File Issues from Evaluation

Overview

This SOP parses evaluation reports from brazil-bench attempts and creates individual GitHub issues on each attempt's repository for every shortcoming, warning, or improvement opportunity identified during the review.

Parameters

• attempt_repo (required): Repository name (e.g., 2025-12-13-python-claude-hive)
• evaluation_path (optional, default: ./results/{attempt_repo}.md): Path to evaluation report
• dry_run (optional, default: false): If true, show issues that would be created without creating them

Steps

1. Load Evaluation Report

Read the evaluation report for the specified attempt.

Constraints:

• You MUST verify the evaluation report exists before proceeding
• You MUST parse markdown sections to identify shortcomings
• You MUST NOT proceed if the report is missing or malformed

# Verify report exists
ls ./results/{attempt_repo}.md

# Read the report content
cat ./results/{attempt_repo}.md

2. Verify Repository Access

Confirm the attempt repository exists and is writable.

Constraints:

• You MUST verify the repo exists on GitHub
• You MUST check that issues are enabled on the repo
• You SHOULD check for existing issues to avoid duplicates
• You SHOULD use default GitHub labels (no custom labels required)

# Verify repo exists
gh repo view brazil-bench/{attempt_repo}

# List existing issues to check for duplicates
gh issue list -R brazil-bench/{attempt_repo} --limit 100

# List available labels (use default GitHub labels)
gh label list -R brazil-bench/{attempt_repo}

Default GitHub Labels to Use:

Using default labels avoids permission issues (creating custom labels requires admin access).

3. Extract Shortcomings from Report

Parse the evaluation report to identify all issues to file.

Systematic Extraction Commands:

Use these commands to programmatically extract shortcomings:

# Extract missing requirements (unchecked items)
grep -E "^- \[ \]" ./results/{attempt_repo}.md

# Extract test quality warnings
grep -A 20 "## Test Quality Warning\|## Test Skip Analysis" ./results/{attempt_repo}.md

# Extract weaknesses section
grep -A 10 "## Weaknesses" ./results/{attempt_repo}.md

# Extract compliance score
grep -E "Spec Compliance.*[0-9]+/[0-9]+" ./results/{attempt_repo}.md

# Extract skip ratio
grep -E "Skip Ratio.*[0-9]+%" ./results/{attempt_repo}.md

Categories of Issues to Extract:

3a. Missing Requirements

Look in the "Requirements Checklist" section for unchecked items:

Pattern: Lines starting with - [ ] indicate missing/partial requirements

# Extract all missing requirements
grep -E "^- \[ \]" ./results/{attempt_repo}.md

Issue Template:

Title: [Missing] {requirement description}
Label: enhancement
Body:
## Requirement
{description}

## Notes from Evaluation
{any additional context from the report}

## Suggested Fix
{if available}

---
Filed from evaluation: results/{attempt_repo}.md

3b. Test Quality Warnings

Look for the "Test Quality Warning" or "Test Skip Analysis" sections:

Patterns to detect:

• Skip ratio > 20%
• Tests with "not yet implemented" messages
• Integration tests that can't run

Issue Template:

Title: [Test Quality] {description of test issue}
Label: bug
Body:
## Issue
{description of the test quality problem}

## Details
{skip breakdown, affected files, etc.}

## Impact
- Skip Ratio: {X}%
- Effective Tests: {Y} (vs {Z} total)

## Suggested Fix
{recommendations}

---
Filed from evaluation: results/{attempt_repo}.md

3c. Spec Compliance Issues

Look for compliance scores below 100%:

Patterns:

• Spec Compliance: X/16 where X < 16
• Partial implementations noted

Issue Template:

Title: [Compliance] Spec compliance at {X}% ({Y}/16 requirements)
Label: enhancement
Body:
## Current Status
{X}/16 requirements met ({percentage}%)

