Skip to content

Improve classify workflow #31

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
amrit110 merged 1 commit into from
Dec 23, 2025
Merged

Improve classify workflow #31

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
188 changes: 44 additions & 144 deletions .github/workflows/fix-remote-pr.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,20 +39,20 @@ jobs:
with:
path: bot-repo

- name: Get PR details
- name: Get PR basic info for checkout
id: pr-details
run: |
REPO="${{ github.event.inputs.target_repo }}"
PR_NUMBER="${{ github.event.inputs.pr_number }}"

# Get PR information including mergeable status
# Get minimal PR info needed for checkout (full details fetched by CLI)
PR_INFO=$(gh pr view $PR_NUMBER --repo "$REPO" --json \
title,author,headRefName,headRepository,headRepositoryOwner,statusCheckRollup,baseRefName,mergeable)
title,author,headRefName,baseRefName,mergeable)

echo "PR Info:"
echo "$PR_INFO" | jq '.'

# Extract details
# Extract details for checkout step
HEAD_REF=$(echo "$PR_INFO" | jq -r '.headRefName')
BASE_REF=$(echo "$PR_INFO" | jq -r '.baseRefName')
PR_TITLE=$(echo "$PR_INFO" | jq -r '.title')
Expand All @@ -65,13 +65,8 @@ jobs:
echo "pr-author=$PR_AUTHOR" >> $GITHUB_OUTPUT
echo "mergeable=$MERGEABLE" >> $GITHUB_OUTPUT

# Get failed checks
FAILED_CHECKS=$(echo "$PR_INFO" | jq -c '[.statusCheckRollup[] | select(.conclusion == "FAILURE")]')
echo "failed-checks=$FAILED_CHECKS" >> $GITHUB_OUTPUT

FAILED_COUNT=$(echo "$FAILED_CHECKS" | jq 'length')
echo "Found $FAILED_COUNT failed checks"
echo "PR mergeable status: $MERGEABLE"
echo "Branch: $HEAD_REF → $BASE_REF"
echo "Mergeable: $MERGEABLE"
env:
GH_TOKEN: ${{ secrets.ORG_ACCESS_TOKEN }}

Expand Down Expand Up @@ -122,63 +117,6 @@ jobs:
fi
continue-on-error: true

- name: Get failure logs
id: get-logs
run: |
REPO="${{ github.event.inputs.target_repo }}"
FAILED_CHECKS='${{ steps.pr-details.outputs.failed-checks }}'

echo "Extracting failure logs from failed checks..."

# Extract job URLs and fetch their logs
> /tmp/failure-logs.txt

echo "$FAILED_CHECKS" | jq -r '.[].detailsUrl' | while read -r JOB_URL; do
if [ -n "$JOB_URL" ]; then
# Extract run ID and job ID from URL
# Format: https://github.com/OWNER/REPO/actions/runs/RUN_ID/job/JOB_ID
RUN_ID=$(echo "$JOB_URL" | grep -oP 'runs/\K[0-9]+')
JOB_ID=$(echo "$JOB_URL" | grep -oP 'job/\K[0-9]+')

echo "Fetching logs for job $JOB_ID in run $RUN_ID"

# Fetch full run logs
gh run view "$RUN_ID" --repo "$REPO" --log > /tmp/full-logs.txt 2>&1 || continue

# Extract relevant failure sections (errors, failures, security issues)
grep -i -E "(error|fail|traceback|exception|exit code [^0]|vulnerability|vulnerabilities|CVE-|GHSA-|audit|found [0-9]+ known)" /tmp/full-logs.txt | tail -2000 >> /tmp/failure-logs.txt || true

# Also get the last 1000 lines which often contain the summary
tail -1000 /tmp/full-logs.txt >> /tmp/failure-logs.txt

rm -f /tmp/full-logs.txt
fi
done

# If we got no logs, try fallback method
if [ ! -s /tmp/failure-logs.txt ]; then
echo "No logs from specific jobs, trying fallback..."
HEAD_REF="${{ steps.pr-details.outputs.head-ref }}"
RUN_ID=$(gh run list --repo "$REPO" --branch "$HEAD_REF" --limit 5 \
--json databaseId,status,conclusion \
--jq '.[] | select(.conclusion == "failure") | .databaseId' | head -1)

if [ -n "$RUN_ID" ]; then
gh run view $RUN_ID --repo "$REPO" --log > /tmp/full-logs.txt 2>&1
grep -i -E "(error|fail|traceback|exception|exit code [^0]|vulnerability|vulnerabilities|CVE-|GHSA-|audit|found [0-9]+ known)" /tmp/full-logs.txt | tail -2000 > /tmp/failure-logs.txt || true
tail -1000 /tmp/full-logs.txt >> /tmp/failure-logs.txt
rm -f /tmp/full-logs.txt
fi
fi

