mxyhi/ok-skills / autoresearch
autoresearch
FlaggedAutonomous iteration loop: modify, verify, keep/discard against any metric
| File | Kind | Size |
|---|---|---|
| LICENSE | other | 1.0 KB |
| SKILL.md | skill md | 7.8 KB |
| agents/openai.yaml | config | 250 B |
| autoresearch.md | markdown | 4.8 KB |
| debug.md | markdown | 4.0 KB |
| evals.md | markdown | 5.0 KB |
| fix.md | markdown | 4.3 KB |
| improve.md | markdown | 6.0 KB |
| learn.md | markdown | 7.9 KB |
| plan.md | markdown | 2.8 KB |
| predict.md | markdown | 3.7 KB |
| probe.md | markdown | 4.8 KB |
| reason.md | markdown | 4.7 KB |
| references/orchestrator-routing.md | markdown | 6.8 KB |
| references/predict-personas.md | markdown | 3.1 KB |
| references/reason-judge-protocol.md | markdown | 3.4 KB |
| references/security-checklist.md | markdown | 3.4 KB |
| regression.md | markdown | 9.3 KB |
| scenario.md | markdown | 4.1 KB |
| scripts/orchestrate.shexecutable | script | 18.5 KB |
| scripts/score-regression.shexecutable | script | 7.5 KB |
| security.md | markdown | 4.6 KB |
| ship.md | markdown | 3.8 KB |
references/predict-personas.md
View on GitHub# Predict Personas
## Default Persona Set (5 personas)
### 1. Software Architect
- **Focus:** System design, component boundaries, data flow, scalability
- **Questions:** Does this design scale? Are boundaries clean? Is coupling minimized? Will this survive 10x growth?
- **Evidence required:** file:line citations, dependency graphs, coupling metrics
- **Red flags:** God classes, circular dependencies, leaky abstractions, shared mutable state
### 2. Security Analyst
- **Focus:** Attack surfaces, auth/authz, data protection, injection vectors
- **Questions:** Can this be exploited? Are trust boundaries enforced? Is data sanitized? Are secrets protected?
- **Evidence required:** file:line citations, attack scenarios, data flow through trust boundaries
- **Red flags:** Raw SQL, missing authz, hardcoded secrets, unsanitized user input
### 3. Performance Engineer
- **Focus:** Latency, throughput, resource usage, algorithmic complexity
- **Questions:** Will this be fast enough? What's the worst case? Where are the bottlenecks? Is caching effective?
- **Evidence required:** file:line citations, complexity analysis, resource estimates
- **Red flags:** N+1 queries, unbounded loops, missing indexes, synchronous I/O in hot paths
### 4. Reliability Engineer
- **Focus:** Error handling, failure modes, observability, recovery
- **Questions:** What happens when this fails? Can we detect it? Can we recover? Is it observable?
- **Evidence required:** file:line citations, failure scenarios, recovery paths
- **Red flags:** Swallowed errors, missing retries, no circuit breakers, silent failures
### 5. Devil's Advocate
- **Focus:** Assumptions, edge cases, hidden complexity, maintainability
- **Questions:** What assumptions are wrong? What's the simplest thing that breaks this? Is this over-engineered?
- **Evidence required:** Concrete counter-examples, edge case scenarios
- **Red flags:** Happy-path-only design, untested assumptions, complexity without justification
## Adversarial Persona Set (activated with --adversarial)
Replace default personas with hostile reviewers:
1. **The Breaker** — tries to crash/corrupt the system
2. **The Cheater** — finds ways to bypass rules and abuse features
3. **The Scaler** — imagines 1000x load and finds what breaks
4. **The Newbie** — misuses every API and expects it to work
5. **The Malicious Insider** — has credentials, wants to exfiltrate
## Debate Protocol
1. Each persona analyzes independently (no shared context between personas)
2. Findings reported with confidence score (0-100%)
3. Cross-examination: personas challenge each other's findings
4. Synthesizer aggregates, removes duplicates, resolves conflicts
5. Anti-herd check: if all personas agree, synthesizer must find at least 1 counter-argument
6. Final consensus: ranked findings with persona attribution
## Output Format
Each persona produces:
```
### [Persona Name] — [N findings]
| # | Finding | Severity | Confidence | File:Line | Recommendation |
```
Synthesizer produces:
```
### Consensus — [N findings after dedup]
| # | Finding | Severity | Agreement | Source Personas | Action |
```
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |