Feature Status Skill

Purpose

This skill counts the number of features marked as @failing and writes the count to a JSON file. This is used by the autonomous coding harness to determine when the implementation loop should end (when failing_count reaches 0).

Output File

File: feature-status.json

Format:

json
1{
2  "failing_count": 3
3}

Loop Termination Logic:

• If failing_count > 0 → Continue coding sessions
• If failing_count == 0 → All features implemented, end loop

How It Works

Step 1: Find Feature Files

Use Glob to find all Gherkin feature files:

Pattern: gherkin.feature_*.feature

Step 2: Count @failing Tags

For each feature file:

1. Read the first few lines
2. Look for @failing tag
3. If found, increment the failing counter

Tag Detection:

@failing
Feature: Some Feature Name
  ...

Read lines until you find either:

• @failing → Count this feature as failing
• @passing → Skip (not failing)
• Feature: line → Stop searching (assume no tag = passing)

Step 3: Write JSON

Write feature-status.json with just the failing count:

json
1{
2  "failing_count": <count>
3}

Usage

Simply invoke the skill:

/feature-status

This will:

1. Scan all gherkin.feature_*.feature files in the current directory
2. Count how many have @failing tags
3. Write the count to feature-status.json

Example Implementation

python
1# Pseudocode for reference
2def count_failing_features(directory):
3    failing_count = 0
4
5    # Find all feature files
6    feature_files = glob("gherkin.feature_*.feature")
7
8    for file in feature_files:
9        with open(file) as f:
10            for line in f:
11                line = line.strip()
12
13                if line.startswith("@failing"):
14                    failing_count += 1
15                    break
16                elif line.startswith("@passing"):
17                    break
18                elif line.startswith("Feature:"):
19                    # No tag found, assume passing
20                    break
21
22    return failing_count

Integration with Autonomous Coding Harness

The harness can check the status file to decide whether to continue:

python
1import json
2
3def should_continue_loop():
4    with open("feature-status.json") as f:
5        status = json.load(f)
6    return status["failing_count"] > 0

Best Practices

1. Run after each coding session to update the failing count
2. Commit the status file to track progress over time
3. Check before starting a new session to avoid unnecessary runs
4. Use as a termination condition in automation scripts