Skip to content
Agentmart
Submit a tool

panel-review

Multi-agent panel review of any code target (a PR by id, a branch, the current branch, or specific paths) against project standards, with confidence-scored…

Skill by TJaenichen

IndexedMIT
Write a review

Chalk Labs

Need help deploying panel-review?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code. Read the repository before you give any tool access to your data.

npx agentmarket install skill/tjaenichen-panel-review

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Multi-agent panel review of any code target (a PR by id, a branch, the current branch, or specific paths) against project standards, with confidence-scored findings grouped by category. Can post the findings to the Azure DevOps PR when the target is a PR. Use when the user asks for a thorough review of a PR, a branch, or a piece of code.

Use cases: devops, cloud, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/panel-review

Tools it pre-approves 6

From allowed-tools in SKILL.md: the agent may use these without asking you each time.

  • Bash(git:*),
  • Bash(powershell:*),
  • Glob,
  • Grep,
  • Read,
  • Task

What's inside

Skill name
panel-review
Description
Multi-agent panel review of any code target (a PR by id, a branch, the current branch, or specific paths) against project standards, with confidence-scored findings grouped by category. Can post the findings to the Azure DevOps PR when the target is a PR. Use when the user asks for a thorough review of a PR, a branch, or a piece of code.
License
MIT
Allowed tools
Bash(git:*),, Bash(powershell:*),, Glob,, Grep,, Read,, Task
Instructions
491 lines in SKILL.md · view on GitHub

Before installing

  • Pre-approves some shell commands

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Panel Review

Command: `/panel-review $ARGUMENTS`

Five reviewers in parallel (standards, bugs, git history, tests, requirements), one scored and filtered report, and the human decides what happens next. Works on a PR, a branch, or a set of files; the PR-specific steps (work item gate, posting threads) only apply when a PR is in play.

## Usage

| Usage | Description |
|-------|-------------|
| `/panel-review` | Current branch vs `main`, plus its PR if one exists |
| `/panel-review 12345` | PR #12345 by ID |
| `/panel-review feature/12345_ShortTitle` | A branch vs `main`, plus its PR if one exists |
| `/panel-review src/Rules/ tests/RulesTests.cs` | Specific files or directories from the working tree; no PR involved |

## Azure DevOps Configuration

- **Server:** https://devops.example.com/
- **Project:** MyProject (ID: `<PROJECT_ID>`)
- **Repository:** Payments (ID: `<REPO_ID>`)
- **Auth:** Windows Integrated Security (UseDefaultCredentials)

## API Endpoints

```
Base URL: https://devops.example.com/DefaultCollection/MyProject/_apis/git/repositories/<REPO_ID>
```

| Operation | Method | Endpoint |
|-----------|--------|----------|
| Get PR by branch | GET | `/pullrequests?searchCriteria.sourceRefName=refs/heads/{branch}&api-version=6.0` |
| Get PR by ID | GET | `/pullrequests/{id}?api-version=6.0` |
| Get PR work items | GET | `/pullrequests/{id}/workitems?api-version=6.0` |
| Post thread | POST | `/pullrequests/{id}/threads?api-version=6.0` |
| Get threads | GET | `/pullrequests/{id}/threads?api-version=6.0` |

## Your Task

Execute the following 10-step code review workflow.

---

### Step 1: Resolve the Target

Parse `$ARGUMENTS` and decide the **mode**:

| Argument shape | Mode | How to resolve |
|---|---|---|
| A number (e.g. `12345`) | `pr` | Fetch the PR by ID via GET `/pullrequests/{id}?api-version=6.0` |
| One or more existing paths (files or directories) | `paths` | Review the working-tree content of those paths. No PR lookup. |
| A branch name (`git rev-parse --verify <name>` or `origin/<name>` succeeds) | `branch` | Diff it against `main`. Also search for a PR on that branch; if one exists, switch to `pr` mode with that PR. |
| Empty | `branch` | Same as above with the current branch (`git rev-parse --abbrev-ref HEAD`). If the current branch is `main`, ask what to review and stop. |

Store: `mode`, `sourceBranch`, `targetBranch` (default `main`), and, in `pr` mode, `prId`, `title`, `status`, `createdBy`, `description`. In `branch` mode set `title` to the branch name and `description` to the last commit message; in `paths` mode set `title` to the path list.