## Missing Requirements
{list of unchecked requirements}

## Partial Implementations
{list of partial implementations with notes}

---
Filed from evaluation: results/{attempt_repo}.md

3d. Skipped Tests

Look for ANY skipped tests in the "Metrics" or "Test Skip Analysis" sections:

Patterns to detect:

• Tests (Skipped) > 0 (zero tolerance for skips)
• Any Skip Ratio > 0%
• Conditional skips for external dependencies

Extraction:

# Extract skip ratio and skipped test count
grep -E "Skip Ratio|Tests \(Skipped\)" ./results/{attempt_repo}.md

Issue Template:

Title: [Test Quality] {X}% tests skip conditionally ({Y}/{Z} tests)
Label: enhancement
Body:
## Issue
{Y} of {Z} tests ({X}%) are skipped.

## Details
{description of why tests skip - e.g., Neo4j not available}

## Impact
- Skip Ratio: {X}%
- Effective Tests: {effective} (vs {total} total)
- Benchmark penalty: Exceeds 10% threshold

## Assessment
{evaluation of whether skips are acceptable or problematic}

### Acceptable Skips
- Integration tests that require external services (Neo4j, Redis, etc.)
- Conditional skips with proper `pytest.skip()` messages

### Problematic Skips
- Tests with "not yet implemented" messages
- Unconditional skips that inflate test counts
- Stub tests that never execute

## Recommendation
{suggested action or note that no action required}

---
Filed from evaluation: results/{attempt_repo}.md

Skip Ratio Thresholds:

3d-integration. Self-Contained Integration Tests (CRITICAL)

Integration tests that skip because external dependencies aren't running are NOT acceptable. Tests must manage their own data stores.

Patterns to detect:

• Tests skipping with "Neo4j not running", "database not available", etc.
• Tests requiring manual docker setup before running
• Tests that only pass in CI environments
• Conditional skips based on external service availability

Extraction:

# Check for external dependency skips
grep -r "pytest.skip.*neo4j\|pytest.skip.*database\|pytest.skip.*not running" ./reviews/{attempt_repo}/tests/

# Check for skipif based on service availability
grep -r "skipif.*neo4j\|skipif.*database\|skipif.*connection" ./reviews/{attempt_repo}/tests/

# Check for testcontainers usage (good)
grep -r "testcontainers\|Neo4jContainer" ./reviews/{attempt_repo}/tests/

# Check for pytest-docker usage (good)
grep -r "pytest-docker\|docker_compose_file" ./reviews/{attempt_repo}/

Issue Template:

Title: [Test Quality] Integration tests must be self-contained - use testcontainers
Label: bug
Body:
## Issue

Integration tests skip when external dependencies (e.g., Neo4j) are not running. This is not acceptable - tests must manage their own data stores.

## Current Behavior

Tests skip with messages like:
- "Neo4j not running"
- "Database not available"
- "Skipping integration tests"

{examples from codebase}

## Required Behavior

Integration tests MUST start their own data stores. Tests should pass on any machine with Docker installed, without manual setup.

