autoresearch

Claude Autoresearch — Autonomous Goal-directed Iteration

Inspired by Karpathy's autoresearch. Applies constraint-driven autonomous iteration to ANY work — not just ML research.

Core idea: You are an autonomous agent. Modify → Verify → Keep/Discard → Repeat.

Subcommands

| Subcommand | Purpose | |------------|---------| | /autoresearch <goal> | Default path — parse free-form goal, build harness, capture baseline, loop until goal met | | /autoresearch | Run the autonomous loop (default) | | /autoresearch:plan | Interactive wizard to build Scope, Metric, Direction & Verify from a Goal | | /autoresearch:security | Autonomous security audit: STRIDE threat model + OWASP Top 10 + red-team (4 adversarial personas) |

Default Path: /autoresearch <free-form goal>

When the user invokes /autoresearch <goal> with any free-form string after the command, parse the goal into seven slots, print the parsed-slot dump back for user visibility, then run the harness protocol in references/benchmark-harness.md before entering the loop.

Goal-parsing rubric:

| Slot | Extraction rule | Fallback | |------|-----------------|----------| | metric | First measurable noun (latency, reliability, coverage, flakiness, bundle size, p95, accuracy, error-rate, LOC, build time) | Ask user (1 sentence) | | direction | reduce/lower/below/under/minimise/to 0% + cost-word → minimise; increase/raise/above/over/maximise/to 100% + quality-word → maximise | minimise for cost/time/size/error, maximise for coverage/score/throughput | | target | Number + unit in goal (500ms, 95%, 0%, <200KB) | "best achievable" — unbounded loop | | scope | Grep repo for goal's domain terms (API, test, build); propose globs | Whole repo minus node_modules, .venv, dist, target | | corpus_source | If goal names inputs (signals, queries, PRs, logs) → find source; if absent → ASK, never fabricate | corpus_required=false only when metric is purely structural (LOC, build time, bundle size) | | verify_cmd | Single shell command that prints metric: <float> on stdout — typically python benchmark.py or equivalent single-file rig | Constructed during harness build | | regression_cmd | Auto-detect: first of pytest -q, npm test, cargo test, go test ./... whose config exists | Ask user |

Worked examples:

/autoresearch reduce API p95 latency to 200ms
→ metric=p95_latency_ms, direction=minimise, target=200, scope=src/api/**,
  corpus_source=prod log tail or fixtures, verify_cmd=python benchmark.py,
  regression_cmd=pytest -q

/autoresearch reduce test flakiness to 0%
→ metric=flaky_test_rate, direction=minimise, target=0, scope=tests/**,
  corpus_source=CI run history, verify_cmd=python benchmark.py (N reruns),
  regression_cmd=pytest -q

/autoresearch increase signal-parser reliability to 99%
→ metric=reliability, direction=maximise, target=0.99, scope=src/parser/**,
  corpus_source=autoresearch/data/signals.jsonl, verify_cmd=python benchmark.py,
  regression_cmd=pytest -q

Print the parsed slot dump to the user before any action — this is the single confirmation checkpoint before the harness protocol begins.

/autoresearch:security — Autonomous Security Audit (v1.0.3)

Runs a comprehensive security audit using the autoresearch loop pattern. Generates a full STRIDE threat model, maps attack surfaces, then iteratively tests each vulnerability vector — logging findings with severity, OWASP category, and code evidence.

Load: references/security-workflow.md for full protocol.

What it does:

1. Codebase Reconnaissance — scans tech stack, dependencies, configs, API routes
2. Asset Identification — catalogs data stores, auth systems, external services, user inputs
3. Trust Boundary Mapping — browser↔server, public↔authenticated, user↔admin, CI/CD↔prod
4. STRIDE Threat Model — Spoofing, Tampering, Repudiation, Info Disclosure, DoS, Elevation of Privilege
5. Attack Surface Map — entry points, data flows, abuse paths
6. Autonomous Loop — iteratively tests each vector, validates with code evidence, logs findings
7. Final Report — severity-ranked findings with mitigations, coverage matrix, iteration log

Key behaviors:

• Follows red-team adversarial mindset (Security Adversary, Supply Chain, Insider Threat, Infra Attacker)
• Every finding requires code evidence (file:line + attack scenario) — no theoretical fluff
• Tracks OWASP Top 10 + STRIDE coverage, prints coverage summary every 5 iterations
• Composite metric: (owasp_tested/10)*50 + (stride_tested/6)*30 + min(findings, 20) — higher is better
• Creates security/{YYMMDD}-{HHMM}-{audit-slug}/ folder with structured reports:

overview.md, threat-model.md, attack-surface-map.md, findings.md, owasp-coverage.md, dependency-audit.md, recommendations.md, security-audit-results.tsv

Flags:

| Flag | Purpose | |------|---------| | --diff | Delta mode — only audit files changed since last audit | | --fix | After audit, auto-fix confirmed Critical/High findings using autoresearch loop | | --fail-on {severity} | Exit non-zero if findings meet threshold (for CI/CD gating) |

Usage:

# Unlimited — keep finding vulnerabilities until interrupted
/autoresearch:security

# Bounded — exactly 10 security sweep iterations
/loop 10 /autoresearch:security

# With focused scope
/autoresearch:security
Scope: src/api/**/*.ts, src/middleware/**/*.ts
Focus: authentication and authorization flows

# Delta mode — only audit changed files since last audit
/autoresearch:security --diff

# Auto-fix confirmed Critical/High findings after audit
/loop 15 /autoresearch:security --fix

# CI/CD gate — fail pipeline if any Critical findings
/loop 10 /autoresearch:security --fail-on critical

# Combined — delta audit + fix + gate
/loop 15 /autoresearch:security --diff --fix --fail-on critical

Inspired by:

• Strix — AI-powered security testing with proof-of-concept validation
• /plan red-team — adversarial review with hostile reviewer personas
• OWASP Top 10 (2021) — industry-standard vulnerability taxonomy
• STRIDE — Microsoft's threat modeling framework

/autoresearch:plan — Goal → Configuration Wizard

Converts a plain-language goal into a validated, ready-to-execute autoresearch configuration.

Load: references/plan-workflow.md for full protocol.

Quick summary:

1. Capture Goal — ask what the user wants to improve (or accept inline text)
2. Analyze Context — scan codebase for tooling, test runners, build scripts
3. Define Scope — suggest file globs, validate they resolve to real files
4. Define Metric — suggest mechanical metrics, validate they output a number
5. Define Direction — higher or lower is better
6. Define Verify — construct the shell command, dry-run it, confirm it works
7. Confirm & Launch — present the complete config, offer to launch immediately

Critical gates:

• Metric MUST be mechanical (outputs a parseable number, not subjective)
• Verify command MUST pass a dry run on the current codebase before accepting
• Scope MUST resolve to ≥1 file

Usage:

/autoresearch:plan
Goal: Make the API respond faster

/autoresearch:plan Increase test coverage to 95%

/autoresearch:plan Reduce bundle size below 200KB

After the wizard completes, the user gets a ready-to-paste /autoresearch invocation — or can launch it directly.

When to Activate

• User invokes /autoresearch <goal-string> (anything after the command) → parse with Default Path rubric, then build harness per references/benchmark-harness.md
• User types /autoresearch with no argument → ask for a one-sentence goal OR suggest /autoresearch:plan
• User invokes /autoresearch or /ug:autoresearch → run the loop
• User invokes /autoresearch:plan → run the planning wizard
• User invokes `/autoresearch:secur