Propel Code Review Smoke

Run async diff-based code reviews using the Propel Review API, poll for completion, retrieve structured findings, and send comment feedback. Use when reviewing the current branch diff or any Git diff with Propel.

Propel Review API Skill

Use this guide to interact with the Propel Review API from an AI agent. Always target the production API unless told otherwise.

Purpose

Run async, diff-based code reviews via the production API and retrieve comments.

Quick Permission Check

Before rollout to a new workspace/repo, run this from the target repository root:

bash

plugins/propel-code-review/skills/propel-code-review/scripts/smoke_test_permissions.sh

It validates:

• good token + connected repo
• good token + unconnected repo
• invalid token + connected repo

Pre-flight: Verify API Key

Before making any API call, check whether PROPEL_API_KEY is set:

bash

if [ -n "$PROPEL_API_KEY" ]; then echo "PROPEL_API_KEY is set"; else echo "PROPEL_API_KEY is not set"; fi

If the variable is empty, unset, or you just received a 401/403 from the Review API, do not attempt any API calls with the current value. Follow these steps to capture a fresh token — each step is a separate action:

Step 1 — Tell the user and open the browser. Send this message and run the Bash command in the same response (in parallel):

Message to user:

PROPEL_API_KEY is not set. Opening the token creation page: https://app.propelcode.ai/administration/settings?tab=review-api-tokens&token_name=Claude+Code&scopes=reviews:read,reviews:write The name and scopes are pre-filled. Click Create token, copy it, and paste it here.

Bash command:

bash

URL="https://app.propelcode.ai/administration/settings?tab=review-api-tokens&token_name=Claude+Code&scopes=reviews:read,reviews:write"
if command -v xdg-open >/dev/null; then xdg-open "$URL"; else open "$URL"; fi

Step 2 — Wait for the user to paste the token. Do not proceed until the user pastes a value starting with rev_. If the value doesn't start with rev_, tell them it doesn't look valid and ask them to try again.

Step 3 — Once you have a valid token, persist it and load it into the session. Run this in a single Bash call (replace <TOKEN> with the actual token):

bash

case "$SHELL" in */zsh) SHELL_RC="$HOME/.zshrc" ;; */bash) SHELL_RC="$HOME/.bashrc" ;; *) SHELL_RC="" ;; esac; if [ -z "$SHELL_RC" ] && [ -f "$HOME/.zshrc" ]; then SHELL_RC="$HOME/.zshrc"; fi; if [ -z "$SHELL_RC" ] && [ -f "$HOME/.bashrc" ]; then SHELL_RC="$HOME/.bashrc"; fi; if [ -n "$SHELL_RC" ]; then printf '\n# Propel Review API token\nexport PROPEL_API_KEY="%s"\n' "<TOKEN>" >> "$SHELL_RC" && echo "Saved to $SHELL_RC"; else echo "No shell profile found"; fi; export PROPEL_API_KEY="<TOKEN>"

Tell the user where the key was saved (e.g. "Saved to ~/.zshrc").

Step 4 — Continue with the review workflow.

Setup (Manual)

If you prefer to set the token yourself ahead of time:

bash

export PROPEL_API_KEY="rev_..."

The token must be a Review API token (scoped to both reviews:write and reviews:read).

Base URL

text

https://api.propelcode.ai

Authentication

Use a bearer token in the Authorization header:

text

Authorization: Bearer $PROPEL_API_KEY

Endpoints (Only These)

Do not assume any other Review APIs exist. Only use the async endpoints below.

Create Review (Async)

POST /v1/reviews

Request body:

json

{
  "diff": "string (required)",
  "repository": "string (required)",
  "base_commit": "string (required)",
  "head_commit_sha": "string (optional)",
  "branch": "string (optional)"
}

Constraints:

• diff max size: 1,000,000 bytes
• repository max length: 255
• base_commit max length: 255
• head_commit_sha max length: 255
• branch max length: 255

Notes:

• base_commit should be a commit that exists in the remote repo history (typically the base commit of the branch you are reviewing).
• repository should be the canonical repo slug (for example, owner/repo) derived from the git remote URL.

Response (202):

json

{
  "review_id": "uuid",
  "status": "queued",
  "repository": "owner/repo",
  "base_commit": "sha",
  "created_at": "...",
  "updated_at": "..."
}

Get Review Status/Results

GET /v1/reviews/:review_id

Response (200):

json

{
  "review_id": "uuid",
  "status": "queued|running|completed|failed",
  "poll_after_ms": 3000,
  "comments": [
    {
      "comment_id": "string",
      "file_path": "path",
      "line": 123,
      "message": "...",
      "severity": "error|warning|info"
    }
  ],
  "error": {
    "code": "generation_failed",
    "message": "..."
  }
}

Post Comment Feedback

POST /v1/reviews/:review_id/comments/feedback

Request body:

json

{
  "comment_id": "string (required)",
  "incorporated": true,
  "notes": "string (optional)"
}

Response (200):

json

{
  "review_id": "uuid",
  "comment_id": "string",
  "incorporated": true
}

Scripts Used by This Skill

Use the helper scripts in scripts/ instead of ad-hoc inline curl loops. Paths below are relative to this skill directory:

• scripts/create_review.sh wraps POST /v1/reviews.
• scripts/poll_review.sh wraps polling GET /v1/reviews/:review_id until terminal status (completed or failed) or timeout.
• scripts/post_comment_feedback.sh wraps POST /v1/reviews/:review_id/comments/feedback.

