IBM Bob 2.0 Hackathon · September 2026

ShipSafe

One number. One decision. Zero Friday fear.
38
🔴 DO NOT SHIP
→
92
🟢 SAFE TO SHIP
+54 pts
in under 2 minutes
View Live Report ↓ View on GitHub
How it works

Scan → Score → Fix

Three steps. One command. Complete release confidence.
Step 1
🔍
Scan
IBM Bob Captain spawns 3 specialist AI agents in parallel. Security scans for OWASP Top 10. Quality checks complexity and dead code. Tests measure coverage and critical paths.
Step 2
🎯
Score
Captain collects all 3 reports and applies the weighted formula: Security 40% + Quality 35% + Tests 25%. Delivers a single Release Readiness Score with a RED or GREEN verdict.
Step 3
🔧
Fix
Captain auto-fixes safe issues: hardcoded secrets, JWT expiry, error leakage, .gitignore. Generates 15 critical security tests. Lists manual items with estimated fix times.
Architecture

4 IBM Bob Custom Modes

Each agent has a dedicated role, scoped tool permissions, and reference skill files.
🎖️
Captain Mode
Orchestrates all 3 agents as parallel subagents. Calculates score. Applies auto-fixes. Generates HTML report.
read · edit · execute · mcp · skill · subagent
🔒
Security Agent
OWASP Top 10 scanner. Finds SQL injection, path traversal, hardcoded secrets, broken auth, IDOR.
Weight: 40%
🔍
Quality Agent
Measures cyclomatic complexity, detects dead code, checks error handling completeness.
Weight: 35%
🧪
Test Agent
Runs test suite, measures coverage, checks for critical security rejection tests.
Weight: 25%
Live Demo

Release Report — AFTER ShipSafe

This is the actual HTML artifact generated by IBM Bob's Captain Mode. Score: 92 🟢 SAFE TO SHIP.
ShipSafe Release Report — generated by IBM Bob 2.0 Captain Mode
Transparent Scoring

The Formula

Not a black box. Every point is explained. Every deduction is shown.
Release Score = (Security × 0.40) + (Quality × 0.35) + (Tests × 0.25)
BEFORE: (20 × 0.40) + (55 × 0.35) + (8 × 0.25) = 38    →    AFTER: (76 × 0.40) + (85 × 0.35) + (80 × 0.25) = 92
score ≥ 80 → 🟢 SAFE TO SHIP
score 70–79 → 🟡 SHIP WITH CAUTION
score < 70 → 🔴 DO NOT SHIP
Why ShipSafe

No tool does all of this

SonarQube gives lists. Snyk gives lists. ShipSafe gives a decision.
Capability SonarQube Snyk ShipSafe
Single pass/fail decision✗✗✓
Auto-fixes code automatically✗✗✓
Generates missing security tests✗✗✓
Multi-agent AI team✗✗✓
Transparent scoring formula✗✗✓
Runs inside your IDE✗✗✓
IBM Bob 2.0

Every advanced feature — working together

🧩
4 Custom Modes
Captain + 3 specialists in custom_modes.yaml with scoped permissions.
📚
4 Skills + 12 Files
OWASP checklist, complexity thresholds, critical paths, scoring formula.
🤖
Parallel Subagents
Captain spawns 3 agents simultaneously in isolated contexts.
🔌
Custom MCP Server
5 tools: run_tests, scan_secrets, check_complexity, audit_deps, list_source_files.
✏️
File Auto-Fix
Captain writes fixes directly to source files using apply_diff and execute_command.
📊
HTML Artifact
create_html_artifact generates a shareable release report inside Bob's chat.