# Log the size of extracted logs
if [ -f /tmp/failure-logs.txt ] && [ -s /tmp/failure-logs.txt ]; then
echo "Extracted $(wc -l < /tmp/failure-logs.txt) lines of failure logs"
else
echo "No failure logs could be extracted" > /tmp/failure-logs.txt
fi
env:
GH_TOKEN: ${{ secrets.ORG_ACCESS_TOKEN }}

- name: Setup Python for classification
uses: actions/setup-python@v6
Expand All @@ -192,80 +130,54 @@ jobs:

- name: Analyze failure type with Claude
id: analyze
working-directory: target-repo
working-directory: bot-repo
run: |
FAILED_CHECKS='${{ steps.pr-details.outputs.failed-checks }}'
MERGEABLE='${{ steps.pr-details.outputs.mergeable }}'

echo "Analyzing failures with Claude-based classifier..."
echo "$FAILED_CHECKS" | jq -r '.[] | "\(.name): \(.conclusion)"'

# Check for merge conflicts first via PR mergeable status
if [ "$MERGEABLE" = "CONFLICTING" ]; then
echo "failure-type=merge_conflict" >> $GITHUB_OUTPUT
echo "failed-check-names=merge-conflict" >> $GITHUB_OUTPUT
echo "confidence=1.0" >> $GITHUB_OUTPUT
echo "reasoning=PR has merge conflicts with base branch (mergeable=CONFLICTING)" >> $GITHUB_OUTPUT
echo "✓ Detected merge conflicts via PR mergeable status"
exit 0
fi

# Also check git status as fallback
if git status | grep -q "Unmerged paths\|merge conflict"; then
echo "failure-type=merge_conflict" >> $GITHUB_OUTPUT
echo "failed-check-names=merge-conflict" >> $GITHUB_OUTPUT
echo "confidence=1.0" >> $GITHUB_OUTPUT
echo "reasoning=Git merge conflicts detected in working tree" >> $GITHUB_OUTPUT
echo "✓ Detected merge conflicts via git status"
exit 0
fi

# Prepare PR context JSON
PR_INFO=$(jq -n \
--arg repo "${{ github.event.inputs.target_repo }}" \
--arg pr_number "${{ github.event.inputs.pr_number }}" \
--arg pr_title "${{ steps.pr-details.outputs.pr-title }}" \
--arg pr_author "${{ steps.pr-details.outputs.pr-author }}" \
--arg base_ref "${{ steps.pr-details.outputs.base-ref }}" \
--arg head_ref "${{ steps.pr-details.outputs.head-ref }}" \
'{repo: $repo, pr_number: $pr_number, pr_title: $pr_title, pr_author: $pr_author, base_ref: $base_ref, head_ref: $head_ref}')

# Check if failure logs file exists
if [ ! -f /tmp/failure-logs.txt ]; then
echo "⚠️ Warning: Failure logs file not found at /tmp/failure-logs.txt"
echo "No failure logs could be extracted" > /tmp/failure-logs.txt
fi

echo "Failure logs file size: $(wc -c < /tmp/failure-logs.txt) bytes"
echo "Failure logs file lines: $(wc -l < /tmp/failure-logs.txt) lines"
REPO="${{ github.event.inputs.target_repo }}"
PR_NUMBER="${{ github.event.inputs.pr_number }}"

# Run Python classifier (using installed CLI entry point)
# Pass file path instead of content to avoid bash variable size limits
cd ../bot-repo
echo "Analyzing PR $REPO#$PR_NUMBER with Claude-based classifier..."
echo ""

echo "Running classifier..."
# Run classify command and save JSON to file
if ! aieng-bot classify \
--pr-info "$PR_INFO" \
--failed-checks "$FAILED_CHECKS" \
--failure-logs-file /tmp/failure-logs.txt \
--output-format github > /tmp/classification-output.txt 2> /tmp/classification-error.txt; then
--repo "$REPO" \
--pr "$PR_NUMBER" \
--output /tmp/classification.json 2> /tmp/classification-error.txt; then
echo "❌ Classification failed with exit code $?"
echo "STDOUT:"
cat /tmp/classification-output.txt || echo "(no stdout)"
echo ""
echo "STDERR:"
cat /tmp/classification-error.txt || echo "(no stderr)"
exit 1
fi

CLASSIFICATION=$(cat /tmp/classification-output.txt)

# Parse and output results
echo "$CLASSIFICATION" >> $GITHUB_OUTPUT
# Parse JSON and output to GitHub Actions
echo "Classification results:"
echo "$CLASSIFICATION"
cat /tmp/classification.json
echo ""