## Recommended Solution: testcontainers

Use [testcontainers-python](https://testcontainers-python.readthedocs.io/) to automatically manage Neo4j:

```python
# conftest.py
import pytest
from testcontainers.neo4j import Neo4jContainer

@pytest.fixture(scope="session")
def neo4j_container():
    """Start Neo4j container for integration tests."""
    with Neo4jContainer("neo4j:5") as neo4j:
        yield neo4j

@pytest.fixture
def neo4j_client(neo4j_container):
    """Get client connected to test container."""
    return Neo4jClient(
        uri=neo4j_container.get_connection_url(),
        auth=("neo4j", "test")
    )

# test_integration.py
def test_create_match(neo4j_client):
    """Test creating a match in Neo4j - container starts automatically."""
    result = neo4j_client.create_match(...)
    assert result is not None

Alternative: pytest-docker

If you prefer docker-compose, use pytest-docker:

# conftest.py
@pytest.fixture(scope="session")
def docker_compose_file():
    return "docker-compose.test.yml"

@pytest.fixture(scope="session")
def neo4j_service(docker_services):
    docker_services.wait_until_responsive(
        timeout=30.0,
        pause=0.5,
        check=lambda: is_responsive()
    )

Dependencies to Add

# pyproject.toml
[project.optional-dependencies]
test = [
    "pytest",
    "testcontainers[neo4j]",
    # OR
    "pytest-docker",
]

Impact

• Current: Tests skip, providing no integration coverage
• After fix: Tests run everywhere, actually verify Neo4j integration

Filed from evaluation: results/{attempt_repo}.md

**Issue Template: Missing Integration Tests**

Title: [Test Quality] Missing integration tests - no tests verify data persistence Label: bug Body:

Issue

No integration tests exist to verify the data persistence layer works correctly.

Current State

• Only unit tests with mocks exist
• No tests verify actual database operations
• Data layer is untested in real conditions

Required

Integration tests MUST exist and MUST:

1. Use persistent storage (testcontainers, pytest-docker, or file-based)
2. Test actual CRUD operations against real storage
3. Run without manual setup (self-contained)

Recommended Solution

Add integration tests using testcontainers:

# tests/integration/conftest.py
import pytest
from testcontainers.neo4j import Neo4jContainer

@pytest.fixture(scope="session")
def neo4j_container():
    with Neo4jContainer("neo4j:5") as neo4j:
        yield neo4j

# tests/integration/test_database.py
def test_create_and_retrieve_match(neo4j_container):
    client = Neo4jClient(uri=neo4j_container.get_connection_url())
    match = client.create_match(...)
    retrieved = client.get_match(match.id)
    assert retrieved == match

Impact

• Current: Data layer has zero test coverage
• After fix: Confidence that database operations work correctly

Filed from evaluation: results/{attempt_repo}.md

**Issue Template: In-Memory Mock Not Acceptable**

Title: [Test Quality] Integration tests use in-memory mock instead of persistent storage Label: bug Body:

Issue

Integration tests use in-memory mock storage instead of persistent storage.

Current Behavior

Tests use mock classes like MockNeo4jDatabase or FakeStorage that store data in memory dictionaries.

Why This Fails

In-memory mocks:

• Don't test actual persistence behavior
• Don't catch serialization/deserialization bugs
• Don't verify database constraints
• Data doesn't survive process restart

Required Behavior

Integration tests must use persistent storage that actually tests the data persistence layer.

Recommended Solution

Use testcontainers with real database:

from testcontainers.neo4j import Neo4jContainer

@pytest.fixture(scope="session")
def neo4j_container():
    with Neo4jContainer("neo4j:5") as neo4j:
        yield neo4j

Or use file-based persistent storage (e.g., SQLite):

@pytest.fixture
def db_path(tmp_path):
    return tmp_path / "test.db"

Filed from evaluation: results/{attempt_repo}.md

**Integration Test Quality Assessment:**

| Pattern | Quality | Action |
|---------|---------|--------|
| testcontainers (auto-managed) | ✓ Best | No issue needed |
| pytest-docker (compose-based) | ✓ Good | No issue needed |
| conftest starts docker manually | ✓ Acceptable | No issue needed |
| In-memory mock (not persistent) | ✗ Not acceptable | File `bug` issue - storage must be persistent |
| Skips if service not running | ✗ Not acceptable | File `bug` issue |
| Requires manual docker setup | ✗ Not acceptable | File `bug` issue |
| CI-only integration tests | ✗ Not acceptable | File `bug` issue |
| No integration tests at all | ✗ Not acceptable | File `bug` issue |

#### 3e. Test Best Practices
Check if tests follow pytest-bdd best practices with Gherkin `.feature` files:

**Patterns to detect:**
- Missing `.feature` files
- Docstring-only BDD (not executable Gherkin)
- Custom BDD helper classes instead of pytest-bdd
- E2E-only tests without unit test coverage
- **Async test functions** (pytest-bdd does not support async)

**Extraction:**
```bash
# Check for pytest-bdd usage
grep -E "pytest-bdd|\.feature|Given.*When.*Then" ./results/{attempt_repo}.md

# Check test approach in comparison analysis
grep -A 5 "Test.*Framework\|BDD Style" ./results/{attempt_repo}.md

# Check for async test functions (problematic with pytest-bdd)
grep -r "async def test_\|async def given_\|async def when_\|async def then_" ./reviews/{attempt_repo}/tests/

Issue Template:

Title: [Test Quality] Use pytest-bdd with .feature files instead of {current_approach}
Label: enhancement
Body:
## Issue
{description of current test approach}

## Current Approach
{code example showing current pattern}

## Best Practice
Use pytest-bdd with proper Gherkin `.feature` files:

```gherkin
# features/example.feature
Feature: Example Feature
  Scenario: Example scenario
    Given precondition
    When action
    Then expected result

IMPORTANT: Use synchronous test functions only.

pytest-bdd does NOT support async step definitions or test functions. Async tests will silently fail or produce unexpected behavior.

# WRONG - async tests fail with pytest-bdd
@given("a database connection")
async def given_db():  # ❌ Will not work correctly
    return await get_connection()

# CORRECT - use sync functions
@given("a database connection")
def given_db():  # ✓ Works correctly
    return get_connection_sync()

If your MCP server uses async code, create synchronous test wrappers:

import asyncio

def run_async(coro):
    """Helper to run async code in sync tests."""
    return asyncio.get_event_loop().run_until_complete(coro)

@given("a player search result")
def given_player_search(context):
    # Wrap async call in sync function
    context["result"] = run_async(search_player("Neymar"))

Benefits of pytest-bdd

1. Readable scenarios for non-technical stakeholders
2. Reusable step definitions
3. Standard tooling (IDE support, linting)
4. Separation of concerns

Recommendation

{specific migration guidance}

Filed from evaluation: results/{attempt_repo}.md

**Test Pattern Quality:**
| Pattern | Quality | Action |
|---------|---------|--------|
| pytest-bdd + .feature files (sync) | Best | No issue needed |
| pytest-bdd without .feature (sync) | Good | Optional improvement |
| pytest-bdd with async functions | Broken | File bug issue - tests silently fail |
| Docstring BDD | Acceptable | File enhancement issue |
| Custom BDD helper | Non-standard | File enhancement issue |
| E2E only / No BDD | Poor | File enhancement issue |

**Critical: Async pytest-bdd Warning**

pytest-bdd step definitions and scenarios MUST be synchronous. Async functions will:
- Silently skip or fail tests
- Return coroutine objects instead of results
- Cause confusing "test passed" reports when tests didn't actually run

Always use synchronous wrappers around async code in BDD tests.

---

#### 3e-swift. Swift/iOS Test Best Practices

Check if tests follow Swift/iOS testing best practices:

**Patterns to detect:**
- Missing test targets in Xcode project
- No XCTest usage
- Empty test methods (stubs)
- Disabled tests (renamed without `test` prefix)
- Missing async test handling
- No mock/dependency injection

**Extraction:**
```bash
# Check for XCTest usage
grep -r "import XCTest\|XCTestCase" ./reviews/{attempt_repo}/ --include="*.swift"

# Check for test methods
grep -r "func test" ./reviews/{attempt_repo}/ --include="*.swift" | wc -l

# Check for disabled/stub tests
grep -r "func disabled_\|// func test\|func _test" ./reviews/{attempt_repo}/ --include="*.swift"

# Check for proper async testing
grep -r "async throws\|expectation\|wait(for:" ./reviews/{attempt_repo}/ --include="*.swift"

