v1.3.0 — Autonomous Debug & Fix: Scientific Method Meets Autoresearch Loop

What's New

/autoresearch:debug — Autonomous Bug Hunter

Scientific method meets autoresearch loop. Doesn't stop at one bug — iteratively hunts ALL bugs using falsifiable hypotheses and evidence-based investigation.

# Hunt all bugs
/loop 20 /autoresearch:debug

# Debug specific error
/autoresearch:debug
Symptom: API returns 500 on POST /users

# Debug then auto-fix everything found
/autoresearch:debug --fix

What makes it different from regular debugging:

Feature	Traditional Debug	/autoresearch:debug
Scope	Find one bug	Hunt ALL bugs iteratively
Method	Ad-hoc	Scientific method (hypothesize → test → prove/disprove)
Memory	Start fresh each time	Results log tracks what was tried
Rollback	Manual	Automatic (disproven hypotheses reverted)
Coverage	Unknown	Composite metric tracks files investigated

Built-in knowledge:

• 7 investigation techniques (binary search, differential, minimal reproduction, trace, pattern search, working backwards, rubber duck)
• 4 cognitive bias guards (confirmation, anchoring, sunk cost, availability)
• 15 common bug patterns across 7 languages (JS, TS, Python, Go, Rust, Java, SQL)
• 5 domain-specific checklists (API, database, auth, async/concurrency, network)
• The 5 Whys root cause drill-down
• "What NOT to Do" anti-patterns table

/autoresearch:fix — Autonomous Error Crusher

Takes a broken state and iteratively repairs it until zero errors remain. ONE fix per iteration. Atomic, committed, verified, auto-reverted on failure.

# Fix everything
/autoresearch:fix

# Fix with guard (no regressions)
/autoresearch:fix
Guard: npm test

# Fix from debug findings
/loop 30 /autoresearch:fix --from-debug

What makes it different:

Feature	Traditional Fix	/autoresearch:fix
Scope	Fix one thing	Fix ALL errors until zero
Priority	Manual	Auto-prioritized (blockers → errors → warnings)
Safety	Hope nothing breaks	Guard command prevents regressions
Atomicity	Sometimes	Always (one fix per iteration)
Rollback	Manual undo	Automatic revert on failure
Completion	Guess	Stops automatically at zero errors

Built-in knowledge:

• Fix strategies for 5 languages (TypeScript, Python, Go, Rust, Java)
• 9-row anti-patterns table (never suppress, never delete tests, never use any)
• Compound fix detection (fixing one reveals another)
• Impact assessment with blast radius analysis
• 5-step rollback protocol
• Parallel fix detection for independent errors
• Escalation path after 3 failed attempts
• Domain-specific patterns: dependency fixes, DB migrations, CI/CD pipelines

Meta: Built by Autoresearch

Both protocols were refined through 35 autoresearch loop iterations across 3 parallel agents:

Score: 90/100 → 143/150 (rubric expanded mid-loop)
Iterations: 35 total (13 debug + 19 fix + 2 manual + 1 scoring)
Protocol size: 490 → 1,092 lines (+123%)

The autoresearch loop was used to improve its own debugging and fixing protocols.

All Commands (v1.3.0)

Command	Purpose	Since
/autoresearch	Autonomous iteration loop	v1.0.0
/autoresearch:plan	Goal → config wizard	v1.0.2
/autoresearch:security	STRIDE + OWASP audit	v1.0.3
/autoresearch:ship	Universal shipping workflow	v1.1.0
/autoresearch:debug	Autonomous bug hunting	v1.3.0
/autoresearch:fix	Autonomous error fixing	v1.3.0
Guard: <command>	Regression prevention	v1.0.4

Install

/plugin install autoresearch@autoresearch

Or manual:

cp -r autoresearch/skills/autoresearch .claude/skills/autoresearch
cp -r autoresearch/commands/autoresearch .claude/commands/autoresearch

Full Changelog

v1.2.0...v1.3.0