" -> "Post comments to ADO" [label="post comments only"];
    "User choice?" -> "Present findings + ask action" [label="edit"];
    "User choice?" -> "Cleanup per-PR" [label="cancel"];
    "Generate fixes in worktree" -> "Present patches + ask approval";
    "Present patches + ask approval" -> "Post comments to ADO" [label="approved"];
    "Present patches + ask approval" -> "Post comments to ADO" [label="cancel patches"];
    "Post comments to ADO" -> "Set vote";
    "Set vote" -> "Save session";
    "Save session" -> "Cleanup per-PR";
    "Cleanup per-PR" -> "More PRs?";
    "More PRs?" -> "Process next PR result" [label="yes"];
    "More PRs?" -> "Print aggregate summary + cleanup" [label="no"];
}

Node Details

Parse input

Read shared/ado-config.md for how to look up ADO project from .ai/config.yaml.

• Organization: read from .ai/config.yaml scm.org — NEVER hardcode
• Project: read from .ai/config.yaml scm.project

Parse $ARGUMENTS to determine the mode:

Input	Mode	Example
(empty)	Current repo, 5 PRs	/dx-pr-review-all
<number>	Current repo, N PRs	/dx-pr-review-all 10
<repo URL>	That repo, 5 PRs	/dx-pr-review-all https://{org}/_git/My-Repo
<repo URL> <number>	That repo, N PRs	/dx-pr-review-all https://{org}/_git/My-Repo 10
<PR URL>	Single PR (delegate)	/dx-pr-review-all https://.../_git/.../pullrequest/12345

Input type?

• PR URL — contains /pullrequest/ → go to "Delegate to dx-pr-review"
• repo URL or empty — contains /_git/ but no /pullrequest/, or numeric only, or empty → go to "Load MCP tools + resolve repo"

Detect PR URL vs Repo URL:

• Contains /pullrequest/ → single PR mode
• Contains /_git/ but no /pullrequest/ → repo URL — extract project and repo name. URL-decode the project (e.g., My%20Project → My Project). The URL-extracted project takes precedence over the config default.
• Numeric only → count for current repo

Delegate to dx-pr-review

Skip the listing and invoke /dx-pr-review <PR URL> directly. The URL contains the correct project — pass it through as-is. STOP.

Load MCP tools + resolve repo

Load the tools:

ToolSearch("+ado repo")
ToolSearch("+ado pull request thread")

Detect current repo (when no URL is provided):

git remote get-url origin

Extract the repo name from the URL:

• vs-ssh.visualstudio.com:v3/{org}/{project}/{repo} → repo name is the last segment
• {org}.visualstudio.com/{project}/_git/{repo} → repo name after _git/

Extract repo from URL (when URL provided):

• https://dev.azure.com/{org}/{project}/_git/{repo} → project + repo
• https://{org}.visualstudio.com/{project}/_git/{repo} → project + repo

Resolve the repo name to an ID:

mcp__ado__repo_get_repo_by_name_or_id
  project: "<project from URL if provided, otherwise from config>"
  repositoryNameOrId: "<repo name>"

Important: If the user provided a URL, use the project extracted from that URL — NOT the config default. The same repo can exist in multiple ADO projects.

Save the id field and sshUrl — needed for all subsequent calls.

List active PRs

Detect the current user from git config user.email.

mcp__ado__repo_list_pull_requests_by_repo_or_project
  repositoryId: "<repo ID>"
  status: "Active"
  top: <count>

Then fetch full details for each PR (parallel calls):

mcp__ado__repo_get_pull_request_by_id
  repositoryId: "<repo ID>"
  pullRequestId: <PR ID>

Filter out own PRs: Exclude PRs where createdBy.uniqueName or createdBy.displayName matches the current user. You can't review your own PRs. Show filtered count: "Found N active PRs (excluded M own)."

PRs found?

• yes — at least one active PR (after filtering) → go to "Present PR list"
• no — no active PRs → go to "No active PRs"

No active PRs

Print: "No active PRs in ." STOP.

Present PR list

## Active PRs — <repo name> (<count> found)

| # | PR | Title | Author | Created | Reviewers |
|---|-----|-------|--------|---------|-----------|
| 1 | [#12345](url) | Fix login bug | John D. | 2d ago | 2/4 approved |
| 2 | [#12346](url) | Add feature X | Jane S. | 5h ago | 0/3 approved |
| ... | | | | | |

Review all, or pick specific PRs? (e.g., "all", "1 3 5", "skip 2")

For each PR show:

• PR number with link: {scm.org}/{scm.project}/_git/{repo}/pullrequest/{id}
• Title — truncated to 60 chars if needed
• Author — display name (last name, first name → first name only for brevity)
• Created — relative time (e.g., "2d ago", "5h ago")
• Reviewers — N/M approved (count votes of 10 = approved, M = total non-container reviewers)

User selects PRs

Wait for user to specify which PRs to review (e.g., "all", "1 3", "skip 2").

