The library
Everything we index — ranked by what works, never by stars.
forSalesMarketingHRFinanceLegalOpsProductEngineeringDataProductivitySupportsetup≤ plug & play≤ + a key≤ multi-tool
● works · ● untested / no effect · ● hurts — every rank is measured against a no-skill baseline
untested★3→untested★0→untested★0→untested★2→untested★3→untested★0→untested★0→untested★2→untested★0→untested★3→untested★0→untested★0→untested★0→untested★2→untested★0→untested★249→untested★0→untested★0→untested★0→untested★1→untested★0→untested★7→untested★0→untested★36→untested★0→untested★38→untested★0→untested★0→untested★0→untested★1→untested★0→untested★1→untested★0→untested★0→untested★36→untested★75→untested★0→untested★0→untested★3→untested★1→untested★0→untested★7→untested★0→untested★0→untested★0→untested★0→untested★0→untested★0→untested★0→untested★2→
Security gate before mergeworkflowEngineeringL3
mishmar-security-gate · Gate a sensitive merge when orthogonal security expertise and adversarial refutation are required.
Audit work-order completion tracking and optimizationworkflowOpsL3
workorder-tracking-audit · Audit a compliance-critical ERP surface when dimension coverage and adversarial verification are essential.
Refactor code modules in parallel with behavior preservationworkflowEngineeringL3
sdlc-refactor-by-module · Refactor multiple modules in parallel when each has an isolated scope and a test suite.
Apply skill documentation edits across multiple filesworkflowEngineeringL2
skill-audit-remediate · Apply bulk skill remediation when decisions span multiple files and lane isolation is guaranteed.
Fuzz and validate KQL translation rulesworkflowEngineeringL3
fuzz-kql-translator · Find bugs in a KQL-to-SQL translator when seeded fuzzing families and differential testing are available.
Identify root cause from disjoint evidenceworkflowEngineeringL3
root-cause · Diagnose a failure when multiple evidence sources exist and hypothesis generation needs to avoid anchoring.
Generate verified reference monograph from researchworkflowMarketingL2
monograph · Write a long-form technical document when depth, consistency, and multi-phase writing are required.
Conduct bounded repository code reviewworkflowEngineeringL2
repo_review · Review a code change for bugs and design issues when structured findings and effort levels matter.
Harden application for production securityworkflowEngineeringL2
sganit-production-hardening · Harden a production system when systematic audit and controlled rollout are required.
Gate deployment readiness with parallel checksworkflowEngineeringL3
mishkan-release-readiness · Gate a release when multi-layer verification and structured blocker tracking are required.
Cross-model audit classification consistencyworkflowDataL2
verify-classify · Fact-check a claim when credibility and source tracing matter.
Detect page ranges in case law PDFsworkflowLegalL2
casebook-boundaries · Establish a boundary rule when precedent and edge-case analysis are critical.
Classify allergen groups across speciesworkflowDataL2
allergen-multispecies · Merge allergen data across species when conflict resolution and taxonomy consistency are needed.
Run comprehensive UI quality checksworkflowProductL2
forge-check-ui · Audit UI components when design consistency and accessibility compliance both matter.
Execute exhaustive code audit with verificationworkflowEngineeringL3
code-review-deep · Review code when multi-lens analysis (correctness + design) and configurable effort are required.
Research meeting participants in parallelworkflowProductivityL2
meeting-prep-research · Prepare for a meeting when rapid research and personalized briefing matter.
