What's New
/autoresearch:scenario — Scenario Explorer
New subcommand that autonomously explores a seed scenario across 12 dimensions to generate situations, edge cases, failure modes, and derivative scenarios. Think brainstorming meets the autoresearch loop.
Just type:
/autoresearch:scenario
Claude asks 4-8 adaptive questions, then iterates — generating one concrete situation per iteration, classifying it (new/variant/duplicate), expanding edge cases, and logging everything.
7-Phase Autonomous Loop
Seed → Decompose → Generate → Classify → Expand → Log → Repeat
Each iteration picks an unexplored dimension/combination and generates a concrete situation with actors, triggers, flow, expected outcome, and severity.
12 Exploration Dimensions
Happy path, error path, edge case, abuse/misuse, scale, concurrent, temporal, data variation, permission, integration, recovery, state transition.
5 Domain Templates
| Domain | Priority Dimensions | Default Output |
|---|---|---|
| Software/API | error, edge case, concurrent, integration | test-scenarios |
| Product/UX | happy path, error, permission, temporal | user-stories |
| Business/Process | happy path, error, permission, recovery | use-cases |
| Security/Compliance | abuse, permission, data variation, concurrent | threat-scenarios |
| Marketing/Sales | happy path, data variation, temporal, scale | user-stories |
Adaptive Interactive Setup
Question count adapts (4-8) based on context provided:
- No input → 8 questions
- Vague scenario (≤5 words, no actor/action) → 7 questions
- Clear scenario, no domain → 5 questions
- Clear scenario + domain → 4 questions
Robustness (from dry-run testing)
- Deferred tool fetch: ToolSearch fallback for AskUserQuestion
- Vague/clear classification with word count + actor+action heuristic
- Inline context parsing rules (flag ordering, conflict resolution)
- Cancel & interruption handling
Flags
| Flag | Purpose |
|---|---|
--domain <type>
| software, product, business, security, marketing |
--depth <level>
| shallow (10), standard (25), deep (50+) |
--format <type>
| use-cases, user-stories, test-scenarios, threat-scenarios, mixed |
--focus <area>
| edge-cases, failures, security, scale |
--scope <glob>
| Limit to specific files/features |
Chaining
/autoresearch:scenario --iterations 15
/autoresearch:debug --scope src/auth/**
/autoresearch:fix --from-debug --iterations 20Documentation
- README.md — commands table, quick decision guide, dedicated section, repo structure
- EXAMPLES.md — 10 new scenario examples with chaining patterns
Files
| File | Status |
|---|---|
commands/autoresearch/scenario.md
| Created (14 lines) |
skills/autoresearch/references/scenario-workflow.md
| Created (353 lines) |
skills/autoresearch/SKILL.md
| Modified (+70 lines, v1.6.0) |
README.md
| Modified (+36 lines) |
EXAMPLES.md
| Modified (+96 lines) |
Upgrade
Plugin: /plugin update autoresearch@autoresearch
Manual: Pull latest, re-copy skills/autoresearch/ and commands/autoresearch/
Full Changelog: v1.5.0...v1.6.0