AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Governance

skill-carbeneai-forge-governance · by CarbeneAI

Organizational governance for PAI agent infrastructure — budget controls, approval gates, goal ancestry, org chart visualization, standing orders, config portability, workspaces, and task locking. USE WHEN user mentions budget, spending, approval, governance, org chart, goals, standing orders, heartbeat, export config, import config, workspace, task lock, OR organizational management.

No reviews yet
0 installs
24 views
0.0% view→install

Install

$ agentstack add skill-carbeneai-forge-governance

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-carbeneai-forge-governance)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
3mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Governance? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Governance

Organizational intelligence for PAI, cherry-picked from Paperclip's zero-human-company model and adapted for CLI-first personal AI infrastructure.

Workflow Routing

| Action | Trigger | Tool/Config | |--------|---------|-------------| | Budget Review | "budget report", "agent spending", "cost tracking" | tools/BudgetReport.ts | | Org Chart | "org chart", "show team", "agent hierarchy" | tools/OrgChart.ts | | Goal Management | "goals", "goal context", "why are we doing this" | tools/GoalManager.ts | | Approval Check | "approval needed", "can I deploy", "governance gate" | tools/ApprovalGate.ts | | Standing Orders | "standing orders", "heartbeat", "scheduled agents" | tools/StandingOrders.ts | | Config Export | "export config", "backup PAI", "portable config" | tools/ConfigExport.ts | | Config Import | "import config", "restore PAI", "load config" | tools/ConfigImport.ts | | Task Locking | "lock task", "task checkout", "atomic task" | tools/TaskLock.ts |

Features

1. Budget Controls

Track and limit per-agent spending with cost estimates based on model pricing.

# Show spending report for current month
bun tools/BudgetReport.ts

# Compare against budget limits
bun tools/BudgetReport.ts --budget

# Filter by agent
bun tools/BudgetReport.ts --agent hiram --period week

Config: config/budgets.yml — per-agent monthly limits with alert thresholds.

2. Approval Gates

Human-in-the-loop governance for high-impact actions via Telegram.

# Check if action needs approval
bun tools/ApprovalGate.ts check deploy --agent hiram

# Request approval (sends Telegram, waits for response)
bun tools/ApprovalGate.ts request deploy --agent hiram --action "Push to carbene.ai production"

# List all gates
bun tools/ApprovalGate.ts list

Config: config/approvals.yml — gate definitions with Telegram integration.

3. Goal Ancestry

Every task traces back to a company goal. Provides "why" context for agent prompts.

# Show goal hierarchy
bun tools/GoalManager.ts list

# Trace project to goal
bun tools/GoalManager.ts trace P4

# Get injectable context block for agent prompts
bun tools/GoalManager.ts context P1

Config: goals/goals.yml — goal hierarchy with projects.

4. Org Chart Visualization

Auto-generated Mermaid diagram from agent configurations.

# Print Mermaid diagram
bun tools/OrgChart.ts

# Save as SVG (requires mmdc)
bun tools/OrgChart.ts --output svg --save org-chart.svg

5. Standing Orders (Heartbeat)

Recurring agent tasks executed on schedules — the PAI version of Paperclip's heartbeat engine.

# List all standing orders
bun tools/StandingOrders.ts list

# Enable/disable
bun tools/StandingOrders.ts enable daily-security-check

# Manually trigger
bun tools/StandingOrders.ts run friday-retro

# Sync to system cron
bun tools/StandingOrders.ts sync

Config: config/standing-orders.yml — agent schedules with goal links.

6. Config Export/Import

Portable, sanitized PAI configuration bundles for backup and sharing.

# Export (sanitized by default)
bun tools/ConfigExport.ts --output ~/exports/pai-config.tar.gz

# Import with dry run
bun tools/ConfigImport.ts pai-config.tar.gz --dry-run

# Import with merge (don't overwrite existing)
bun tools/ConfigImport.ts pai-config.tar.gz --merge

7. Multi-Workspace Isolation

Group agents, budgets, and goals by business domain.

Config: config/workspaces.yml — workspace definitions mapping agents to goals with budget pools.

8. Atomic Task Checkout

File-based locking prevents two agents from claiming the same task.

# Acquire lock
bun tools/TaskLock.ts acquire TASK-001 --agent hiram

# Release lock
bun tools/TaskLock.ts release TASK-001 --agent hiram

# Check status
bun tools/TaskLock.ts status TASK-001

# Clean stale locks (>30 min)
bun tools/TaskLock.ts cleanup --stale 30

Architecture

Governance/
├── SKILL.md                    # This file
├── tools/
│   ├── OrgChart.ts             # Mermaid org chart generator
│   ├── BudgetReport.ts         # Cost tracking and budget comparison
│   ├── GoalManager.ts          # Goal hierarchy management
│   ├── ApprovalGate.ts         # Telegram-based approval gates
│   ├── StandingOrders.ts       # Heartbeat/recurring agent tasks
│   ├── ConfigExport.ts         # Portable config export
│   ├── ConfigImport.ts         # Config import with merge
│   └── TaskLock.ts             # Atomic task checkout
├── config/
│   ├── budgets.yml             # Per-agent spend limits
│   ├── approvals.yml           # Approval gate definitions
│   ├── standing-orders.yml     # Recurring agent schedules
│   └── workspaces.yml          # Workspace isolation
└── reference/
    └── PaperclipInspiration.md # Design decisions and Paperclip comparison

Design Principles

  1. CLI-first — Every feature is a TypeScript CLI tool (PAI Constitution Principle 8)
  2. File-based state — No database required; YAML configs and file locks
  3. Telegram-native governance — Approvals via existing Telegram bot infrastructure
  4. Goal-aware — Every agent action can trace back to a company objective
  5. Sanitize by default — Config exports strip secrets automatically
  6. Composable — Each tool works independently; combine via shell pipelines

Examples

Example 1: Morning governance check

User: "How are we doing on budgets this month?"
→ Runs BudgetReport.ts --budget
→ Shows per-agent spend vs limits
→ Flags any agents approaching thresholds

Example 2: Pre-deploy approval

User: "Deploy carbene.ai changes"
→ ApprovalGate.ts check deploy --agent hiram
→ Gate triggered → sends Telegram approval request
→ Waits for the user's "approve" reply
→ Proceeds with deployment

Example 3: Why are we doing this?

User: "What goal does the trading system serve?"
→ GoalManager.ts trace P7
→ "P7 (Joseph Trading System) → G3 (Generate investment returns via automated trading)"
→ Provides full context for decision-making

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.