If a number was given but no such PR exists, tell the user and stop.

**PowerShell template:**
```powershell
$repoId = "<REPO_ID>"
$baseUrl = "https://devops.example.com/DefaultCollection/MyProject/_apis/git/repositories/$repoId"

# By branch
$result = Invoke-RestMethod -Uri "$baseUrl/pullrequests?searchCriteria.sourceRefName=refs/heads/$branch&api-version=6.0" -UseDefaultCredentials
$pr = $result.value[0]

# By ID
$pr = Invoke-RestMethod -Uri "$baseUrl/pullrequests/$prId`?api-version=6.0" -UseDefaultCredentials
```

---

### Step 2: Eligibility Checks

**`pr` mode only.** In `branch` mode, try to parse a work item ID from the branch name (`^(feature|bugfix)/(\d+)[-_]`) and fetch it so Agent 5 has requirements context; if there is none, note "no work item context" and skip Agent 5. In `paths` mode there is no work item context; skip Agent 5. Then go to Step 3.

In `pr` mode, check the following before proceeding:

- **No linked work item**: Fetch linked work items via GET `/pullrequests/{id}/workitems?api-version=6.0` and check the `value` array. If empty, **stop the review** and tell the user: "PR has no linked DevOps work item. Link a User Story, Bug, or Task to the PR before requesting a review." This is a hard blocker — do not proceed. Store the work item IDs for use by Agent 5.
- If **status is `completed`** (merged/closed): Warn "PR is already completed" but proceed since user explicitly invoked the review.
- If **status is `abandoned`**: Warn "PR is abandoned" but proceed.
- If **a PR thread comment already contains `### Panel Review`**: Fetch threads via GET `/pullrequests/{id}/threads?api-version=6.0` and check if any thread's first comment content contains `### Panel Review`. If found, warn "PR already has a review comment" but proceed (user re-invoked = intent to re-review).

For warnings: display the warning, then continue with the review. For the missing work item check: **stop and do not proceed**.

---

### Step 3: Collect CLAUDE.md Review Criteria

Read all CLAUDE.md files in the repository to gather project-specific review standards:

```bash
find "$(git rev-parse --show-toplevel)" -name "CLAUDE.md" -not -path "*/node_modules/*" -not -path "*/.git/*"
```

Read each file. These contain the review criteria (Clean Architecture rules, testing standards, Dapper patterns, etc.).

---

### Step 4: Get the Code Under Review

- **`pr` / `branch` mode** — the complete diff against the target branch, using local git:
  ```bash
  git fetch origin
  git diff origin/main...origin/{sourceBranch} --stat
  git diff origin/main...origin/{sourceBranch}
  ```
  Use the target branch from the PR if it's not `main` (replace `origin/main` with `origin/{targetBranch}`).
- **`paths` mode** — the current content of every file under the given paths (use Glob for directories, skip binaries and generated files). If the branch has changes to those paths vs `main`, also compute `git diff main...HEAD -- <paths>` and hand both to the agents: they review the whole content but flag changed lines with priority. If there is no diff, everything is in scope and the "only flag changed code" instruction below does not apply.

If the diff is very large (>5000 lines), note this and still proceed — the parallel agents will handle chunks.

Store: list of changed files, total lines changed, which architectural layers are affected.

**Determine affected layers** from file paths:
- `Contoso.Payments.API/` = Presentation
- `Contoso.Payments.API.Application/` = Application
- `Contoso.Payments.API.Domain/` = Domain
- `Contoso.Payments.API.Infrastructure/` = Infrastructure

From TJaenichen/claude-code-harness, MIT.

More from TJaenichen/claude-code-harness

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Aug 16, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from TJaenichen

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get panel-review running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying panel-review?

CallGet help
Cite this page

panel-review skill. Agent Mart. https://agentmart.dev/skills/tjaenichen-panel-review (last updated 2026-10-24)