$ whoami wan-huiyan — Creator
@wan-huiyan
agent-review-panel · v1.0.0
Claude Code skill: Multi-agent adversarial review panel — 4-6 AI reviewers debate your code/plans, then a supreme judge delivers the verdict. 9 auto-detected signal groups, built-in domain checklists, anti-groupthink mechanisms.
subagent-driven-development · v1.0.0
Claude Code skill: execute implementation plans with fresh subagents per task, two-stage review gates, and parallel mode via worktree isolation. Based on obra/superpowers, parallel mode adapted from theradengai/spawn-tasks.
overnight-workflows · v1.0.0
Two sister Claude Code plugins for running autonomous overnight work sessions that land a polished deliverable on your desk by morning, with multi-agent review panels baked in to catch factual errors before they reach the client.
context-baton · v1.0.0
Two Claude Code skills for passing context across boundaries: session-handoff (end-of-session, agent→human) + successor-handoff (mid-run, agent→agent). Install via Claude Code plugin marketplace.
causal-impact-campaign · v1.0.0
Claude Code skill for measuring causal impact of marketing campaigns using Bayesian structural time series
google-trends-browser-fetch · v1.0.0
Fetch Google Trends data via browser automation, with multi-chunk daily-resolution stitching for date ranges longer than 90 days.
cloud-run-batch-experiment · v1.0.0
Running 50+ experiment variants locally ties up your machine for hours. This skill deploys them to GCP Cloud Run Jobs in parallel — for pennies — and collects results to GCS automatically.
dbt-model-planner · v1.0.0
Jumping straight from a dashboard brief to dbt SQL skips the hard part: mapping every target field to a source table, validating it exists in production, and agreeing on the logic before writing a line of code. This skill guides you through
external-data-scout · v1.0.0
   
publish-skill · v1.0.0
Publishing a Claude Code skill involves a dozen error-prone steps — repo structure, README, screenshots, marketplace submission, and client data scanning. This skill handles the full workflow so you only have to do each mistake once.
claude-statusline · v1.0.0
Build and customize the Claude Code status line (bottom info bar) and session name banner.
claude-client-proposal-slide · v1.0.0
A Claude Code skill that turns technical analysis into stakeholder-ready HTML presentations — with business activation framing so your audience knows exactly what to do next.
skill-anonymizer · v1.0.0
Scan Claude Code skills for client-specific data and anonymize them for safe public sharing. Handles current file content and git history cleaning via git-filter-repo.
plan-review-integrator · v1.0.0
After a review panel generates 14 findings with mixed severity, someone has to cross-reference each one against the plan, classify what's a real blocker versus pre-existing debt, and edit the document without breaking coherence. This skill
test-effectiveness-auditor · v1.0.0
 
api-billing-toggle · v1.0.0
Toggle Claude Code between Anthropic API-key billing (Console) and Pro/Max subscription — without ever putting your API key in plaintext.
skill-sync · v1.0.0
After improving skills locally across multiple sessions, pushing changes back to GitHub means cloning, diffing, and copying files for every skill. This skill automates that for your entire published portfolio at once.
gcp-dataform-rest-api-deploy · v1.0.0
Deploying Dataform .sqlx files via REST API requires base64-encoding, undocumented field names, and a polling loop — six places to get it wrong. This skill handles the full lifecycle without the Dataform CLI.
memory-hygiene · v1.0.0
Audit and optimize Claude Code's persistent memory system — axioms, phase templates, MEMORY.md, lessons, memory files, and ADRs — using a research-backed tiered architecture with promotion/demotion lifecycle and agency-aware staleness detec
ml-training-window-assessor · v1.0.0
A Claude Code skill that answers "can we train on more data?" with a structured, drift-aware diagnostic — before you extend your training window and silently degrade your model.
ml-feature-evaluator · v1.0.0
A Claude Code skill that answers "should we add this feature to the model?" with a structured, evidence-backed diagnostic — before you spend a week integrating something that adds no predictive value.
git-squash · v1.0.0
Noisy commit history buries meaningful changes and makes PRs hard to review. This skill auto-detects trivial commits — using message patterns, diff size, and timing heuristics — and squashes them with a single command.
claude-ecosystem-hygiene · v1.0.0
Five complementary skills across two tracks. The Claude-ecosystem track (4 plugins) audits, measures, cleans, and keeps consistent your ~/.claude/ stack: identify what's HOT vs DORMANT, measure whether the HOT artifacts actually improve tas
client-signal-triage · v1.0.0
When a client sends 15 CRM fields to evaluate for a model, manually checking each one for coverage, temporal safety, and composite potential takes hours. This skill does it systematically in minutes — GO/DEFER/NO-GO with evidence.
batch-experiment-cost-optimizer · v1.0.0
Minimize cloud batch experiment costs by checking for cached results before submitting new tasks. Works with any cloud batch system (GCP Cloud Run Jobs, AWS Batch, Azure Container Instances) that writes results to object storage.
data-provenance-verifier · v1.0.0
Catch fabricated data files before they contaminate your analysis. A Claude Code skill that audits external data files (CSV, JSON, Excel) for authenticity by checking provenance documentation, spot-checking against real APIs, and flagging s
permutation-validation · v2.0.0
|
session-handoff · v1.0.0
> 📦 Also available in the context-baton marketplace, paired with successor-handoff for mid-run agent-to-agent handoffs during long autonomous work. This standalone repo tracks the same SKILL.md as the bundle.
Are you wan-huiyan? Claim this profile.