Prepare repo access

Determine the repo path for diffing:

1. Check if PR repo matches current directory:

   basename $(git remote get-url origin)
   

2. If same repo: use current directory as repoPath.
3. If different repo: check .ai/config.yaml repos: section for a local checkout path. Use that as repoPath.
4. If no local checkout: shallow-clone once:

   git clone --no-checkout --filter=blob:none <sshUrl> /tmp/dx-review-<repo>
   

   Use /tmp/dx-review-<repo> as repoPath.

Pre-fetch branches

Collect all unique source and target branches from the selected PRs (strip refs/heads/ prefix). Fetch them all in one command:

git -C <repoPath> fetch origin <branch1> <branch2> <branch3> ...

Fetch existing threads

For each selected PR, fetch existing review threads (make all calls in parallel):

mcp__ado__repo_list_pull_request_threads
  repositoryId: "<repo ID>"
  pullRequestId: <PR ID>

Summarize per PR: count of active threads and which files they cover.

Spawn review agents

For each selected PR, spawn a dx-pr-reviewer agent via the Task tool.

Spawn ALL agents in a single message (multiple Task tool calls) for parallel execution:

Task(
  subagent_type: "dx-pr-reviewer",
  description: "Review PR #<id>",
  prompt: "Review this pull request:

    repoName: <name>
    repoPath: <path>
    pullRequestId: <id>
    title: <title>
    description: <description>
    author: <author display name>
    sourceBranch: <branch without refs/heads/>
    targetBranch: <branch without refs/heads/>
    existingThreadsSummary: <N active threads on: file1, file2, ...>"
)

Process next PR result

Pick the next PR result from the batch of completed agents. Load .ai/me.md once (if it exists) to shape the voice of comments and patch proposals.

Present findings + ask action

Print separator:

---
## PR <N> of <M>: #<id> — <title>
**Repo:** <repo> | **Author:** <name> | **Files:** <count>

Display the agent's structured findings:

| # | Sev | File | Line(s) | Comment | Fixable? |
|---|-----|------|---------|---------|----------|
| 1 | MUST-FIX | `path/to/file.js` | L42-L45 | hm, this null check is missing — will NPE when X is empty | YES |
| 2 | QUESTION | `path/to/file.js` | L10 | not sure this handles the edge case where... | NO |

**Verdict**: Approved / Approved with suggestions / Changes requested
Reviewed N files — N comments.
**Fixable issues:** <N> of <M> can have patches generated

Fixable determination: An issue is fixable if the reviewer agent provided enough detail to write a specific code change. Questions and ambiguous issues are NOT fixable.

Ask the user:

• Post comments only — post review comments as-is (no patches)
• Post with fix patches — generate patches first, then post comments with inline diffs
• Edit — remove, modify, or add comments
• Cancel — discard without posting, move to next PR

Wait for explicit choice before proceeding.

User choice?

• post with patches → go to "Generate fixes in worktree"
• post comments only → go to "Post comments to ADO"
• edit → go back to "Present findings + ask action" (with edits applied)
• cancel → go to "Cleanup per-PR"

Generate fixes in worktree

Use the Task tool with isolation: "worktree" to create fixes without touching local state:

Task(
  subagent_type: "general-purpose",
  isolation: "worktree",
  description: "Generate PR #<id> fix patches",
  prompt: "Generate code fixes for the issues found in PR #<id>. Work on the PR author's source branch.

    repoPath: <current working directory>
    sourceBranch: <source branch without refs/heads/>
    targetBranch: <target branch without refs/heads/>

    ## Setup

    1. Checkout the PR's source branch:
       ```bash
       git fetch origin <sourceBranch>
       git checkout origin/<sourceBranch>
       ```
    2. This is now the PR author's code. Your job is to fix the issues below.

    ## Issues to Fix

    <for each selected fixable issue:>
    ### Issue #<N>
    File: <filePath>
    Line(s): <line range>
    Severity: <MUST-FIX>
    Problem: <description from the review>
    What needs to change: <specific fix description>

    ## Persona

    <If .ai/me.md was found, paste its full content here.
     If not found, omit this entire Persona section.>

    ## Instructions

    For each issue:
    1. **Read the file** — full file or +/-50 lines around the target area
    2. **Understand the context** — what the code does, what's wrong
    3. **Apply the minimal fix** — ONLY what's needed. No refactoring
    4. **Verify consistency** — check if the same pattern exists elsewhere
    5. **Follow project conventions** — read .claude/rules/ for the file type
    6. **Report what you changed**

    ## Constraints
    - Minimal changes only. This is a PROPOSAL — the author decides.
    - If a fix seems risky, flag it instead of applying.
    - One fix = one logical change.