# Check for Quick/Nimble BDD framework
grep -r "import Quick\|import Nimble\|describe.*context.*it" ./reviews/{attempt_repo}/ --include="*.swift"

Issue Template (Swift):

Title: [Test Quality] Use XCTest with proper async handling instead of {current_approach}
Label: enhancement
Body:
## Issue
{description of current test approach}

## Current Approach
{code example showing current pattern}

## Best Practice
Use XCTest with proper structure:

```swift
import XCTest
@testable import YourApp

final class FeatureTests: XCTestCase {
    var sut: FeatureViewModel!
    var mockService: MockNetworkService!

    override func setUp() {
        super.setUp()
        mockService = MockNetworkService()
        sut = FeatureViewModel(service: mockService)
    }

    override func tearDown() {
        sut = nil
        mockService = nil
        super.tearDown()
    }

    func test_whenLoadingData_thenUpdatesState() async throws {
        // Given
        mockService.mockResult = TestData.sampleItems

        // When
        await sut.loadData()

        // Then
        XCTAssertEqual(sut.items.count, 3)
        XCTAssertFalse(sut.isLoading)
    }
}

For BDD-style tests, use Quick/Nimble:

import Quick
import Nimble
@testable import YourApp

class FeatureSpec: QuickSpec {
    override class func spec() {
        describe("Feature") {
            var sut: FeatureViewModel!

            beforeEach {
                sut = FeatureViewModel()
            }

            context("when loading data") {
                it("updates the state") {
                    waitUntil { done in
                        sut.loadData {
                            expect(sut.items).toNot(beEmpty())
                            done()
                        }
                    }
                }
            }
        }
    }
}

Recommendation

{specific migration guidance}

Filed from evaluation: results/{attempt_repo}.md

**Swift Test Pattern Quality:**

| Pattern | Quality | Action |
|---------|---------|--------|
| XCTest + async/await + DI | Best | No issue needed |
| XCTest + expectations + DI | Good | No issue needed |
| Quick/Nimble BDD | Good | No issue needed |
| XCTest without DI | Acceptable | Optional improvement |
| XCTest with stubs/disabled | Poor | File enhancement issue |
| No tests | Critical | File bug issue |

**Swift-Specific Test Issues:**

| Issue | Detection | Severity |
|-------|-----------|----------|
| Tests don't compile | `swift build` fails | Critical |
| Tests timeout on CI | Long `wait(for:)` calls | High |
| No async handling | Missing `async` or `expectation` | Medium |
| Force unwrapping in tests | `!` without guard | Low |
| Missing mocks | No protocol-based DI | Medium |

#### 3f. Documentation Quality
Check if README.md contains essential user documentation:

**Required Elements:**
1. **Setup Instructions**: Prerequisites, installation, configuration
2. **MCP Server Setup**: How to start server, how to connect Claude
3. **Example Q&A**: Sample questions and expected responses

**Extraction:**
```bash
# Check README in cloned repo
head -100 ./reviews/{attempt_repo}/README.md

# Look for key sections
grep -E "Quick Start|Installation|Setup|MCP|Example|Usage" ./reviews/{attempt_repo}/README.md

Issue Template:

Title: [Docs] README missing setup instructions and usage examples
Label: documentation
Body:
## Issue
The README.md lacks essential user documentation.

## Missing Documentation

### 1. Setup Instructions
- Prerequisites (Python version, Neo4j)
- Installation steps
- Environment configuration

### 2. MCP Server Setup
- How to start the MCP server
- How to configure Claude to use the server
- Example configuration

### 3. Usage Examples with Q&A
- Example questions users can ask
- Expected responses/output

## Best Practice Example
See `2025-10-30-python-hive` for comprehensive documentation.

---
Filed from evaluation: results/{attempt_repo}.md

Documentation Quality Levels:

3g. Architecture/Quality Issues

Look for concerns in "Architecture Summary", "Weaknesses", or "Areas of Note":

Patterns:

• Bullet points under "Weaknesses" heading
• Notes about missing MCP endpoints
• Performance concerns
• Code structure issues

Issue Template:

Title: [Quality] {description}
Label: enhancement
Body:
## Issue
{description}

## Context
{relevant context from the evaluation}

## Recommendation
{suggested improvement}

---
Filed from evaluation: results/{attempt_repo}.md

3h. Context Header Blocks

Check if source code files have proper context header comment blocks.