Approval-Friendly Prefixes (One-Time)

If your client supports prefix-based trust/approval, approve these once before running this skill:

• scripts/create_review.sh
• scripts/poll_review.sh
• scripts/post_comment_feedback.sh
• git diff
• git rev-parse
• git remote get-url
• gh pr view
• jq

Workflow (Recommended)

1. Resolve the base branch (PR base when available; otherwise remote default branch):
   • BASE_BRANCH=$(gh pr view --json baseRefName -q '.baseRefName' 2>/dev/null || git remote show origin | sed -n '/HEAD branch/s/.*: //p')
2. Compute the base commit (must exist in the remote repo history):
   • git rev-parse "$BASE_BRANCH"
3. Compute the current head commit and branch (for local review dedup context):
   • git rev-parse HEAD
   • git rev-parse --abbrev-ref HEAD
4. Compute the repository slug:
   • git remote get-url origin | sed -E 's#([e-mail]:|https://github.com/)##; s/\\.git$//'
5. Generate the diff:
   • git diff "$BASE_BRANCH" > /tmp/review_api.diff
6. Call scripts/create_review.sh with diff, base commit, repository slug, head commit, and branch.
7. Handle create-review failures before polling (script exits non-zero and prints the API response body):
   • 401/403: token invalid/expired/missing scope. Stop and ask user to refresh token.
   • 404: repository is not connected to the Propel workspace (or slug is wrong). Stop and ask user to connect/fix repo slug.
   • 400/413: invalid request or diff too large. Stop and show actionable fix.
   • 5xx: transient API error. Retry with bounded backoff, then stop and report if still failing.
8. Poll with scripts/poll_review.sh until status is completed or failed. The script keeps the existing 15 minute timeout budget, and honors poll_after_ms from the API when present.
9. Present comments to the user with file/line context.
10. For each comment, determine whether it is valid and applicable to the code.
11. If valid, incorporate the change in the codebase. If invalid, do not change the codebase.
12. Immediately call scripts/post_comment_feedback.sh for each comment with comment_id, incorporated true/false, and brief notes explaining the decision. Do not wait for user confirmation.

Example (Production)

bash

BASE_BRANCH=$(gh pr view --json baseRefName -q '.baseRefName' 2>/dev/null || git remote show origin | sed -n '/HEAD branch/s/.*: //p')
BASE_COMMIT=$(git rev-parse "$BASE_BRANCH")
HEAD_COMMIT=$(git rev-parse HEAD)
BRANCH=$(git rev-parse --abbrev-ref HEAD)
REPO_SLUG=$(git remote get-url origin | sed -E 's#([e-mail]:|https://github.com/)##; s/\\.git$//')
git diff "$BASE_BRANCH" > /tmp/review_api.diff

CREATE_RESPONSE=$(
  scripts/create_review.sh \
    --diff-file /tmp/review_api.diff \
    --repo "$REPO_SLUG" \
    --base-commit "$BASE_COMMIT" \
    --head-commit-sha "$HEAD_COMMIT" \
    --branch "$BRANCH"
)

REVIEW_ID=$(echo "$CREATE_RESPONSE" | jq -r '.review_id // empty')
if [ -z "$REVIEW_ID" ]; then
  echo "$CREATE_RESPONSE"
  exit 1
fi

scripts/poll_review.sh \
  --review-id "$REVIEW_ID" \
  --max-attempts 30 \
  --sleep-seconds 30 \
  --output-file /tmp/review_api.result.json

# The poller uses the 30x30s budget above, but will follow API-provided
# poll_after_ms hints when the server returns them.

cat /tmp/review_api.result.json

# Example feedback posts after deciding incorporate true/false per comment.
jq -c '.comments[]?' /tmp/review_api.result.json | while read -r comment; do
  COMMENT_ID=$(echo "$comment" | jq -r '.comment_id // empty')
  if [ -z "$COMMENT_ID" ]; then
    continue
  fi
  scripts/post_comment_feedback.sh \
    --review-id "$REVIEW_ID" \
    --comment-id "$COMMENT_ID" \
    --incorporated true \
    --notes "Applied in this branch."
done

Troubleshooting

• 401/403 — re-run the pre-flight check above. The token may be missing, expired, or missing scopes. Guide the user to generate a new one at: https://app.propelcode.ai/administration/settings?tab=review-api-tokens&token_name=Claude+Code&scopes=reviews:read,reviews:write
• 404 {"error":"Repository not found"} — the repository string does not match a repo connected to the account. Treat this as an access/config problem, not a retryable failure.
• 413 — the diff exceeded the 1,000,000 byte limit.

Permission Handling Contract

If review creation returns 401, 403, or 404, do all of the following:

1. Stop the review flow immediately (do not poll and do not retry in a loop).
2. Report the exact repository slug used.
3. Report an actionable next step:
   • 401/403: "refresh token and confirm reviews:read + reviews:write scopes".
   • 404: "connect this repository in Propel workspace, or correct owner/repo slug".
4. Mark the run as blocked until user intervention.

Notes for Agents

• Do not log or expose tokens in output.
• Always use https://api.propelcode.ai until told otherwise.
• Use scripts/create_review.sh, scripts/poll_review.sh, and scripts/post_comment_feedback.sh instead of inline curl commands.
• The agent must decide whether each comment is valid, incorporate fixes when valid, and report feedback automatically via the feedback endpoint using the comment_id from the review response (no user confirmation required).