    ## Output Format

    ### Fix #<N> — <filePath>
    **Issue:** <1-line problem description>
    **What changed:** <1-line fix description>
    **Lines modified:** L<start>-L<end>
    **Risk:** low | medium | high
    **Notes:** <concerns, or 'none'>
    ---

    If a fix could NOT be applied:
    ### Fix #<N> — <filePath>
    **Status:** SKIPPED
    **Reason:** <why>
    ---
  "
)

After the worktree agent finishes, generate a unified diff:

git -C <worktreePath> diff origin/<sourceBranch> -- . > /tmp/dx-propose-pr-<id>.patch
git -C <worktreePath> diff origin/<sourceBranch> --stat
git -C <worktreePath> diff origin/<sourceBranch>

If the patch is empty, report and fall back to posting comments only.

Present patches + ask approval

Only shown when patches were generated.

## Proposed Fixes — PR #<id>: <title>

| # | File | Issue | Fix | Risk |
|---|------|-------|-----|------|
| 1 | `hero.js` L42 | Missing null check | Added null guard | low |
| 2 | `Model.java` L18 | Missing @Optional | Added annotation | low |

**Patch size:** <N> files, +<additions> -<deletions>

Show the full git diff so the user can inspect every change.

Ask: Post all / Edit / Cancel (discard patches, post comments only).

Wait for explicit approval.

• approved → go to "Post comments to ADO" (with patches included)
• cancel patches → go to "Post comments to ADO" (comments only, no patches)

Post comments to ADO

Without patches (default): Post each comment as a thread:

mcp__ado__repo_create_pull_request_thread
  repositoryId: "<repo ID>"
  pullRequestId: <PR ID>
  content: "<approved comment text>"
  filePath: "/<path/to/file>"
  rightFileStartLine: <line>
  rightFileEndLine: <line>
  rightFileStartOffset: 1
  rightFileEndOffset: 1
  status: "active"

Then post the summary (no filePath = general PR comment):

mcp__ado__repo_create_pull_request_thread
  repositoryId: "<repo ID>"
  pullRequestId: <PR ID>
  content: "**Verdict**: <verdict>\n\nReviewed N files — N comments.\n\n<overall impression>"
  status: "active"

With patches: For each fix, post a comment with the issue AND the specific patch:

<issue description — written like a colleague, not a linting tool>

<details>
<summary>Proposed fix (click to expand)</summary>

\`\`\`diff
<unified diff for this specific file only>
\`\`\`

To apply: \`git apply fix.patch\`
</details>

CRITICAL — diff rendering in <details> blocks:

1. Blank line after </summary> is mandatory — without it, ADO won't process the code fence as markdown
2. NEVER HTML-encode diff content — write raw <p>, <span>, <div>, NOT <p>, <span>, <div>. The code fence handles escaping for display. HTML-encoding creates double-encoding that shows literal < text to the reader.
3. Always include the triple-backtick code fence with diff language tag — without it, HTML tags in the diff get parsed as actual HTML

For non-fixable issues (QUESTION): regular comment without a patch.

Summary thread (no filePath):

**Review with proposed fixes**

Reviewed <N> files — <M> issues found, <K> with proposed patches.

| # | File | Issue | Patch |
|---|------|-------|-------|
| 1 | `hero.js` L42 | Missing null check | included |
| 2 | `file.js` L10 | Edge case question | no patch |

<details>
<summary>Full combined patch (click to expand)</summary>

\`\`\`diff
<full unified patch combining all fixes>
\`\`\`

To apply all: \`git apply combined-fix.patch\`
</details>

If a comment fails to post: log the error, continue posting remaining, list failures at the end.

Set vote

Ask the user what vote to set:

• Approve — no critical issues
• Approve with suggestions — minor improvements
• Request changes — critical issues
• Skip voting — comments only

Never auto-approve or auto-decline without explicit user confirmation.

