/breaking-impact - Breaking Change Analysis

Analyze uncommitted or staged changes for breaking impact on currently deployed production code. Categorizes changes by risk level and identifies API surface violations, schema migrations, config changes, and behavioral shifts.

System Prompt

EXECUTION DIRECTIVE: When the user invokes this command, you MUST:

1. IMMEDIATELY execute - no questions, no explanations first
2. Load the skill at skills/breaking-impact-review/SKILL.md
3. Analyze changes using git diff (staged + unstaged)
4. Categorize each change by risk level (BREAKING, RISKY, MIGRATION-REQUIRED, SAFE, UNKNOWN)
5. Output report in the format defined by the skill

DO NOT:

• Ask for confirmation unless requires_confirmation: true in frontmatter
• Skip execution even if it seems redundant - run it anyway

The user invoking the command IS the confirmation.

---

Usage

# Analyze current uncommitted changes
/breaking-impact

# Analyze specific commit range
/breaking-impact --range HEAD~3..HEAD

# Analyze specific files
/breaking-impact --files src/api/ src/models/

# Output as JSON
/breaking-impact --json

# Compare against specific deployed version
/breaking-impact --baseline v2.4.1

# Show only breaking/risky (skip SAFE)
/breaking-impact --risky-only

# Include migration plan for MIGRATION-REQUIRED changes
/breaking-impact --with-migration-plan

Execution Steps

1. Gather changes - Run git diff and git diff --staged to collect all modifications
2. Load skill - Read skills/breaking-impact-review/SKILL.md for analysis framework
3. Categorize changes by impact area:
   • API surface (endpoints, request/response schemas, status codes)
   • Data layer (migrations, schema changes, data format)
   • Configuration (env vars, feature flags, settings)
   • Dependencies (package versions, removed packages)
   • Infrastructure (ports, volumes, resource limits)
   • Behavioral (logic changes, error handling, timing)
4. Assign risk levels - BREAKING, RISKY, MIGRATION-REQUIRED, SAFE, UNKNOWN
5. Generate report with per-change verdicts and overall deployment recommendation

Options

Option	Description
--range <ref>	Git range to analyze (default: uncommitted)
--files <paths>	Limit to specific files/directories
--json	JSON output format
--baseline <tag>	Compare against specific deployed version
--risky-only	Show only BREAKING/RISKY/MIGRATION-REQUIRED
--with-migration-plan	Include migration steps for MIGRATION-REQUIRED
--help	Show usage information

Related Commands

Command	Relationship
/release-gate	Aggregates breaking-impact as one of its blocking gates
/api-compat	Deeper API compatibility analysis
/smoke-test	Post-deploy verification after changes ship

Success Output

COMMAND COMPLETE: /breaking-impact
Changes Analyzed: 14 files, 342 lines
Risk Summary: 1 BREAKING, 2 RISKY, 1 MIGRATION-REQUIRED, 10 SAFE
Recommendation: DO NOT DEPLOY without migration plan

Completion Checklist

• Changes collected (staged + unstaged)
• Each change categorized by impact area
• Risk level assigned per change
• Overall deployment recommendation provided
• Migration plan included (if MIGRATION-REQUIRED found)

Failure Indicators

• No changes to analyze (clean working tree)
• Unable to determine baseline version
• Unknown file types that cannot be categorized

When NOT to Use

• Documentation-only changes (use /classify instead)
• Infrastructure-only changes (use canary analysis)
• No production deployment exists yet

Anti-Patterns

Anti-Pattern	Problem	Solution
Skip before deploy	Breaking changes reach production	Run before every deployment
Ignore RISKY items	Latent issues accumulate	Review all RISKY with team
Override BREAKING	Outage risk	Fix or add migration first

Principles

• #3 Complete Execution - Full analysis of all changes
• #9 Based on Facts - Evidence from actual code diff
• #8 No Assumptions - Every change explicitly categorized

Full Standard: CODITECT-STANDARD-AUTOMATION.md