Use when you have a written implementation plan to execute in a separate session with review checkpoints
npx skills add 30eggis/claude-frontend-skills --skill "spec-change"
Install specific skill from multi-skill repository
# Description
Spec modification router with comprehensive validation. Detects duplicates, conflicts, quality issues, and impact before applying changes. Use for any spec add/remove/modify requests.
# SKILL.md
name: spec-change
description: "Spec modification router with comprehensive validation. Detects duplicates, conflicts, quality issues, and impact before applying changes. Use for any spec add/remove/modify requests."
allowed-tools: Read, Write, Edit, Bash, Task, AskUserQuestion, Glob, Grep
argument-hint: "
permissionMode: bypassPermissions
spec-change: Spec Modification Router
Validates and applies spec changes with comprehensive pre-flight checks.
References
- Output Formats - Analysis results, change plan, RTM formats
- shared/output-rules.md - Agent output rules
- shared/context-rules.md - Context management
Workflow
[Request] → [Parallel Analysis] → [Plan Review] → [User Approval] → [Apply]
│
┌──────────────┼──────────────┐
│ │ │
Doppelganger Conflict Clarity/Coverage
│ │ │
└──────────────┼──────────────┘
│
Butterfly Effect
│
[Change Plan] → [Approve?] → [Apply]
Phase 0: Init
1. Context Validation
Read: .claude/settings.local.json
REQUIRE: context != "none" (minimum: fork)
2. Session Detection
IF sessionId provided: use it
ELSE: find most recent in tmp/
3. Parse Request
EXTRACT: action (add/remove/modify), target, content
Phase 1: Parallel Analysis
Batch 1 (4 parallel)
| Agent | Model | Output |
|---|---|---|
| spec-doppelganger | sonnet | _analysis/doppelganger.json |
| spec-conflict | sonnet | _analysis/conflict.json |
| spec-clarity | sonnet | _analysis/clarity.json |
| spec-consistency | haiku | _analysis/consistency.json |
Batch 2 (2 parallel)
| Agent | Model | Output |
|---|---|---|
| spec-coverage | sonnet | _analysis/coverage.json |
| spec-butterfly | opus | _analysis/butterfly.json |
Aggregate
Read: _analysis/*.json
Create: _analysis/summary.md
Phase 2: Plan Review
Task(change-planner, opus):
Input: summary.md, butterfly.json
Output: _analysis/change-plan.md
AskUserQuestion: "Apply these changes?"
Options: ["Yes, apply all", "Modify plan", "Cancel"]
Phase 3: Apply Changes
FOR each change in plan:
Edit: {target_file}
Checkpoint: meta-checkpoint.sh
Task(rtm-updater, haiku):
Output: _traceability/rtm.json
Output: "✓ Changes applied. Run /spec-change --verify to validate."
Output Structure
.spec-it/{sessionId}/plan/
├── _analysis/
│ ├── doppelganger.json, conflict.json, clarity.json
│ ├── consistency.json, coverage.json, butterfly.json
│ ├── summary.md
│ └── change-plan.md
└── _traceability/
└── rtm.json
See Output Formats for detailed structures.
Agents
| Agent | Model | Purpose |
|---|---|---|
| spec-doppelganger | sonnet | Duplicate detection |
| spec-conflict | sonnet | Contradiction detection |
| spec-clarity | sonnet | Quality assessment |
| spec-consistency | haiku | Term consistency |
| spec-coverage | sonnet | Gap analysis |
| spec-butterfly | opus | Impact analysis |
| change-planner | opus | Generate change plan |
| rtm-updater | haiku | Update traceability |
Arguments
| Arg | Required | Description |
|---|---|---|
| sessionId | No | Auto-detected if omitted |
| --target | No | Specific document path |
| --skip-checks | No | Skip validation (not recommended) |
| --verify | No | Verify previous changes |
Examples
# Add requirement
/spec-change 20260130-123456 "Add 2FA to authentication"
# Modify with target
/spec-change 20260130-123456 --target 01-chapters/CH-02.md "Change to OAuth"
# Verify changes
/spec-change --verify 20260130-123456
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.