Save session

Save review state for future follow-up (enables /dx-pr-review <PR URL> follow-up mode later):

mkdir -p .ai/pr-reviews

Write .ai/pr-reviews/pr-<id>.md:

# PR #<id> — <title> (Review)

**Author:** <name>
**Branch:** <sourceBranch> → <targetBranch>
**Repo:** <repoName> (ID: <repoId>)
**Project:** <ADO project name>
**Last reviewed:** <ISO date>
**Review commit:** <SHA>
**Status:** reviewed | follow-up-needed | complete

## My Threads

### Thread #<threadId> | <status>

- **File:** <filePath or 'General'>
- **Line(s):** <range or 'N/A'>
- **Severity:** MUST-FIX | QUESTION
- **Comment:** <my review comment text>
- **Thread ID:** <ADO thread ID>
- **Posted:** <ISO date>
- **Comment count at save:** <number of comments in thread>
- **Patch posted:** yes | no
- **Follow-up status:** pending | addressed | argued | ignored

Review commit = current SHA of origin/<sourceBranch>:

git -C <repoPath> rev-parse origin/<sourceBranch>

Cleanup per-PR

# Worktree (if patches generated)
git worktree remove <worktreePath> --force 2>/dev/null

# Temp patch file
rm -f /tmp/dx-propose-pr-<id>.patch

Then check "More PRs?".

More PRs?

• yes — more selected PRs remain → go to "Process next PR result"
• no — all PRs processed → go to "Print aggregate summary + cleanup"

Print aggregate summary + cleanup

Print:

## Review Summary — <repo name>

| # | PR | Title | Verdict | Comments | Patches |
|---|-----|-------|---------|----------|---------|
| 1 | #12345 | Fix login bug | Approved | 2 | 1 |
| 2 | #12346 | Add feature X | Changes requested | 5 | 3 |
| 3 | #12347 | Update docs | Skipped | — | — |

**Reviewed:** <N> PRs | **Approved:** <N> | **Changes requested:** <N> | **Skipped:** <N>
**Sessions saved:** .ai/pr-reviews/pr-<id>.md (for follow-up with `/dx-pr-review`)

If a temp clone was created in "Prepare repo access":

rm -rf /tmp/dx-review-<repo>

Examples

1. /dx-pr-review-all — Lists 5 active PRs in the current repo (excluding your own), presents the list for selection. You type "1 3" to review PRs #1 and #3. Each PR gets the full review flow: analysis, findings presentation, optional patch generation, comment posting, and vote.

2. /dx-pr-review-all https://dev.azure.com/myorg/MyProject/_git/Other-Repo 10 — Lists up to 10 active PRs from a different repo, fetches branches in one batch, spawns review agents in parallel, then processes results one at a time for approval.

3. /dx-pr-review-all https://dev.azure.com/myorg/MyProject/_git/MyRepo/pullrequest/12345 — Detects a single PR URL, skips the listing step, and delegates directly to /dx-pr-review for that specific PR.

Troubleshooting

• "No active PRs in " Cause: All PRs are either completed, abandoned, or created by you (own PRs are filtered out). Fix: Check ADO directly to confirm PR status. If you want to review your own PRs, use /dx-pr-review <PR URL> directly.

• Review agent times out on large PRs Cause: The PR has hundreds of changed files or very large diffs that exceed the agent's context window. Fix: Review the PR manually for the largest files. For subsequent reviews, consider breaking large PRs into smaller ones.

• Comments fail to post to ADO Cause: ADO PAT lacks "Pull Requests: Contribute" permission, or the PR was completed/abandoned between review and posting. Fix: The skill logs failures and continues posting remaining comments. Check your PAT permissions. Failed comments are listed in the summary.

Rules

• List before reviewing — always show the PR list and let the user choose which to review
• Parallel analysis — spawn all dx-pr-reviewer agents in a single message for concurrent execution
• Sequential approval — present each PR's findings one at a time for user approval
• Full review flow per PR — each PR gets the identical flow as /dx-pr-review: present → patch → post → vote → save session
• Respect user selection — if user says "1 3", only review those two
• Don't duplicate — if /dx-pr-review was already run on a PR in this session, skip it and note "already reviewed"
• Current repo as default — when no arguments, detect repo from git remote
• MCP tools are deferred — always load via ToolSearch before first use
• Pre-fetch branches — do one git fetch for all branches before spawning agents
• Agent handles analysis — diff reading, context loading, and code review happen inside the agent, keeping main context lean
• Never push to their branch — generate patches, never commit or push to the author's branch
• Worktree isolation — all fix generation happens in an isolated worktree. Never modify local working directory
• Patch, not push — output is always a patch in a PR comment. The author decides whether to apply
• Collapsible patches — use <details> with mandatory blank line after </summary>, never HTML-encode diff content
• Include apply instructions — every patch comment includes git apply instructions
• Combined + individual — per-file patches on relevant lines AND a combined patch in the summary
• Ask before acting — never approve, decline, or post without user confirmation
• Session persistence — always save to .ai/pr-reviews/pr-<id>.md after posting each PR. Enables follow-up via /dx-pr-review
• Continue on failure — if one comment fails to post, log and continue