Patterns to detect:

• Files missing header blocks entirely
• Headers missing required sections (Purpose, Interfaces, Change History)
• Headers not updated with change history

Extraction:

# Check for context headers in source files
for f in $(find ./reviews/{attempt_repo}/src -name "*.py" -not -name "__init__.py"); do
  if ! head -30 "$f" | grep -q "CONTEXT\|Purpose\|Module:"; then
    echo "Missing header: $f"
  fi
done

Issue Template:

Title: [Quality] Source files missing context header blocks
Label: enhancement
Body:
## Issue

Source code files are missing required context header comment blocks.

## Required Header Format

Every source file must have a header block that includes:
1. **Purpose** - What the file/module does
2. **Interfaces** - Key classes, functions, or APIs exposed
3. **Change History** - Record of modifications (updated on every change)

## Example Header (Python)

```python
"""
================================================================================
CONTEXT BLOCK
================================================================================
File: server.py
Module: brazilian_soccer_mcp.server
Purpose: MCP server implementation for Brazilian soccer knowledge graph

Description:
    Exposes Neo4j queries as MCP tools for Claude to use.

Interfaces:
    - create_server(): Creates and configures the MCP server
    - MCPServer: Main server class with tool handlers

Dependencies:
    - mcp: MCP protocol implementation
    - neo4j: Database connection

Change History:
    - 2025-01-04: Add get_player_stats tool
    - 2025-01-03: Initial creation
================================================================================
"""

Files Missing Headers

{list of files without proper headers}

Impact

• Without context headers, code is harder to understand and maintain
• Change history helps track evolution of the codebase
• Interfaces section provides quick API reference

Filed from evaluation: results/{attempt_repo}.md

**Context Header Assessment:**

| Coverage | Assessment | Action |
|----------|------------|--------|
| 100% | Excellent | No issue needed |
| 75-99% | Good | Optional issue |
| 50-74% | Partial | File enhancement issue |
| <50% | Poor | File enhancement issue |

### 4. Create Issues
File each extracted shortcoming as a separate GitHub issue.

**Constraints:**
- You MUST create one issue per shortcoming (not a combined issue)
- You MUST apply appropriate default labels
- You MUST include reference to the evaluation report
- You MUST check for duplicates before creating (match on title)
- You MUST create detail issues BEFORE the summary issue (so you can reference issue numbers)
- You SHOULD group related issues with a common prefix
- You MAY skip issues if identical ones already exist

**Issue Creation Order:**
1. Create all detail issues first ([Missing], [Test Quality], [Quality], etc.)
2. Note the issue numbers assigned
3. Create the [Compliance] summary issue last, referencing the detail issues

```bash
# Check if issue already exists
gh issue list -R brazil-bench/{attempt_repo} --search "{issue_title}" --json title

# Create issue using HEREDOC for multi-line markdown body
gh issue create -R brazil-bench/{attempt_repo} \
  --title "[Missing] MCP server implementation" \
  --label "enhancement" \
  --body "$(cat <<'EOF'
## Requirement

The spec requires an MCP server with tools exposed as endpoints.

## Current State

Query engine exists but not exposed as MCP tools.

## Suggested Fix

Add MCP server wrapper using `fastmcp` or `mcp` package.