# Extract fields from JSON
FAILURE_TYPE=$(jq -r '.failure_type' /tmp/classification.json)
CONFIDENCE=$(jq -r '.confidence' /tmp/classification.json)
FAILED_CHECK_NAMES=$(jq -r '.failed_check_names | join(",")' /tmp/classification.json)
REASONING=$(jq -r '.reasoning' /tmp/classification.json)
RECOMMENDED_ACTION=$(jq -r '.recommended_action' /tmp/classification.json)

# Output to GitHub Actions
echo "failure-type=$FAILURE_TYPE" >> $GITHUB_OUTPUT
echo "confidence=$CONFIDENCE" >> $GITHUB_OUTPUT
echo "failed-check-names=$FAILED_CHECK_NAMES" >> $GITHUB_OUTPUT
{
echo "reasoning<<EOF_REASONING"
echo "$REASONING"
echo "EOF_REASONING"
} >> $GITHUB_OUTPUT
{
echo "recommended-action<<EOF_ACTION"
echo "$RECOMMENDED_ACTION"
echo "EOF_ACTION"
} >> $GITHUB_OUTPUT
env:
ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
GH_TOKEN: ${{ secrets.ORG_ACCESS_TOKEN }}

- name: Prepare for agent execution
id: prepare
Expand Down Expand Up @@ -303,21 +215,12 @@ jobs:
cp -r bot-repo/.claude target-repo/.claude
echo "✓ Skills copied to target-repo/.claude/"

# Copy failure logs to target-repo so agent can read it
if [ -f /tmp/failure-logs.txt ]; then
cp /tmp/failure-logs.txt target-repo/.failure-logs.txt
echo "✓ Failure logs copied to target-repo/.failure-logs.txt"
else
echo "⚠️ Warning: No failure logs found at /tmp/failure-logs.txt"
fi

# LAYER 1: Add bot files to git's local exclude list (safety net)
echo "Excluding bot files from git..."
{
echo "# AI Engineering Bot temporary files - DO NOT COMMIT"
echo ".claude/"
echo ".pr-context.json"
echo ".failure-logs.txt"
} >> target-repo/.git/info/exclude
echo "✓ Bot files added to .git/info/exclude"

Expand Down Expand Up @@ -386,7 +289,7 @@ jobs:
echo "Verifying bot temporary files are not staged or committed..."

# Check if bot files are in staging area
BOT_FILES_STAGED=$(git diff --cached --name-only | grep -E "^\.claude/|^\.pr-context\.json$|^\.failure-logs\.txt$" || true)
BOT_FILES_STAGED=$(git diff --cached --name-only | grep -E "^\.claude/|^\.pr-context\.json$" || true)

if [ -n "$BOT_FILES_STAGED" ]; then
echo "❌ ERROR: Bot temporary files found in staging area!"
Expand All @@ -399,7 +302,7 @@ jobs:

# Check if bot files exist in any unpushed commits
BASE_REF="${{ steps.pr-details.outputs.base-ref }}"
BOT_FILES_COMMITTED=$(git diff --name-only "origin/$BASE_REF...HEAD" | grep -E "^\.claude/|^\.pr-context\.json$|^\.failure-logs\.txt$" || true)
BOT_FILES_COMMITTED=$(git diff --name-only "origin/$BASE_REF...HEAD" | grep -E "^\.claude/|^\.pr-context\.json$" || true)

if [ -n "$BOT_FILES_COMMITTED" ]; then
echo "⚠️ WARNING: Bot temporary files found in commits!"
Expand All @@ -413,7 +316,7 @@ jobs:

# Cleanup bot files from working directory
echo "Cleaning up bot temporary files..."
rm -rf .claude/ .pr-context.json .failure-logs.txt 2>/dev/null || true
rm -rf .claude/ .pr-context.json 2>/dev/null || true
echo "✓ Bot files cleaned up"

- name: Upload trace to GCS
Expand Down Expand Up @@ -548,9 +451,6 @@ jobs:
id: push-fixes
working-directory: target-repo
run: |
# Remove failure logs file (don't commit it to the repo)
rm -f .failure-logs.txt

# Check if Agent SDK created a commit (check if HEAD differs from remote)
git fetch origin ${{ steps.pr-details.outputs.head-ref }}

Expand Down
1 change: 1 addition & 0 deletions pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,6 +26,7 @@ dependencies = [
"click>=8.0.0",
"pyyaml>=6.0.2",
"rich>=13.0.0",
"python-dotenv>=1.0.0",
]

[project.urls]
Expand Down
Loading