Aryan's productivity toolkit — skills and plugins for AI coding agents
npx claudepluginhub aryanbhats/ab-stackAutomatic Todoist task tracking at session boundaries. Maps git repos to Todoist sections, shows due tasks on session start, offers completion on stop.
A productivity toolkit for AI coding agents -- skills and plugins.
Instead of asking one AI for advice, convene a board of directors, investment committee, code review panel, or research council — each member genuinely independent, each bringing a distinct evaluation lens, each forced to give one specific actionable suggestion. Then a judge synthesizes the decision with all arguments anonymized.
| Component | Type | What it does |
|---|---|---|
/council | Skill | Multi-persona deliberation with anonymized judging |
todoist-sync | Plugin | Automatic Todoist task tracking at session boundaries |
/council — Multi-Persona Deliberation/council finance "Should I buy NVDA at current levels?"
/council code # reviews your recent changes
/council planning # challenges the current plan
/council --deep "Should we pivot to B2B?"
/council [preset] "question"
│
┌─────────────────┼─────────────────┐
│ │ │
┌─────▼─────┐ ┌──────▼──────┐ ┌───────▼───────┐
│ Persona A │ │ Persona B │ │ Persona C │
│ (Security) │ │ (Performance)│ │ (Devil's Adv) │
│ │ │ │ │ │
│ ONE idea │ │ ONE idea │ │ ONE idea │
└─────┬──────┘ └──────┬───────┘ └───────┬───────┘
│ │ │
│ all run in parallel, independent │
│ no shared context, no anchoring │
└─────────────────┼─────────────────────┘
│
┌───────▼────────┐
│ ANONYMIZATION │
│ │
│ Strip labels → │
│ "Analysis A" │
│ "Analysis B" │
│ "Analysis C" │
└───────┬─────────┘
│
┌───────▼────────┐
│ JUDGE │
│ │
│ Assistant mode: │
│ synthesizes │
│ │
│ User mode: │
│ YOU decide │
└───────┬─────────┘
│
┌───────▼────────┐
│ SAVE PROCEEDINGS│
│ (.md file) │
└────────────────┘
# Clone
git clone https://github.com/aryanbhats/ab-stack.git
# Install (symlinks skills to ~/.agents/skills/)
cd ab-stack && ./setup
# Use
/council "your question here"
| Preset | Invoke | Judge | Personas | Output |
|---|---|---|---|---|
| General | /council | Assistant | Skeptic, Pragmatist, Innovator, Risk Analyst, Devil's Advocate | Decision + reasoning + dissent |
| Planning | /council planning | User | Architect, Pragmatist, Skeptic, User Advocate, Devil's Advocate | Proceed / Revise / Rethink |
| Code Review | /council code | Assistant | Security, Performance, Correctness, Maintainability, Devil's Advocate | Approve / Revise / Reject (tiered) |
| Finance | /council finance | User | Value Investor, Growth Investor, Contrarian, Macro Analyst, Risk Manager | BUY / HOLD / SELL + conviction |
| Content | /council content | User | Brand Guardian, SEO, Audience Resonance, Conversion, Growth | Publish / Revise / Rethink + scores |
| Business | /council business | User | CEO, CFO, COO, CTO, CMO | Proceed / Pivot / Hold / Kill |
| Research | /council research | Assistant | Methodology, Literature Gap, Baseline Scout, Statistics, Applicability | Accept → Reject + scores |
The core principle. Each persona delivers one specific, actionable suggestion — not a list, not a ramble, not hedged advice.
BAD (generic, hedging):
"There are several security concerns to consider. You might want to
look at input validation, and it could be worth checking the auth
flow. Also, error messages might leak information."
Development marketplace for Superpowers core skills library
Harness-native ECC skills, hooks, rules, MCP conventions, and operator workflows
Open Design — local-first design app exposed to coding agents over MCP. Install once with your agent's plugin command and projects/files/skills are reachable through stdio.