Classify allergen data with adversarial reviewworkflowDataL2
allergen-classify-misses · Multi-pass allergen data cleanup when rules are complex and precision matters over speed
Fix translation strings with glossary preservationworkflowOpsL2
translate-rework · Bulk game localization when idiom fidelity + formatting preservation are non-negotiable
Build website sections from design systemworkflowProductL2
gc-build-sections · Parallel page section development when design system is immutable + ten builders can work simultaneously
Benchmark extraction and verification stagesworkflowEngineeringL2
bench-s4-parallel-2stage · Measuring wall-clock cost of serialization barriers vs pipelined concurrency
Route heterogeneous reviewers to consensusworkflowOpsL2
stamina-dispatch · Deterministic dispatch orchestration when you need role-diverse review with a procedural verdict, not a human gate
Review changes across multiple dimensionsworkflowEngineeringL3
compass-review · High-confidence PR review when 5 concurrent reviewers + skeptic-per-finding beats single-pass
Verify allergen detection completenessworkflowDataL2
allergen-completeness · Final completeness pass over rejected rows when regex false-negatives are a known risk
Select best agent approach via tournamentworkflowProductL2
tournament · Strategy selection when solution space is open-ended and head-to-head beats rubric scoring
Review changes with adversarial verificationworkflowEngineeringL2
review-changes · Quick lightweight review when changes are small + reasoning-only suffices
Author minimal fixes for remaining bugsworkflowEngineeringL2
nvptx-author-fixes · Compiler bug fixing when root cause is documented + one-off minimal edits suffice (not redesigns)
Run deterministic code review verdicts in parallelworkflowEngineeringL2
sdlc-review-team · Deterministic multi-role PR review when you want procedural routing but human retains merge gate
Implement PRD modules with isolated ownershipworkflowProductL2
sdlc-prd-sharding · Parallel PRD analysis when sections are truly independent + worktree overhead is acceptable
Evaluate docs against cold-start personasworkflowProductL2
docs-reader-personas · Persona extraction when docs are rich + you want diverse concurrent reads with deterministic aggregation
Remove host-copy bridges from metal traitworkflowEngineeringL2
metal-trait-bridge-removal · Rust-specific refactoring when manual trait bridge removal is tedious + well-scoped
Verify rendered sections against design cropsworkflowEngineeringL2
gc-review-fidelity · QA pass after section build when pixel-perfect fidelity is required
Fix visual defects in report sectionsworkflowMarketingL2
cadence-report-repair · Report synthesis when logs are noisy + visual defects must be auto-repaired
Repair translation edge cases for localizationworkflowMarketingL2
translate-round2 · Translation refinement when round-1 coverage is ~95% but edge cases need manual TLC
Validate changes across multiple verification lanesworkflowEngineeringL2
verify-council · Verification when author needs raw multi-lane input + bears adjudication burden themselves
Loop discovery until no new findings emergeworkflowL2
loop-until-dry · Open-ended discovery when you need to loop until exhaustion + K is user-configurable
Cluster and rank customer feedback themesworkflowProductL2
customer-feedback-theme-extractor · Bulk customer feedback analysis when you want quick theme extraction + ranking by frequency
Audit sprint retro transcripts across lensesworkflowOpsL2
retro-audit-engine · Continuous monitoring of sprint health through multi-lens transcript audit when retro findings need adversarial validation.
Complete partially-written content tracksworkflowMarketingL2
finish-track · Graduate-level monographs on statistical/mathematical topics where step-by-step derivations and honest real-world examples are mandatory.
Validate infrastructure changes across dimensionsworkflowEngineeringL2
migdal-infra-change · Non-trivial infrastructure changes (network, IAM, scaling) where blast-radius analysis spans design, systems, devops, observability, and health.
Dispatch parallel agent tasks with capped concurrencyworkflowEngineeringL2
fanout · Batched parallel work where cost/quota capping is required and partial-failure detection is needed.
Extract grant facets into controlled taxonomiesworkflowFinanceL2
layer2-facet · Exploratory grant faceting where open-ended vocabularies evolve (seed values provided but new entries created per grant).
Scan Python codebases for TODO commentsworkflowEngineeringL2
todo-scanner · Quick inventory of technical debt across a Python codebase when distributed priorities and actionable top-3 recommendations are needed.
Validate and refine translation quality at scaleworkflowMarketingL2
sonnet-validate-fix · Second-tier quality gate on bulk machine-translated game localization when register, terminology, formatting preservation, and idiomatic rewrite are critical.
Review and repair language-specific code findingsworkflowEngineeringL2
systems-check-workflow · Iterative code review when language-specific inspectors and optional specialist agents (security, perf, migration) are needed and findings must be actionable.
Discover and refute bugs with majority verificationworkflowEngineeringL2
bug-hunt · Exploratory correctness/security/resource-leak testing when multiple independent finders and majority-refute skeptics must verify candidates.
Process finance theme through critical review pipelineworkflowFinanceL2
fi-theme-pipeline · Off-the-battlefield historical narrative when locked house voice, fact-pack validation, and parallel critic review are mandatory.
Migrate portfolio template to Sanity with visual parityworkflowEngineeringL3
kokimoto-migration · Version upgrades of complex config systems when automated transformation and validation must preserve backward compatibility.
Re-author failed carousel content daysworkflowMarketingL2
author-carousel-missing-days · Editorial scheduling when author carousel coverage must be contiguous and gaps auto-filled.
Rewrite paper sections in parallel for consistencyworkflowMarketingL2
paper-redraft · Batch academic/technical paper authoring when spec-driven drafting and multi-phase review are needed.
Detect CLAUDE.md audit issues across codebasesworkflowEngineeringL2
claude-md-audit-detect · Multi-repo governance when CLAUDE.md documentation standards must be enforced and drift detected.