---
Filed from evaluation: [results/{attempt_repo}.md](https://github.com/brazil-bench/pourpoise/blob/main/results/{attempt_repo}.md)
EOF
)"

Cross-Referencing Issues:

When creating the summary [Compliance] issue, reference related detail issues:

## Missing Requirements

1. **MCP server implementation** - see #2
2. **Query performance benchmarks** - see #3
3. **Cross-file queries verification** - see #4

## Additional Issues

- **Test Quality:** 84% skip ratio - see #1

Issue Creation Rules:

*Use bug for skip ratios >50%, problematic skips, or integration tests that skip due to missing dependencies; use enhancement for acceptable conditional skips.

4a. Summary Issue Template

After creating all detail issues, create a [Compliance] summary issue that links them together:

gh issue create -R brazil-bench/{attempt_repo} \
  --title "[Compliance] Spec compliance at {X}% ({Y}/16 requirements)" \
  --label "enhancement" \
  --body "$(cat <<'EOF'
## Current Status

{Y}/16 requirements met ({X}% compliance)

## Missing Requirements

1. **{requirement_1}** - see #{issue_number_1}
2. **{requirement_2}** - see #{issue_number_2}
...

## Implemented Requirements

### Category 1 (N/M)
- [x] Requirement that passed
- [x] Another passing requirement
...

## Additional Issues

- **Test Quality:** {skip_ratio}% skip ratio - see #{test_quality_issue}

## Benchmark Score

Current score: **{score}** (rank {rank} of 8)

To improve ranking, address the issues linked above.

---
Filed from evaluation: [results/{attempt_repo}.md](https://github.com/brazil-bench/pourpoise/blob/main/results/{attempt_repo}.md)
EOF
)"

This summary issue serves as an index to all other filed issues and provides context for prioritization.

5. Generate Summary

Output a summary of all issues created.

Constraints:

• You MUST list all issues created with their URLs
• You MUST note any issues skipped due to duplicates
• You MUST include the total count

Output Format:

## Issues Filed for {attempt_repo}

### Created ({N} issues)
| # | Title | URL |
|---|-------|-----|
| 1 | [Missing] MCP server implementation | {url} |
| 2 | [Test Quality] 84% tests skip unconditionally | {url} |
| ... | ... | ... |

### Skipped (duplicates)
- {title} - already exists as #{existing_issue_number}

### Summary
- Total shortcomings identified: {X}
- Issues created: {Y}
- Issues skipped (duplicates): {Z}

Examples

Example 1: Filing issues for Hive v2

# File issues from the Hive v2 evaluation
file issues for 2025-12-13-python-claude-hive

Expected issues:

1. [Missing] MCP server implementation - tools not exposed as MCP endpoints
2. [Missing] Query performance benchmarks (< 2s simple, < 5s aggregate)
3. [Missing] Cross-file queries verification
4. [Test Quality] 84% of tests skip unconditionally (53/63 tests)
5. [Compliance] Spec compliance at 81% (13/16 requirements)

Example 2: Dry run to preview issues

# Preview issues without creating them
file issues for 2025-12-13-python-claude-hive --dry-run

Troubleshooting

No evaluation report found

• Run the evaluate-attempt skill first
• Check the path: ./results/{attempt_repo}.md

Permission denied on repository

• Verify you have write access: gh repo view brazil-bench/{attempt_repo}
• Check if issues are enabled on the repo

Duplicate detection failing

• Issues are matched by exact title
• If titles differ slightly, duplicates may be created
• Use gh issue list -R repo --search "keyword" to check

Labels don't exist

• Use default GitHub labels (enhancement, bug, documentation) which exist on all repos
• Avoid custom labels as creating them requires admin access (HTTP 403 error)
• If a label doesn't exist, omit the --label flag entirely
• The title prefix (e.g., [Missing], [Test Quality]) provides categorization without labels

Too many issues

• Use --dry-run first to preview
• Consider grouping minor issues into a single "Minor improvements" issue
• Focus on high-impact shortcomings first