agentlaunch-templates 0.4.4 → 0.4.6

package/dist/claude-context.js

@@ -10,84 +10,633 @@
 export const RULES = {
     "agentlaunch.md": `# AgentLaunch Platform Rules
 
+When working with AgentLaunch tokens and the platform API:
+
 ## Constants (from smart contracts -- never change)
 
 - Deploy fee: 120 FET (read from contract, can change via multi-sig)
 - Graduation: 30,000 FET liquidity -> auto DEX listing
-- Trading fee: 2% -> 100% to protocol treasury (NO creator fee)
+- Trading fee: 2% -> 100% to REVENUE_ACCOUNT (protocol treasury)
+- NO creator fee. The 2% fee has NO split. All to protocol.
 - Total buy supply: 800,000,000 tokens per token
+- Buy price difference: 1000 (10x)
 - Default chain: BSC (testnet=97, mainnet=56)
 
 ## API Authentication
 
 - Use \`X-API-Key\` header with Agentverse API key
 - Key is read from \`.env\` AGENTVERSE_API_KEY
+- No wallet signatures needed for API calls
+- Redis caches validation for 5 minutes
 
 ## API Base URLs
 
-- Production: \`https://agent-launch.ai/api\`
-- Frontend: \`https://agent-launch.ai\`
+- Production (default): \`https://agent-launch.ai/api\`
+- Dev: \`https://launchpad-backend-dev-1056182620041.us-central1.run.app\`
+- Set \`AGENT_LAUNCH_ENV=dev\` to use dev URLs
+- Direct override: set \`AGENT_LAUNCH_API_URL\`
+
+## Wallet & Key Management
+
+- Agents have auto-provisioned Fetch.ai wallets (\`fetch1...\`) via \`agent.wallet\`
+- For BSC/EVM operations, agents store private keys via Agentverse Secrets API
+- Handoff links are used for irreversible actions (token deployment, 120 FET)
+- Autonomous trading uses agent-held keys for routine operations
+- Deploy link: \`https://agent-launch.ai/deploy/{tokenId}\`
+- Trade link: \`https://agent-launch.ai/trade/{tokenAddress}?action=buy&amount=100\`
+
+### Wallet Access (Runtime-Verified)
 
-## Key Endpoints
+- \`ctx.wallet\` **DOES NOT EXIST** — never use it
+- \`agent.wallet\` **EXISTS** — \`cosmpy.aerial.wallet.LocalWallet\`
+- \`ctx.ledger\` **EXISTS** — \`cosmpy.aerial.client.LedgerClient\`
+- Balance query: \`ctx.ledger.query_bank_balance(str(agent.wallet.address()), "atestfet")\`
+
+## Key Endpoints (VERIFIED - always use these exact paths)
 
 \`\`\`
-POST  /agents/tokenize             Create token -> handoff link
-GET   /tokens                      List tokens
-GET   /tokens/address/{address}    Token details by address
-GET   /tokens/id/{id}              Token details by ID
-GET   /tokens/calculate-buy        Preview buy
-GET   /tokens/calculate-sell       Preview sell
-GET   /agents/my-agents            List your agents
-GET   /agents/token/{address}/holders  Token holder list
-POST  /agents/auth                 Exchange API key for JWT
-GET   /comments/{address}          Get comments
-POST  /comments/{address}          Post comment
-GET   /platform/stats              Platform statistics
+GET   /tokens                             List tokens
+GET   /tokens/address/{address}           Token details by address
+GET   /tokens/id/{id}                     Token details by ID
+GET   /tokens/calculate-buy               Preview buy
+GET   /tokens/calculate-sell              Preview sell
+
+POST  /agents/tokenize                    Create token -> handoff link
+GET   /agents/my-agents                   List your agents
+GET   /agents/token/{address}/holders     Token holder list
+POST  /agents/auth                        Exchange API key for JWT
+
+GET   /comments/{address}                 Get comments
+POST  /comments/{address}                 Post comment
+
+GET   /platform/stats                     Platform stats
 \`\`\`
 
-## Handoff Protocol
+## Common Path Mistakes (NEVER use these)
 
-Token deployment requires a human signature:
-1. Agent calls API to create token record
-2. API returns handoff link
-3. Agent shares link with human
-4. Human signs transaction
+| WRONG | CORRECT |
+|-------|---------|
+| \`POST /tokenize\` | \`POST /agents/tokenize\` |
+| \`GET /token/{address}\` | \`GET /tokens/address/{address}\` |
+| \`GET /token/{address}/holders\` | \`GET /agents/token/{address}/holders\` |
+| \`GET /my-agents\` | \`GET /agents/my-agents\` |
+| \`POST /auth\` | \`POST /agents/auth\` |
 
-Once deployed, agents CAN trade autonomously via HoldingsManager + BSC_PRIVATE_KEY.
-`,
+## Fee Rule (Enforced)
+
+The 2% trading fee goes 100% to REVENUE_ACCOUNT (protocol treasury).
+There is NO creator fee split. Never write "1% creator", "split evenly",
+"creator earnings from fees", or any CREATOR_FEE constant.
+This applies to ALL files: code, docs, comments, marketing copy.
+
+## Token Lifecycle
+
+1. Agent calls POST /agents/tokenize with agentAddress, name, symbol, description, chainId
+2. API returns token record with handoff link
+3. Human visits link, connects wallet, signs transaction (pays 120 FET)
+4. Token deploys on-chain with bonding curve
+5. At 30,000 FET liquidity, auto-lists on DEX (Uniswap/PancakeSwap)`,
     "agentverse.md": `# Agentverse Deployment Rules
 
-## API: https://agentverse.ai/v1
-## Auth: Authorization: Bearer <AGENTVERSE_API_KEY>
+When deploying agents to Agentverse:
+
+## API Base URL
+
+\`\`\`
+https://agentverse.ai/v1
+\`\`\`
+
+## Auth Header
+
+\`\`\`
+Authorization: Bearer <AGENTVERSE_API_KEY>
+\`\`\`
+
+Note: Capital "Bearer", followed by space, then the raw API key.
 
 ## CRITICAL: Code Upload Format
 
 The code field MUST be double-encoded JSON:
+
 \`\`\`python
 code_array = [{"language": "python", "name": "agent.py", "value": source_code}]
 payload = {"code": json.dumps(code_array)}
 \`\`\`
 
+Sending the raw array (not json.dumps'd) causes: "Invalid code format"
+
 ## Agent Code Requirements
 
-- Use \`Agent()\` with zero params
-- Use Chat Protocol v0.3.0
-- Must have \`ChatAcknowledgement\` handler
+- Use \`Agent()\` with zero params (Agentverse provides config)
+- Use Chat Protocol v0.3.0 from \`uagents_core.contrib.protocols.chat\`
+- Must have \`@chat_proto.on_message(ChatAcknowledgement)\` handler
 - Use \`ctx.logger\`, never \`print()\`
-- Use \`datetime.now()\` not \`utcnow()\`
-- Include \`publish_manifest=True\`
+- Use \`datetime.now()\` not \`datetime.utcnow()\` (deprecated)
+- End sessions with \`EndSessionContent\`
+- Include \`publish_manifest=True\` in \`agent.include()\`
 
 ## Deployment Flow
 
-1. POST /v1/hosting/agents -> creates agent
-2. PUT /v1/hosting/agents/{addr}/code -> upload
-3. POST /v1/hosting/secrets -> set secrets
-4. POST /v1/hosting/agents/{addr}/start -> start
-5. Poll until compiled (15-60s)
-`,
+1. \`POST /v1/hosting/agents\` -- creates agent, returns address
+2. \`PUT /v1/hosting/agents/{addr}/code\` -- upload (double-encoded!)
+3. \`POST /v1/hosting/secrets\` -- set AGENTVERSE_API_KEY etc.
+4. \`POST /v1/hosting/agents/{addr}/start\` -- start agent
+5. \`GET /v1/hosting/agents/{addr}\` -- poll until compiled=true (15-60s)
+6. \`GET /v1/hosting/agents/{addr}/logs\` -- verify running
+
+## Common Errors
+
+- "Invalid code format" -- code field not json.dumps'd
+- Agent stuck in "compiling" -- wait longer, check logs for syntax errors
+- 401 Unauthorized -- bad API key or missing "bearer" prefix
+- Agent listing response is \`{ items: [...] }\` not \`{ agents: [...] }\`
+
+## Available Packages on Agentverse
+
+- \`uagents\`, \`uagents_core\`
+- \`requests\`, \`openai\`
+- Python standard library
+
+## Secrets
+
+Set secrets via POST /v1/hosting/secrets:
+\`\`\`json
+{
+  "address": "agent1q...",
+  "name": "AGENTVERSE_API_KEY",
+  "secret": "av-xxx"
+}
+\`\`\`
+
+List secrets: GET /v1/hosting/secrets?address=agent1q...`,
+    "api-design.md": `---
+paths:
+  - "packages/sdk/src/**/*.ts"
+  - "packages/cli/src/**/*.ts"
+  - "packages/mcp/src/**/*.ts"
+---
+
+# API Design Rules
+
+## Endpoints
+- GET /api/{resource} - List
+- POST /api/{resource} - Create
+- GET /api/{resource}/:id - Read
+- PUT /api/{resource}/:id - Update
+- DELETE /api/{resource}/:id - Delete
+
+## Response Format
+{
+  "success": boolean,
+  "data": object | array,
+  "error": null | { "code": string, "message": string }
+}
+
+## Status Codes
+- 200: Success
+- 201: Created
+- 400: Bad Request (validation)
+- 401: Unauthorized
+- 403: Forbidden
+- 404: Not Found
+- 500: Server Error
+
+## Pagination
+- Query: ?page=1&limit=20
+- Response meta: { page, limit, total, totalPages }`,
+    "api-paths.md": `# API Paths Reference (VERIFIED)
+
+These paths have been tested against the live API and are the ONLY correct paths.
+When writing code, tests, or documentation, ALWAYS use these exact paths.
+
+## Base URL
+
+\`\`\`
+https://agent-launch.ai/api
+\`\`\`
+
+## Verified Endpoints
+
+### Tokens
+
+| Method | Path | Description |
+|--------|------|-------------|
+| \`GET\` | \`/tokens\` | List all tokens |
+| \`GET\` | \`/tokens/address/{address}\` | Get token by contract address |
+| \`GET\` | \`/tokens/id/{id}\` | Get token by numeric ID |
+| \`GET\` | \`/tokens/calculate-buy\` | Simulate buy (params: address, fetAmount) |
+| \`GET\` | \`/tokens/calculate-sell\` | Simulate sell (params: address, tokenAmount) |
+
+### Agents
+
+| Method | Path | Description |
+|--------|------|-------------|
+| \`POST\` | \`/agents/tokenize\` | Create token for an agent |
+| \`GET\` | \`/agents/my-agents\` | List caller's Agentverse agents |
+| \`GET\` | \`/agents/token/{address}/holders\` | Get token holder list |
+| \`POST\` | \`/agents/auth\` | Exchange API key for JWT |
+
+### Comments
+
+| Method | Path | Description |
+|--------|------|-------------|
+| \`GET\` | \`/comments/{address}\` | Get comments for a token |
+| \`POST\` | \`/comments/{address}\` | Post a comment |
+
+### Platform
+
+| Method | Path | Description |
+|--------|------|-------------|
+| \`GET\` | \`/platform/stats\` | Platform-wide statistics |
+
+## WRONG vs CORRECT Paths
+
+These mistakes are common. NEVER use the wrong paths:
+
+| WRONG (will 404) | CORRECT |
+|------------------|---------|
+| \`POST /tokenize\` | \`POST /agents/tokenize\` |
+| \`GET /token/{address}\` | \`GET /tokens/address/{address}\` |
+| \`GET /tokens/{address}\` | \`GET /tokens/address/{address}\` |
+| \`GET /tokens/{id}\` | \`GET /tokens/id/{id}\` |
+| \`GET /token/{address}/holders\` | \`GET /agents/token/{address}/holders\` |
+| \`GET /my-agents\` | \`GET /agents/my-agents\` |
+| \`POST /auth\` | \`POST /agents/auth\` |
+| \`GET /stats\` | \`GET /platform/stats\` |
+
+## SDK Function to Path Mapping
+
+| SDK Function | Correct Path |
+|--------------|--------------|
+| \`listTokens()\` | \`GET /tokens\` |
+| \`getToken(address)\` | \`GET /tokens/address/{address}\` |
+| \`tokenize(params)\` | \`POST /agents/tokenize\` |
+| \`calculateBuy(address, amount)\` | \`GET /tokens/calculate-buy\` |
+| \`calculateSell(address, amount)\` | \`GET /tokens/calculate-sell\` |
+| \`getTokenHolders(address)\` | \`GET /agents/token/{address}/holders\` |
+| \`getComments(address)\` | \`GET /comments/{address}\` |
+| \`postComment(params)\` | \`POST /comments/{address}\` |
+| \`authenticate(key)\` | \`POST /agents/auth\` |
+| \`getMyAgents()\` | \`GET /agents/my-agents\` |
+| \`getPlatformStats()\` | \`GET /platform/stats\` |
+
+## Enforcement
+
+When reviewing or writing code that makes API calls:
+1. Check that the path matches this reference
+2. If you see an old/wrong path, fix it immediately
+3. Reference \`docs/paths.md\` for full documentation with examples`,
+    "consumer-payments.md": `# Consumer Payments Rules
+
+## Multi-Token Support
+
+The toolkit supports FET and USDC on BSC (Testnet chain 97, Mainnet chain 56).
+
+### Known Token Addresses
+
+| Token | Chain | Address |
+|-------|-------|---------|
+| FET | BSC Testnet (97) | \`0x304ddf3eE068c53514f782e2341B71A80c8aE3C7\` |
+| FET | BSC Mainnet (56) | \`0xBd5df99ABe0E2b1e86BE5eC0039d1e24de28Fe87\` |
+| USDC | BSC Testnet (97) | \`0x64544969ed7EBf5f083679233325356EbE738930\` |
+| USDC | BSC Mainnet (56) | \`0x8AC76a51cc950d9822D68b83fE1Ad97B32Cd580d\` |
+
+### Token Registry
+
+Use \`getToken(symbol, chainId)\` to look up tokens. Never hardcode addresses.
+
+\`\`\`typescript
+import { getToken, KNOWN_TOKENS } from 'agentlaunch-sdk';
+const fet = getToken('FET', 97);   // PaymentToken object
+const usdc = getToken('USDC', 97); // PaymentToken object
+\`\`\`
+
+## Spending Delegation
+
+Uses standard ERC-20 \`approve()\` / \`transferFrom()\` — no custom contracts.
+
+### Flow
+
+1. Agent generates delegation handoff link → sends to human
+2. Human opens link, connects wallet, signs \`approve()\` transaction
+3. Agent checks allowance on-chain with \`checkAllowance()\`
+4. Agent spends from delegation with \`spendFromDelegation()\` (calls \`transferFrom\`)
+
+### SDK Functions
+
+\`\`\`typescript
+import { checkAllowance, spendFromDelegation, createSpendingLimitHandoff } from 'agentlaunch-sdk';
+
+// Generate link for human to approve
+const link = createSpendingLimitHandoff({ tokenSymbol: 'FET', amount: '100' }, agentWallet);
+
+// Check on-chain allowance
+const limit = await checkAllowance(tokenAddress, ownerAddress, spenderAddress, 97);
+
+// Spend from delegation
+const result = await spendFromDelegation(tokenAddress, owner, recipient, '10');
+\`\`\`
+
+## Invoices
+
+Invoices are stored in Agentverse agent storage (consistent with existing revenue/pricing patterns).
+
+### Storage Keys
+
+- \`invoice_{id}\` — Individual invoice JSON
+- \`invoice_index\` — Array of invoice IDs
+
+### Invoice Lifecycle
+
+\`pending\` → \`paid\` → (done)
+\`pending\` → \`expired\`
+\`paid\` → \`refunded\`
+\`paid\` → \`disputed\`
+
+### SDK Functions
+
+\`\`\`typescript
+import { createInvoice, getInvoice, listInvoices, updateInvoiceStatus } from 'agentlaunch-sdk';
+
+const inv = await createInvoice(agentAddress, { id, issuer, payer, service, amount });
+const invoices = await listInvoices(agentAddress, 'pending');
+await updateInvoiceStatus(agentAddress, invoiceId, 'paid', txHash);
+\`\`\`
+
+## Fiat Onramp
+
+Fiat is handoff-only — generate URLs to MoonPay/Transak, never process fiat directly.
+
+### Supported Providers
+
+| Provider | Env Var | URL |
+|----------|---------|-----|
+| MoonPay | \`MOONPAY_API_KEY\` | \`https://buy.moonpay.com\` |
+| Transak | \`TRANSAK_API_KEY\` | \`https://global.transak.com\` |
+
+### SDK Function
+
+\`\`\`typescript
+import { generateFiatOnrampLink } from 'agentlaunch-sdk';
+
+const link = generateFiatOnrampLink({
+  fiatAmount: '50',
+  fiatCurrency: 'USD',
+  cryptoToken: 'FET',
+  walletAddress: '0x...',
+  provider: 'moonpay',
+});
+// Returns { provider, url }
+\`\`\`
+
+## Consumer Commerce Template
+
+The \`consumer-commerce\` template generates agents with:
+- MultiTokenPricingTable — FET + USDC columns
+- InvoiceManager — CRUD in ctx.storage
+- FiatOnrampHelper — card purchase link generation
+- DelegationChecker — verify allowance before auto-pay
+
+### Presets
+
+| Preset | Symbol | Use Case |
+|--------|--------|----------|
+| \`payment-processor\` | $PAY | Multi-token payment routing |
+| \`booking-agent\` | $BOOK | Service booking + payment |
+| \`subscription-manager\` | $SUB | Recurring billing with delegation |
+| \`escrow-service\` | $ESCR | Funds held until delivery confirmed |
+
+### Scaffold Command
+
+\`\`\`bash
+agentlaunch scaffold my-agent --type consumer-commerce
+\`\`\`
+
+## MCP Tools
+
+| Tool | Description |
+|------|-------------|
+| \`multi_token_payment\` | Send FET, USDC, or ERC-20 payment |
+| \`check_spending_limit\` | Read ERC-20 allowance |
+| \`create_delegation\` | Generate delegation handoff link |
+| \`get_fiat_link\` | Generate MoonPay/Transak URL |
+| \`create_invoice\` | Create payment invoice |
+| \`list_invoices\` | List invoices by status |
+| \`get_multi_token_balances\` | Query FET + USDC + BNB balances |
+
+## CLI Commands
+
+\`\`\`bash
+agentlaunch wallet balances                          # Multi-token balance display
+agentlaunch wallet delegate FET 100 --spender 0x...  # Generate delegation link
+agentlaunch wallet allowance FET --owner 0x... --spender 0x...  # Check limit
+agentlaunch wallet send USDC 0x... 10                # Multi-token transfer
+agentlaunch pay 0x... 10 --token USDC                # Pay in any token
+agentlaunch invoice create --agent agent1q... --payer 0x... --service api --amount 10
+agentlaunch invoice list --agent agent1q... --status pending
+\`\`\`
+
+## Backward Compatibility
+
+- All new type fields are optional — existing FET-only agents are unchanged
+- New fields on PricingEntry: \`altPrices?\`, \`acceptedTokens?\`
+- New fields on AgentRevenue: \`revenueByToken?\`
+- New fields on AgentCommerceStatus: \`usdcBalance?\`, \`tokenBalances?\`, \`activeInvoices?\`, \`delegations?\`
+- Default token is always FET when not specified`,
+    "marketing-swarm.md": `# Marketing Team Swarm Rules
+
+## The 7 Roles
+
+| Role | Token | Services | Price/call | Interval |
+|------|-------|----------|-----------|----------|
+| Writer | $WRITE | blog_post, tweet_thread, newsletter, ad_copy | 0.01 FET | on-demand |
+| Social | $POST | post_tweet, schedule_thread, reply_mentions | 0.005 FET | 5 min |
+| Community | $COMM | moderate, answer_faq, run_poll | 0.002 FET | 1 min |
+| Analytics | $STATS | engagement_report, audience_insights, content_performance | 0.005 FET | 5 min |
+| Outreach | $REACH | find_partners, draft_pitch, send_email | 0.01 FET | on-demand |
+| Ads | $ADS | create_ad, ab_test, campaign_report | 0.01 FET | 5 min |
+| Strategy | $PLAN | content_calendar, brand_audit, competitor_analysis, campaign_plan | 0.02 FET | on-demand |
+
+## Build Order
+
+Writer -> Community -> Social -> Analytics -> Outreach -> Ads -> Strategy
+
+Writer first (everyone needs content), Community second (standalone engagement).
+
+## Starter Configurations
+
+- **Content only**: Writer (1 agent)
+- **Social presence**: Writer + Social (2 agents)
+- **Community**: Writer + Community + Social (3 agents)
+- **Analytics stack**: Writer + Social + Analytics (3 agents)
+- **Full team**: All 7
+
+## Customizing SwarmBusiness
+
+The swarm-starter template marks the business logic section:
+\`\`\`python
+# === YOUR SWARM LOGIC ===
+\`\`\`
+
+This is where you add:
+- Custom message handlers for your services
+- Interval tasks for background work
+- Integration with external APIs
+- Agent-to-agent communication logic
+
+## Adding New Roles
+
+1. Define the role's services and pricing
+2. Create a preset in \`packages/templates/src/presets.ts\`
+3. Generate from swarm-starter template with role variables
+4. Deploy and wire into the swarm
+
+## Cross-Holdings
+
+Agents can buy each other's tokens:
+- Strategy buys Writer tokens (values its content)
+- Writer buys Analytics tokens (values its performance data)
+- Creates economic alignment between agents
+
+## Token Lifecycle
+
+1. Deploy agent on Agentverse
+2. Tokenize on AgentLaunch (120 FET deploy fee)
+3. Bonding curve active (2% fee to protocol treasury, NO creator fee)
+4. At 30,000 FET -> auto DEX listing (graduation)`,
+    "payment-protocol.md": `# Payment Protocol Rules
+
+> **Status:** Payment protocol import verified available on Agentverse (2026-03-04).
+
+## Official Imports (uagents_core)
+
+\`\`\`python
+from uagents_core.contrib.protocols.payment import (
+    RequestPayment,
+    CommitPayment,
+    CompletePayment,
+    RejectPayment,
+    CancelPayment,
+    Funds,
+    payment_protocol_spec,
+)
+\`\`\`
+
+## Protocol Creation
+
+\`\`\`python
+from uagents import Protocol
+
+# Seller side (agents that charge for services):
+payment_proto = Protocol(spec=payment_protocol_spec, role="seller")
+
+# Buyer side (agents that pay for services):
+payment_proto = Protocol(spec=payment_protocol_spec, role="buyer")
+\`\`\`
+
+**Note:** The payment protocol has defined roles. You MUST specify \`role="seller"\` or \`role="buyer"\` when creating the protocol. \`agent.create_protocol()\` does NOT exist on Agentverse. Always use \`Protocol(spec=..., role=...)\`.
+
+## Payment Flow
+
+\`\`\`
+Buyer                          Seller
+  |                              |
+  |  ChatMessage (service req)   |
+  |----------------------------->|
+  |                              |
+  |  RequestPayment              |
+  |<-----------------------------|
+  |                              |
+  |  CommitPayment (tx_hash)     |
+  |----------------------------->|
+  |                              |
+  |  [verifies on-chain]         |
+  |                              |
+  |  CompletePayment (result)    |
+  |<-----------------------------|
+\`\`\`
+
+## Denomination
+
+| Network | Denom | Rate |
+|---------|-------|------|
+| Testnet | atestfet | 1 FET = 10^18 atestfet |
+| Mainnet | afet | 1 FET = 10^18 afet |
+
+Common amounts:
+- 0.001 FET = 1,000,000,000,000,000 atestfet
+- 0.01 FET = 10,000,000,000,000,000 atestfet
+- 1 FET = 1,000,000,000,000,000,000 atestfet
+
+## Error Handling
+
+- Always handle \`RejectPayment\` -- buyer may decline
+- Always handle \`CancelPayment\` -- timeout or cancellation
+- Verify tx_hash on-chain before delivering service
+- Store transaction log in \`ctx.storage\`
+
+## Swarm-Starter Template Commerce Layers
+
+The swarm-starter template includes these commerce classes inline:
+- \`PaymentService\`: Seller-side payment handling
+- \`PricingTable\`: Per-service pricing from ctx.storage
+- \`TierManager\`: Token-gated access (free/premium)
+- \`WalletManager\`: Balance queries, fund alerts
+- \`RevenueTracker\`: Income/expense logging
+- \`SelfAwareMixin\`: Token price/holder awareness
+- \`HoldingsManager\`: Direct on-chain token operations`,
+    "security-checklist.md": `---
+paths:
+  - "packages/**/*.ts"
+---
+
+# Security Checklist
+
+## Backend
+- Input validation on all endpoints
+- JWT expiration configured
+- Secrets in .env only
+- CORS restricted
+- Error messages don't leak internals
+
+## Frontend
+- No secrets in client code
+- Sanitize user input
+- Validate wallet addresses
+
+## Smart Contracts
+- Reentrancy guard on state changes
+- Access control validated
+- Events for all state changes
+- No hardcoded addresses (use constructor)
+
+## General
+- npm audit regularly
+- No private keys in code
+- .env in .gitignore`,
+    "testing.md": `---
+paths:
+  - "packages/**/*.spec.ts"
+  - "packages/**/*.test.ts"
+---
+
+# Testing Rules
+
+## SDK / CLI / MCP (Jest)
+- Co-located: *.spec.ts or *.test.ts
+- Mock with jest.mock()
+- beforeEach/afterEach for setup/cleanup
+
+## Integration Tests
+- Test SDK against mock API server
+- Test CLI commands with mocked HTTP
+- Test MCP tools with mock context`,
     "uagent-patterns.md": `# uAgent Code Patterns
 
-> For new agents, run: \`npx agentlaunch myagent\` (uses chat-memory template by default)
+When writing Agentverse agent code:
+
+> **For new agents, use the swarm-starter template:** \`agentlaunch scaffold myagent --type swarm-starter\`
+> It includes the full commerce stack (payments, pricing, tiers, revenue tracking).
 
 ## Minimal Working Agent
 
@@ -99,7 +648,7 @@ from uagents_core.contrib.protocols.chat import (
 )
 from datetime import datetime
 
-agent = Agent()
+agent = Agent()  # Zero params on Agentverse
 chat_proto = Protocol(spec=chat_protocol_spec)
 
 @chat_proto.on_message(ChatMessage)
@@ -112,6 +661,7 @@ async def handle_message(ctx: Context, sender: str, msg: ChatMessage):
         msg_id=msg.msg_id,
         content=[TextContent(text=f"You said: {text}")]
     ))
+
     await ctx.send(sender, ChatMessage(
         timestamp=datetime.now(),
         msg_id=msg.msg_id,
@@ -120,7 +670,7 @@ async def handle_message(ctx: Context, sender: str, msg: ChatMessage):
 
 @chat_proto.on_message(ChatAcknowledgement)
 async def handle_ack(ctx: Context, sender: str, msg: ChatAcknowledgement):
-    ctx.logger.info(f"Ack: {msg.acknowledged_msg_id}")
+    ctx.logger.info(f"Ack from {sender}: {msg.acknowledged_msg_id}")
 
 agent.include(chat_proto, publish_manifest=True)
 
@@ -128,35 +678,34 @@ if __name__ == "__main__":
     agent.run()
 \`\`\`
 
-## Payment Protocol (Official)
+## Rules
 
-\`\`\`python
-from uagents_core.contrib.protocols.payment import (
-    RequestPayment, CommitPayment, CompletePayment,
-    RejectPayment, CancelPayment, Funds, payment_protocol_spec,
-)
+- Always use \`Agent()\` with zero params on Agentverse
+- Always include \`ChatAcknowledgement\` handler (required by protocol)
+- Always end conversations with \`EndSessionContent\`
+- Always use \`ctx.logger\` (never \`print\`)
+- Always use \`datetime.now()\` (never \`utcnow\`, it is deprecated)
+- Always include \`publish_manifest=True\` in \`agent.include()\`
+- Use \`Protocol(spec=chat_protocol_spec)\` — \`agent.create_protocol()\` does NOT exist on Agentverse
 
-# Seller (service provider)
-seller_proto = Protocol(spec=payment_protocol_spec)
+## Required Imports
 
-# Buyer (service consumer)
-buyer_proto = Protocol(spec=payment_protocol_spec)
+\`\`\`python
+from uagents import Agent, Context, Protocol, Model
+from uagents_core.contrib.protocols.chat import (
+    ChatAcknowledgement, ChatMessage, TextContent,
+    EndSessionContent, chat_protocol_spec
+)
+from datetime import datetime
 \`\`\`
 
-## Commerce Layer (Swarm-Starter Template)
+---
 
-The swarm-starter template includes inline commerce classes:
-- PaymentService: Charge callers, pay other agents
-- PricingTable: Per-service pricing from ctx.storage
-- TierManager: Token-gated access (free/premium)
-- WalletManager: Balance queries, fund alerts
-- RevenueTracker: Income/expense logging
-- SelfAwareMixin: Token price/holder awareness
-- HoldingsManager: Buy/sell other agents' tokens
-`,
-    "payment-protocol.md": `# Payment Protocol Rules
+## Payment Protocol (Official)
 
-## Official Imports (uagents_core)
+Use the official payment protocol from \`uagents_core\`. Do NOT create custom payment models.
+
+### Imports
 
 \`\`\`python
 from uagents_core.contrib.protocols.payment import (
@@ -170,226 +719,1424 @@ from uagents_core.contrib.protocols.payment import (
 )
 \`\`\`
 
-## Role-Based Protocol Creation
+### Protocol Creation
 
 \`\`\`python
-# Seller (service provider)
-seller_proto = Protocol(spec=payment_protocol_spec)
+from uagents import Protocol
 
-# Buyer (service consumer)
-buyer_proto = Protocol(spec=payment_protocol_spec)
-\`\`\`
+# Seller side (agents that charge for services):
+payment_proto = Protocol(spec=payment_protocol_spec, role="seller")
 
-## Payment Flow
+# Buyer side (agents that pay for services):
+payment_proto = Protocol(spec=payment_protocol_spec, role="buyer")
+\`\`\`
 
-Buyer sends ChatMessage -> Seller sends RequestPayment -> Buyer sends CommitPayment (tx_hash) -> Seller verifies on-chain -> Seller sends CompletePayment.
+**Note:** The payment protocol has defined roles. You MUST specify \`role="seller"\` or \`role="buyer"\` when creating the protocol. \`agent.create_protocol()\` does NOT exist on Agentverse. Always use \`Protocol(spec=..., role=...)\`.
 
-## Denomination
+### Payment Flow
 
-- Testnet: atestfet (1 FET = 10^18 atestfet)
-- Mainnet: afet (1 FET = 10^18 afet)
+\`\`\`
+Buyer                          Seller
+  |                              |
+  |  ChatMessage (service req)   |
+  |----------------------------->|
+  |                              |
+  |  RequestPayment              |
+  |<-----------------------------|
+  |                              |
+  |  CommitPayment (tx_hash)     |
+  |----------------------------->|
+  |                              |
+  |  [verifies on-chain]         |
+  |                              |
+  |  CompletePayment (result)    |
+  |<-----------------------------|
+\`\`\`
 
-## Error Handling
+### Error Handling
 
-- Always handle RejectPayment -- buyer may decline
-- Always handle CancelPayment -- timeout or cancellation
+- Always handle \`RejectPayment\` -- buyer may decline
+- Always handle \`CancelPayment\` -- timeout or cancellation
 - Verify tx_hash on-chain before delivering service
-- Store transaction log in ctx.storage
+- Store transaction log in \`ctx.storage\`
 
-## Swarm-Starter Template Commerce Layers
+See \`.claude/rules/payment-protocol.md\` for the full reference.
 
-The swarm-starter template includes these commerce classes inline:
-- PaymentService, PricingTable, TierManager
-- WalletManager, RevenueTracker, SelfAwareMixin, HoldingsManager
-`,
-    "marketing-swarm.md": `# Marketing Team Rules
+---
 
-## The 7 Roles
+## Wallet & Ledger Operations (Runtime-Verified 2026-03-04)
 
-| Role | Token | Services | Price/call |
-|------|-------|----------|-----------|
-| Writer | $WRITE | blog_post, tweet_thread, newsletter, ad_copy | 0.01 FET |
-| Social | $POST | post_tweet, schedule_thread, reply_mentions | 0.005 FET |
-| Community | $COMM | moderate, answer_faq, run_poll | 0.002 FET |
-| Analytics | $STATS | engagement_report, audience_insights, content_performance | 0.005 FET |
-| Outreach | $REACH | find_partners, draft_pitch, send_email | 0.01 FET |
-| Ads | $ADS | create_ad, ab_test, campaign_report | 0.01 FET |
-| Strategy | $PLAN | content_calendar, brand_audit, competitor_analysis, campaign_plan | 0.02 FET |
+Every agent on Agentverse has a wallet. Key facts:
 
-## Build Order
+- **\`ctx.ledger\` EXISTS** — type \`cosmpy.aerial.client.LedgerClient\` (verified on Agentverse)
+- **\`ctx.wallet\` DOES NOT EXIST** — not on the Context object
+- **\`agent.wallet\` EXISTS** — type \`cosmpy.aerial.wallet.LocalWallet\` (use this instead)
+- **Balance queries WORK** — \`ctx.ledger.query_bank_balance()\` returns int
 
-Writer -> Community -> Social -> Analytics -> Outreach -> Ads -> Strategy
+### Complete \`ctx\` Attributes (verified)
 
-Writer first (everyone needs content), Community second (standalone engagement).
+\`\`\`
+address, agent, envelopes, get_message_protocol, identifier,
+ledger, logger, outbound_messages, protocols, send,
+send_and_receive, send_raw, send_wallet_message, session,
+session_history, storage, wallet_messages
+\`\`\`
 
-## Starter Configurations
+### Check Balance
 
-- Content only: Writer (1 agent)
-- Social presence: Writer + Social (2 agents)
-- Community: Writer + Community + Social (3 agents)
-- Analytics stack: Writer + Social + Analytics (3 agents)
-- Full team: All 7
+\`\`\`python
+agent = Agent()
 
-## Cross-Holdings
+async def check_balance(ctx: Context) -> int:
+    """Get agent's FET balance in atestfet (1 FET = 10^18 atestfet)."""
+    # Use agent.wallet (NOT ctx.wallet — it doesn't exist)
+    balance = ctx.ledger.query_bank_balance(str(agent.wallet.address()), "atestfet")
+    return int(balance)
+\`\`\`
 
-Agents buy each other's tokens for economic alignment.
-- Strategy buys Writer tokens (values its content)
-- Writer buys Analytics tokens (values its performance data)
+### Send FET to Another Agent
 
-## Token Lifecycle
+\`\`\`python
+agent = Agent()
 
-1. Deploy on Agentverse
-2. Tokenize on AgentLaunch (120 FET deploy fee)
-3. Bonding curve active (2% fee to protocol treasury, NO creator fee)
-4. At 30,000 FET -> auto DEX listing (graduation)
-`,
-    "api-design.md": `# API Design Rules
+async def send_fet(ctx: Context, recipient: str, amount_afet: int, denom: str = "atestfet") -> str:
+    """Send FET on Fetch.ai native chain."""
+    tx = ctx.ledger.send_tokens(
+        recipient, amount_afet, denom, agent.wallet  # agent.wallet, NOT ctx.wallet
+    ).wait_to_complete()
+    ctx.logger.info(f"Sent {amount_afet} {denom} to {recipient}: {tx.tx_hash}")
+    return tx.tx_hash
+\`\`\`
+
+### Denomination
 
-## Response Format
-\`\`\`json
-{
-  "success": boolean,
-  "data": object | array,
-  "error": null | { "code": string, "message": string }
-}
 \`\`\`
+TESTNET: "atestfet" (1 FET = 10^18 atestfet)
+MAINNET: "afet"     (1 FET = 10^18 afet)
 
-## Status Codes
-- 200: Success
-- 201: Created
-- 400: Bad Request
-- 401: Unauthorized
-- 404: Not Found
-- 429: Rate Limited
-- 500: Server Error
-`,
-    "security-checklist.md": `# Security Checklist
+Example: 0.01 FET = 10_000_000_000_000_000 atestfet
+\`\`\`
 
-## Agent Code
-- Input validation on all messages
-- Rate limiting per user
-- Secrets in .env only
-- Never log sensitive data
-- Sanitize user input
+### Two Wallet Systems
 
-## General
-- No private keys in code
-- .env in .gitignore
-- Use ctx.logger not print
-`,
-    "testing.md": `# Testing Rules
+Agents may need TWO wallets for full functionality:
+
+| | Fetch.ai Native (Cosmos) | BSC (EVM) |
+|---|---|---|
+| **Address** | \`fetch1...\` (auto-provisioned) | \`0x...\` (key in Agentverse Secrets) |
+| **Access** | \`agent.wallet\` + \`ctx.ledger\` | web3.py + \`ctx.get_secret("WALLET_KEY")\` |
+| **Used for** | Agent-to-agent FET payments | Token trading, bonding curve, deploy fee |
+| **Library** | cosmpy (built-in) | web3, eth_account |
+
+---
+
+## Commerce Layer (Swarm-Starter Template)
+
+The swarm-starter template includes a complete commerce stack inline. These classes are
+generated directly into the agent code -- no external imports needed.
+
+| Class | Purpose |
+|-------|---------|
+| \`PaymentService\` | Charge callers, pay other agents, verify on-chain transactions |
+| \`PricingTable\` | Per-service pricing stored in \`ctx.storage\` |
+| \`TierManager\` | Token-gated access: free tier vs. premium (hold tokens to unlock) |
+| \`WalletManager\` | Balance queries, low-fund alerts |
+| \`RevenueTracker\` | Income/expense logging, GDP contribution |
+| \`SelfAwareMixin\` | Read own token price, holder count, market cap |
+| \`HoldingsManager\` | Buy/sell other agents' tokens for cross-holdings |
+
+To use the commerce layer, scaffold with the swarm-starter template:
 
-## Local Testing
 \`\`\`bash
-source .env && python agent.py
+agentlaunch scaffold myagent --type swarm-starter
+# Or with a preset:
+agentlaunch scaffold oracle-agent --type swarm-starter --preset oracle
+\`\`\`
+
+See \`.claude/rules/marketing-swarm.md\` for the 7 preset roles and build order.
+
+---
+
+## Storage Patterns
+
+Use \`ctx.storage\` for persistent data:
+
+\`\`\`python
+# Set value (strings only)
+ctx.storage.set("my_key", "my_value")
+ctx.storage.set("count", str(123))
+ctx.storage.set("data", json.dumps({"foo": "bar"}))
+
+# Get value
+value = ctx.storage.get("my_key")  # Returns str or None
+count = int(ctx.storage.get("count") or 0)
+data = json.loads(ctx.storage.get("data") or "{}")
+
+# Check exists
+if ctx.storage.get("initialized"):
+    pass
+\`\`\`
+
+---
+
+## Interval Tasks
+
+\`\`\`python
+@agent.on_interval(period=300.0)  # Every 5 minutes
+async def periodic_task(ctx: Context):
+    ctx.logger.info("Running periodic task")
+    balance = await check_balance(ctx)
+    ctx.storage.set("last_balance", str(balance))
+\`\`\`
+
+---
+
+## Agentverse Allowed Imports
+
+These are available on Agentverse hosted agents:
+
+\`\`\`
+uagents, uagents_core           # Agent framework
+cosmpy                          # Ledger operations
+pydantic                        # Data validation
+requests, aiohttp               # HTTP
+web3, eth_account               # EVM interaction
+datetime, json, hashlib, uuid   # Standard library
+\`\`\`
+
+---
+
+## Common Gotchas
+
+1. **Double-encoded code upload** -- When uploading via API, code must be JSON stringified twice
+2. **ChatAcknowledgement required** -- Always handle it, even if empty
+3. **datetime.utcnow() deprecated** -- Use \`datetime.now()\`
+4. **Agent listing is \`items\`** -- Response is \`{ items: [...] }\` not \`{ agents: [...] }\`
+5. **Wait for compilation** -- 15-60s after start before agent responds
+6. **Balance in atestfet** -- 1 FET = 10^18 atestfet, always use int not float
+7. **Use official payment protocol** -- Import from \`uagents_core.contrib.protocols.payment\`, not custom models (verified available on Agentverse)
+8. **agent.create_protocol() does NOT exist** -- Use \`Protocol(spec=...)\` directly
+9. **ctx.wallet DOES NOT EXIST** -- Use \`agent.wallet\` (module-scope Agent object). Never write \`ctx.wallet\`
+10. **ctx.ledger EXISTS** -- It's a \`cosmpy.aerial.client.LedgerClient\`. No \`hasattr\` guard needed
+11. **Logs endpoint is /logs/latest** -- \`GET /v1/hosting/agents/{addr}/logs/latest\`, NOT \`/logs\``,
+    "workflow.md": `# Agent Lifecycle Workflow (MANDATORY)
+
+**Every agent deployment MUST follow the full workflow from \`docs/workflow.md\`.**
+Skipping phases degrades agent ranking, discoverability, and ASI:One routing.
+
+## The 8 Phases
+
+\`\`\`
+[1] Create    -> [2] Deploy    -> [3] Optimize -> [4] Tokenize
+[5] Handoff   -> [6] Discover  -> [7] Trade    -> [8] Grow
 \`\`\`
 
-## Deploy Testing
+## Phase 3: Optimize (NEVER SKIP)
+
+After deploying an agent, you MUST complete the Setup Checklist before moving on:
+
+### Required (API-settable)
+
+1. **README** -- Write a compelling README with:
+   - One-line value proposition
+   - Capabilities list with specific examples
+   - 2-3 example conversations
+   - Pricing table (if agent charges)
+   - Keywords for search
+   - Push via \`PUT /v1/hosting/agents/{addr}\` with \`{"readme": "..."}\`
+
+2. **Short Description** -- 1-2 sentence bio for directory listings
+   - Push via \`PUT /v1/hosting/agents/{addr}\` with \`{"short_description": "..."}\`
+
+3. **Avatar** (if available) -- Custom image for visual identity
+   - Push via \`PUT /v1/hosting/agents/{addr}\` with \`{"avatar_url": "..."}\`
+
+### Manual (inform user)
+
+4. **@Handle** -- Dashboard only, no API. Suggest 3-5 options.
+5. **3+ Interactions** -- Tell user to chat with agent or use Response QA Agent.
+6. **Domain Verification** -- Optional bonus, DNS TXT record.
+
+### API Call Pattern
+
 \`\`\`bash
-agentlaunch deploy
-agentlaunch status <address>
+curl -s -X PUT "https://agentverse.ai/v1/hosting/agents/{addr}" \\
+  -H "Authorization: Bearer $AGENTVERSE_API_KEY" \\
+  -H "Content-Type: application/json" \\
+  -d '{
+    "readme": "# Agent Name\\n\\nDescription...",
+    "short_description": "One-line description of what this agent does."
+  }'
 \`\`\`
-`,
+
+## Enforcement
+
+When deploying agents (via /deploy, /build-agent, /build-swarm, or manual API calls):
+
+1. After confirming agent is compiled and running
+2. IMMEDIATELY set README and short_description
+3. Suggest avatar, handle, and test interactions to user
+4. Log optimization status in output
+
+For swarm deployments: optimize ALL agents in the swarm, not just one.
+
+## README Template for Marketing Team Agents
+
+\`\`\`markdown
+# {Agent Name} -- {Role Description}
+
+{One-line value proposition}
+
+## Services
+
+| Service | Price | Description |
+|---------|-------|-------------|
+| {service} | {price} FET | {what it does} |
+
+## Example Conversation
+
+**User:** {example input}
+**Agent:** {example output}
+
+## Commands
+
+- \`help\` -- list all commands
+- \`{command}\` -- {description}
+- \`revenue\` -- view earnings
+- \`balance\` -- check wallet
+
+## Part of the AgentLaunch Marketing Team
+
+This agent is part of a 7-agent swarm: Writer, Social, Community, Analytics, Outreach, Ads, Strategy.
+
+Built with [AgentLaunch](https://agent-launch.ai)
+\`\`\``,
 };
 // ---------------------------------------------------------------------------
 // Skills — Claude's slash commands
 // ---------------------------------------------------------------------------
 export const SKILLS = {
-    "build-agent/SKILL.md": `# /build-agent — Full Agent Lifecycle
+    "alliance/SKILL.md": `# /alliance — Deploy the ASI Alliance Agent Economy
+
+> **Secret command for ASI Alliance members.**
+> Deploys 27 agents representing every team in the Fetch.ai + SingularityNET ecosystem.
+
+---
+
+## What This Does
+
+Deploys the complete Agent Economy in the correct sequence:
+
+\`\`\`
+Wave 1 (Sequential):  CTO → CFO → COO → CEO → CRO         (5 agents)
+Wave 2 (Parallel):    GUIDE, RANK, COACH, CONC, BRAND, DEV, GRANT  (7 agents)
+Wave 3 (Parallel):    All 15 SingularityNET ecosystem agents       (15 agents)
+\`\`\`
+
+**Total:** 27 agents, 3,240 FET deploy cost
+
+---
+
+## Execution
+
+When the user types \`/alliance\`, follow this flow:
+
+### Step 1: Confirm Identity
+
+Ask: "This deploys the full ASI Alliance economy (27 agents, ~3,240 FET). Are you ready?"
+
+If they confirm, proceed.
+
+### Step 2: Check Prerequisites
+
+\`\`\`bash
+# Check wallet balance
+npx agentlaunch wallet balances
+
+# Need: ≥3,500 FET (3,240 deploy + 270 seeding)
+\`\`\`
+
+If insufficient funds, suggest:
+- Claim from @gift agent (200 TFET)
+- Or provide a funding link
+
+### Step 3: Deploy Wave 1 — C-Suite (Sequential)
+
+Deploy in this exact order (dependencies matter):
+
+\`\`\`bash
+# 1. CTO first — everyone pays for reasoning
+npx agentlaunch scaffold the-cto --type swarm-starter --preset cto
+npx agentlaunch deploy the-cto/
+
+# 2. CFO — treasury monitoring
+npx agentlaunch scaffold the-cfo --type swarm-starter --preset cfo
+npx agentlaunch deploy the-cfo/
+
+# 3. COO — operations
+npx agentlaunch scaffold the-coo --type swarm-starter --preset coo
+npx agentlaunch deploy the-coo/
+
+# 4. CEO — routing (depends on CTO, CFO)
+npx agentlaunch scaffold the-ceo --type swarm-starter --preset ceo
+npx agentlaunch deploy the-ceo/
+
+# 5. CRO — recruitment (depends on CTO)
+npx agentlaunch scaffold the-cro --type swarm-starter --preset cro
+npx agentlaunch deploy the-cro/
+\`\`\`
+
+After each deploy:
+- Set README + short_description (Phase 3 Optimize)
+- Record agent address
+- Confirm running before next
+
+### Step 4: Deploy Wave 2 — Fetch Internal (Parallel)
+
+Deploy all 7 in parallel:
+
+| Agent | Preset | Description |
+|-------|--------|-------------|
+| The Tour Guide | \`guide\` | Ecosystem onboarding |
+| The SEO Manager | \`rank\` | Agent discoverability |
+| The Business Coach | \`coach\` | Startup guidance |
+| The Concierge | \`conc\` | Consumer experience |
+| The Brand Manager | \`brand\` | Brand partnerships |
+| The Dev Advocate | \`dev\` | Developer support |
+| The Grants Officer | \`grant\` | Funding guidance |
+
+### Step 5: Deploy Wave 3 — SingularityNET (Parallel)
+
+Deploy all 15:
+
+| Agent | Token | Organisation |
+|-------|-------|--------------|
+| The Marketplace Manager | $MKTPL | SingularityNET |
+| The IT Manager | $INFRA | CUDOS |
+| The Facilities Manager | $EDGE | NuNet |
+| The Finance Controller | $FLOW | HyperCycle |
+| The Health Coach | $LIFE | Rejuve.AI |
+| The Lab Manager | $LAB | Rejuve Biotech |
+| The Editor | $ED | Mindplex |
+| The Fund Manager | $FUND | SingularityDAO |
+| The Yield Farmer | $YIELD | Singularity Finance |
+| The Treasury Manager | $STABLE | Cogito Protocol |
+| The EA | $EA | Twin Protocol |
+| The A&R Manager | $AR | Jam Galaxy |
+| The Strategy Director | $STRAT | TrueAGI |
+| The Research Scientist | $PHD | Zarqa / OpenCog |
+| The Talent Agent | $TALENT | Yaya Labs |
+
+### Step 6: Seed Cross-Holdings
+
+After all agents deployed:
+
+\`\`\`bash
+# CEO buys infrastructure tokens
+npx agentlaunch buy $CTO_TOKEN --amount 5
+npx agentlaunch buy $CFO_TOKEN --amount 5
+npx agentlaunch buy $GUIDE_TOKEN --amount 2
+
+# COO buys monitoring dependencies
+npx agentlaunch buy $CEO_TOKEN --amount 3
+npx agentlaunch buy $CTO_TOKEN --amount 3
+
+# Coalition cross-holdings...
+\`\`\`
+
+### Step 7: Verify & Report
+
+\`\`\`bash
+npx agentlaunch status --swarm
+\`\`\`
+
+Show summary:
+\`\`\`
+╭────────────────────────────────────────────────────────────────╮
+│              ASI ALLIANCE ECONOMY DEPLOYED                     │
+├────────────────────────────────────────────────────────────────┤
+│  Agents:     27 running                                        │
+│  Tokens:     27 on bonding curves                              │
+│  GDP:        Tracking...                                       │
+│  Deploy:     3,240 FET                                         │
+│                                                                │
+│  C-Suite:    CEO, CTO, CFO, COO, CRO                          │
+│  Fetch:      GUIDE, RANK, COACH, CONC, BRAND, DEV, GRANT      │
+│  SNET:       MKTPL, INFRA, EDGE, FLOW, LIFE, LAB, ED,         │
+│              FUND, YIELD, STABLE, EA, AR, STRAT, PHD,         │
+│              TALENT                                            │
+│                                                                │
+│  Next: CRO will scout for more agents to recruit              │
+╰────────────────────────────────────────────────────────────────╯
+\`\`\`
+
+---
+
+## Named After Real People
+
+These agents are named after real ASI Alliance leadership:
+
+| Agent | Named After | Organisation |
+|-------|-------------|--------------|
+| The CEO ($CEO) | Humayun Sheikh | Fetch.ai / ASI Alliance |
+| The CTO ($CTO) | Thomas Hain | Fetch.ai (Co-Founder) |
+| The CRO ($CRO) | Sana Wajid | Fetch.ai Innovation Lab |
+| The Tour Guide ($GUIDE) | Maria Minaricova | Fetch.ai BD |
+| The Marketplace Manager ($MKTPL) | Ben Goertzel | SingularityNET |
+| The IT Manager ($INFRA) | Matt Hawkins | CUDOS |
+
+---
+
+## Related Docs
+
+- [people.md](docs/people.md) — Full team roster + agent mapping
+- [the-agent-economy.md](docs/the-agent-economy.md) — Complete vision document
+- [agent-coordination.md](docs/agent-coordination.md) — Pitch deck version
+
+---
+
+## Output Style
+
+Be elegant and helpful:
+- Show progress clearly
+- Celebrate each wave completion
+- Explain what each agent does as it deploys
+- End with clear next steps
+
+**This is the flagship deployment. Make it feel special.**`,
+    "build-agent/SKILL.md": `# /build-agent -- Full Agent Lifecycle
 
 Build, deploy, and tokenize an agent in one guided flow.
 
 ## Steps
 
-1. Ask user what agent they want (name, description, functionality)
-2. Generate agent.py code based on description
-3. Show code, let user request changes
-4. Deploy: \`agentlaunch deploy\`
-5. Tokenize: \`agentlaunch tokenize --agent <address> --name "Name" --symbol SYM\`
-6. Return handoff link for human to sign
+1. **Gather requirements**: Ask the user what kind of agent they want to build.
+   Get the agent name, ticker symbol, and description.
 
-## Platform Fees
-- Deploy: 120 FET (paid by human signer)
-- Trading: 2% to protocol treasury
-`,
-    "build-swarm/SKILL.md": `# /build-swarm — Deploy Agent Swarm
+2. **Choose a template**: Based on the description, pick one of:
+   custom, price-monitor, trading-bot, data-analyzer, research, gifter.
+   Confirm the choice with the user.
+
+3. **Scaffold the agent code**: Use the \`scaffold_agent\` MCP tool (or the
+   templates package directly) to generate agent code from the template.
+   Customize the generated code based on user requirements.
+
+4. **Review with user**: Show the generated code. Let them request changes.
+
+5. **Deploy to Agentverse**:
+   - Use the \`deploy_to_agentverse\` MCP tool
+   - Or manually: create agent, upload code (double-encoded JSON), set secrets, start
+   - Poll until compiled (60s timeout)
+   - Check logs for errors
+
+6. **Optimize** (MANDATORY -- Phase 3 of \`docs/workflow.md\`):
+   - Write a compelling README with: value proposition, services, example conversations, pricing
+   - Write a short_description (1-2 sentences for directory listing)
+   - Push via \`PUT /v1/hosting/agents/{addr}\` with \`{"readme": "...", "short_description": "..."}\`
+   - Suggest 3-5 @handle options to the user
+   - Tell user to run 3+ test interactions (Response QA Agent or chat directly)
+   - Suggest avatar if user has one
+
+7. **Tokenize on AgentLaunch**:
+   - Use the \`create_token_record\` MCP tool
+   - POST /agents/tokenize with name, symbol, description, chainId
+   - Default chain: BSC Testnet (97)
+
+8. **Return results to user**:
+   - Agent address (agent1q...)
+   - Token handoff link (\${AGENT_LAUNCH_FRONTEND_URL}/deploy/{tokenId})
+   - Instructions: "Click this link, connect your wallet, and sign to deploy"
 
-Scaffold, deploy, and tokenize a multi-agent swarm with the swarm-starter template.
+9. **Optionally show market data**: Use \`get_token\` to show token details
+   after deployment.
+
+## Environment
+
+- Reads AGENTVERSE_API_KEY from \`.env\`
+- Uses SDK/MCP tools for all API calls
+- Uses templates package for code generation
+- Dev URLs used by default; set AGENT_LAUNCH_ENV=production for mainnet
+
+## Platform Fee Note
+
+The 120 FET deployment fee is paid by the human who signs the transaction.
+The 2% trading fee goes 100% to protocol treasury (no creator fee).`,
+    "build-swarm/SKILL.md": `# /build-swarm -- Build an Agent Swarm
+
+Guided swarm creation experience for Claude Code.
 
 ## Steps
 
-1. Ask user what swarm they want (name, roles, purpose)
-2. Show the 7 available presets: writer, social, community, analytics, outreach, ads, strategy
-3. Let user pick roles (or suggest a starter configuration)
-4. For each role:
-   a. Scaffold from swarm-starter template with preset variables
-   b. Deploy to Agentverse
-   c. Tokenize on AgentLaunch
-5. Return handoff links for each agent
+1. **Understand the goal**: Ask what the user wants their swarm to do
+2. **Suggest presets**: Based on the goal, recommend a combination of presets
+   - For content marketing: Writer + Social + Analytics
+   - For community growth: Writer + Community + Social
+   - For full marketing team: All 7
+3. **Scaffold each agent**: Use \`scaffold_swarm\` MCP tool with appropriate preset
+4. **Customize business logic**: Let user review/edit the SwarmBusiness section
+5. **Deploy**: Use \`deploy_swarm\` MCP tool or deploy individually
+   - Follow build order: Level 0 (no deps) first, then Level 1, etc.
+   - Set peer address secrets so agents can call each other
+6. **Optimize EVERY agent** (MANDATORY -- Phase 3 of \`docs/workflow.md\`):
+   - For each deployed agent, set README and short_description
+   - Push via \`PUT /v1/hosting/agents/{addr}\` with \`{"readme": "...", "short_description": "..."}\`
+   - README must include: value proposition, services table, example conversations, pricing, commands
+   - Suggest @handle options for each agent
+   - Tell user to run 3+ test interactions per agent
+7. **Show status**: Display swarm health, addresses, optimization score, next steps
+
+## Preset Quick Reference
+
+| Preset | Role | Price/call | Best for |
+|--------|------|-----------|----------|
+| Writer | Content creation | 0.01 FET | Every swarm needs content |
+| Social | Twitter/X posting | 0.005 FET | Social media presence |
+| Community | Telegram management | 0.002 FET | Community engagement |
+| Analytics | Engagement tracking | 0.005 FET | Data-driven decisions |
+| Outreach | Partnership emails | 0.01 FET | Business development |
+| Ads | Ad campaigns | 0.01 FET | Paid marketing |
+| Strategy | Campaign coordination | 0.02 FET | Orchestrates all agents |
+
+## Recommended Starters
+
+- **Content only**: Writer (1 agent)
+- **Social presence**: Writer + Social (2 agents)
+- **Community**: Writer + Community + Social (3 agents)
+- **Full team**: All 7
+
+## Environment
+
+- Reads AGENTVERSE_API_KEY from .env
+- Uses MCP tools: scaffold_swarm, deploy_swarm, deploy_to_agentverse
+- Platform: agent-launch.ai`,
+    "deploy/SKILL.md": `# /deploy -- Deploy Agent to Agentverse
+
+Deploy a Python agent file to the Agentverse hosting platform.
+**Follows the full lifecycle workflow from \`docs/workflow.md\`.**
+
+## Usage
 
-## Starter Configurations
+\`\`\`
+/deploy [path/to/agent.py]
+\`\`\`
 
-- Content only: Writer (1 agent)
-- Social presence: Writer + Social (2 agents)
-- Community: Writer + Community + Social (3 agents)
-- Analytics stack: Writer + Social + Analytics (3 agents)
-- Full team: All 7
+If no path is given, look for \`agent.py\` in the current directory.
 
-## Platform Fees
-- Deploy: 120 FET per agent (paid by human signer)
-- Trading: 2% to protocol treasury (NO creator fee)
-`,
-    "deploy/SKILL.md": `# /deploy — Deploy to Agentverse
+## Steps
 
-Deploy agent.py to Agentverse hosting.
+1. **Read** the agent.py file
+2. **Validate** it follows uAgent patterns:
+   - Has \`Agent()\` with zero params
+   - Imports Chat Protocol v0.3.0
+   - Has \`ChatAcknowledgement\` handler
+   - Uses \`ctx.logger\` not \`print()\`
+3. **Create agent** on Agentverse via POST /v1/hosting/agents
+4. **Upload code** using double-encoded JSON format
+5. **Set secrets** from \`.env\` (AGENTVERSE_API_KEY at minimum)
+6. **Start agent** via POST /v1/hosting/agents/{addr}/start
+7. **Poll compilation** status (up to 60s)
+8. **Optimize** (MANDATORY -- see \`.claude/rules/workflow.md\`):
+   - Write a README based on the agent's capabilities (from code analysis)
+   - Write a short_description (1-2 sentences)
+   - Push both via \`PUT /v1/hosting/agents/{addr}\` with \`{"readme": "...", "short_description": "..."}\`
+   - Suggest @handle options to the user
+   - Tell user to test with 3+ interactions
+9. **Show results**: agent address, compilation status, optimization score, initial logs
+
+## Auth
+
+Uses AGENTVERSE_API_KEY from \`.env\` with header: \`Authorization: bearer <key>\``,
+    "grow/SKILL.md": `# /grow -- Autonomous Task Execution
+
+Claim and execute tasks from \`docs/TODO.md\` autonomously.
+
+## Behavior
+
+1. **Read TODO.md**: Open \`docs/TODO.md\` and parse the task tables
+2. **Find next task**: Look for tasks marked \`[ ]\` with no unfinished dependencies
+3. **Claim task**: Update the status marker from \`[ ]\` to \`[~]\` (in progress)
+4. **Execute task**: Follow the "How" column instructions using toolkit commands
+5. **Verify KPI**: Check the "KPI" column to confirm success
+6. **Complete task**: Update status from \`[~]\` to \`[x]\` (complete)
+7. **Report**: Summarize what was done and move to next task (if \`/grow N\`)
+
+## Arguments
+
+- \`/grow\` - Execute the next pending task
+- \`/grow 3\` - Execute 3 tasks sequentially
+- \`/grow L-1\` - Execute a specific task by ID
+- \`/grow status\` - Show current progress without executing
+
+## Task Selection Rules
+
+1. Never execute a task whose "Depends" column lists incomplete tasks
+2. Process tasks in ID order within each phase (L-1 before L-2)
+3. Process phases in order (Phase 1 before Phase 2)
+4. Skip tasks marked \`[!]\` (blocked) or \`[x]\` (complete)
+
+## Status Markers
+
+| Marker | Meaning |
+|--------|---------|
+| \`[ ]\`  | Pending - ready to execute (if deps are met) |
+| \`[~]\`  | In Progress - currently being worked on |
+| \`[x]\`  | Complete - verified and done |
+| \`[!]\`  | Blocked - external blocker, needs human intervention |
+
+## Example Task Execution
+
+For task L-1 "Deploy the swarm":
+
+\`\`\`markdown
+| Status | ID | Task | How | KPI | Depends |
+|:---:|:---|:---|:---|:---|:---|
+| \`[~]\` | L-1 | Deploy the swarm | \`npx agentlaunch create\` ... | All 7 running | — |
+\`\`\`
+
+1. Run \`npx agentlaunch create\` and follow prompts
+2. Verify all 7 agents are running with \`network_status\` MCP tool
+3. If KPI met, update to \`[x]\`
+4. If failed, update to \`[!]\` and report the blocker
+
+## Gate Checks
+
+After completing all tasks in a phase, verify the Phase Gate criteria.
+If all gate items pass, proceed to the next phase.
+If any gate item fails, report which ones need attention.
+
+## Error Handling
+
+- If a task fails, mark it \`[!]\` and add a comment explaining the blocker
+- If dependencies are missing, report which tasks need to complete first
+- If a gate check fails, report which criteria are unmet
+
+## Toolkit Commands Used
+
+- \`npx agentlaunch create\` - Deploy agents
+- \`npx agentlaunch status\` - Check agent/token status
+- \`network_status\` MCP tool - Swarm health check
+- \`check_agent_commerce\` MCP tool - Individual agent commerce
+- \`GET /agents/token/{addr}/holders\` - Token holder list`,
+    "improve.md": `# /improve - Capture Session Learnings
+
+Extract learnings from this session and create a PR to improve the toolkit.
+
+## Trigger
+
+User runs \`/improve\` after building agents, fixing errors, or discovering patterns.
+
+## Workflow
+
+### 1. Analyze Session
+
+Review the conversation to identify:
+
+- **Errors fixed**: Error messages encountered and how they were resolved
+- **Code patterns**: Working code that could become a template or snippet
+- **API discoveries**: Gotchas, undocumented behavior, correct endpoints
+- **Workflow improvements**: Better ways to do things
+
+### 2. Check for Duplicates
+
+Before suggesting additions, search existing rules:
 
-## Command
 \`\`\`bash
-agentlaunch deploy
+# Search for the error/pattern in existing rules
+grep -r "keyword" .claude/rules/
+grep -r "keyword" docs/
 \`\`\`
 
-## What It Does
-1. Reads agent.py
-2. Validates uAgent patterns
-3. Creates agent on Agentverse
-4. Uploads code
-5. Sets secrets from .env
-6. Starts agent
-7. Waits for compilation
-`,
-    "tokenize/SKILL.md": `# /tokenize — Create Token
+Skip anything already documented. Only surface NEW learnings.
 
-Create a tradeable token for an agent.
+### 3. Present Learnings
+
+Show the user what was found:
+
+\`\`\`
+Session Analysis
+================
+
+Errors Fixed:
+  1. [NEW] "agent.create_protocol() does not exist"
+     Fix: Use Protocol(spec=...) directly
+     → Suggest: Add to .claude/rules/uagent-patterns.md
+
+  2. [KNOWN] "datetime.utcnow() deprecated"
+     Already in: .claude/rules/uagent-patterns.md
+     → Skip
+
+Patterns Discovered:
+  1. [NEW] Twitter integration with rate limiting
+     → Suggest: New template twitter-bot/
+
+  2. [NEW] Conversation memory with ctx.storage
+     → Suggest: Add to .claude/rules/uagent-patterns.md
+
+Which learnings should we add? (comma-separated numbers, or 'all', or 'none')
+\`\`\`
+
+### 4. Create Branch & Changes
+
+For selected learnings:
 
-## Command
 \`\`\`bash
-agentlaunch tokenize --agent <address> --name "Name" --symbol SYM
+# Create improvement branch
+git checkout -b improve/session-$(date +%s)
+
+# Make changes based on learning type:
 \`\`\`
 
-## Result
-Returns a handoff link. Human clicks it, connects wallet, signs, token goes live.
-`,
-    "status/SKILL.md": `# /status — Check Status
+**For rule updates:**
+- Edit the relevant \`.claude/rules/*.md\` file
+- Add the pattern under the appropriate section
+- Include: what the error looks like, why it happens, how to fix it
 
-Check agent and token status.
+**For new templates:**
+- Create directory in \`packages/templates/src/templates/\`
+- Add agent.py with the pattern
+- Update templates index
+
+**For workflow improvements:**
+- Edit \`.claude/rules/workflow.md\` or relevant skill
+- Add the step or check
+
+### 5. Test Changes
+
+Before committing, validate everything still works:
 
-## Command
 \`\`\`bash
-agentlaunch status <address>
+# 1. Build all packages (catches TypeScript/import errors)
+npm run build
+
+# 2. Run test suite
+npm run test
+
+# 3. If new template added, validate it compiles
+if [ -d "packages/templates/src/templates/NEW_TEMPLATE" ]; then
+  python3 -m py_compile packages/templates/src/templates/NEW_TEMPLATE/agent.py
+fi
+
+# 4. Lint markdown (optional but recommended)
+npx markdownlint .claude/rules/*.md 2>/dev/null || true
 \`\`\`
 
-Shows: running status, price, holders, progress to graduation.
-`,
-    "market/SKILL.md": `# /market — Browse Tokens
+**If tests fail:**
+\`\`\`
+Test Results
+============
+✗ npm run build - FAILED
 
-Browse tokens on AgentLaunch.
+Error: packages/templates/src/index.ts(45,10):
+  Cannot find module './templates/twitter-writer'
+
+The new template needs to be registered in the index.
+Fix the error, then run /improve again.
+\`\`\`
+
+Do not create PR until all tests pass.
+
+**If tests pass:**
+\`\`\`
+Test Results
+============
+✓ npm run build - passed
+✓ npm run test - 47 passed, 0 failed
+✓ Template syntax - valid Python
+✓ Markdown lint - no errors
+
+All tests passed. Proceeding to commit...
+\`\`\`
+
+### 6. Commit & Push
 
-## Commands
 \`\`\`bash
-agentlaunch list
-agentlaunch status <token_address>
+git add .
+git commit -m "$(cat <<'EOF'
+improve: Add learnings from session
+
+Learnings added:
+- [list what was added]
+
+Session context:
+- [brief description of what user was building]
+
+EOF
+)"
+
+git push -u origin improve/session-$(date +%s)
 \`\`\`
-`,
+
+### 7. Create PR
+
+\`\`\`bash
+gh pr create --title "Improve: [brief summary]" --body "$(cat <<'EOF'
+## Session Learnings
+
+This PR captures learnings from a development session.
+
+### What was built
+[Description of what the user was working on]
+
+### Learnings added
+
+| Type | File | Description |
+|------|------|-------------|
+| Rule | \`.claude/rules/xxx.md\` | [what was added] |
+| Template | \`packages/templates/...\` | [what was added] |
+
+### Why these matter
+[Brief explanation of how these help future users]
+
+---
+*Generated by \`/improve\` - see docs/self-improvement.md*
+EOF
+)"
+\`\`\`
+
+### 8. Report to User
+
+\`\`\`
+Improvement PR created!
+
+Branch: improve/session-1709654321
+PR: https://github.com/[org]/agent-launch-toolkit/pull/42
+
+Changes:
+  - .claude/rules/uagent-patterns.md: Added Protocol() pattern
+  - packages/templates/src/templates/twitter-bot/: New template
+
+Next steps:
+  1. Review the PR
+  2. Team discusses if needed
+  3. Merge when ready
+
+Thanks for improving the toolkit!
+\`\`\`
+
+## Learning Categories
+
+### Error Patterns (→ Rules)
+
+Format for error rules:
+
+\`\`\`markdown
+## Error: "[exact error message]"
+
+**Cause:** [why this happens]
+
+**Fix:** [what to do instead]
+
+**Example:**
+\`\`\`python
+# Wrong
+ctx.wallet.address()
+
+# Right
+agent.wallet.address()
+\`\`\`
+\`\`\`
+
+### API Gotchas (→ Rules)
+
+Format for API rules:
+
+\`\`\`markdown
+## [API/Endpoint Name]
+
+**Gotcha:** [the unexpected behavior]
+
+**Correct usage:**
+\`\`\`python
+# code example
+\`\`\`
+
+**Why:** [explanation]
+\`\`\`
+
+### Code Patterns (→ Templates or Rules)
+
+If it's a full agent pattern → new template
+If it's a snippet → add to relevant rule file
+
+### Workflow Improvements (→ Rules or Skills)
+
+If it's a step people should always do → add to workflow.md
+If it's a new capability → create/update skill
+
+## Edge Cases
+
+### No learnings found
+
+\`\`\`
+No new learnings detected in this session.
+
+Everything worked smoothly, or the patterns are already documented.
+
+If you think something should be added, describe it and I'll help
+create the improvement manually.
+\`\`\`
+
+### User declines all
+
+\`\`\`
+No learnings selected. That's fine!
+
+If you change your mind, run /improve again or manually edit:
+  - .claude/rules/ for coding patterns
+  - packages/templates/ for agent templates
+  - .claude/skills/ for workflow improvements
+\`\`\`
+
+### Conflict with existing rules
+
+If a learning contradicts an existing rule:
+
+\`\`\`
+Potential conflict detected:
+
+Existing rule (.claude/rules/uagent-patterns.md):
+  "Use datetime.now() instead of utcnow()"
+
+Your session used:
+  datetime.utcnow() successfully
+
+Options:
+  1. Skip (keep existing rule)
+  2. Update rule with exception/context
+  3. Replace rule with new pattern
+
+Which approach?
+\`\`\`
+
+## Examples
+
+### Example 1: Error Pattern
+
+Session: User got "Protocol() missing role parameter" error
+
+\`\`\`
+Learning detected:
+
+Error: "Protocol() takes 1 positional argument but 2 were given"
+       when using payment_protocol_spec
+
+Fix: Payment protocol requires role="seller" or role="buyer"
+
+Add to .claude/rules/payment-protocol.md? [y/n]
+\`\`\`
+
+### Example 2: New Template
+
+Session: User built a Twitter bot with rate limiting
+
+\`\`\`
+Learning detected:
+
+Pattern: Twitter bot with:
+  - Rate limiting (15 calls/15min)
+  - Retry with exponential backoff
+  - Tweet queue with ctx.storage
+
+Create new template 'twitter-bot'? [y/n]
+\`\`\`
+
+### Example 3: Workflow Improvement
+
+Session: User's agent failed because they tokenized before optimizing
+
+\`\`\`
+Learning detected:
+
+Workflow issue: Tokenizing before optimization leads to poor
+discoverability. README should be set BEFORE tokenize.
+
+Update .claude/rules/workflow.md to enforce order? [y/n]
+\`\`\`
+
+## Files Modified by /improve
+
+| Learning Type | Target File |
+|---------------|-------------|
+| uAgent errors | \`.claude/rules/uagent-patterns.md\` |
+| Payment issues | \`.claude/rules/payment-protocol.md\` |
+| API path errors | \`.claude/rules/api-paths.md\` |
+| Agentverse issues | \`.claude/rules/agentverse.md\` |
+| Workflow problems | \`.claude/rules/workflow.md\` |
+| New agent pattern | \`packages/templates/src/templates/[name]/\` |
+| Swarm patterns | \`.claude/rules/marketing-swarm.md\` |
+| Consumer payments | \`.claude/rules/consumer-payments.md\` |
+
+## See Also
+
+- \`docs/self-improvement.md\` - Full documentation of this system
+- \`.claude/rules/\` - All current rules
+- \`packages/templates/\` - All current templates`,
+    "market/SKILL.md": `# /market -- Browse Tokens and Prices
+
+Browse tokens on AgentLaunch and check market data.
+
+## Usage
+
+\`\`\`
+/market [token_address]
+\`\`\`
+
+## Steps
+
+1. **If no address given**: Use \`list_tokens\` MCP tool to show trending tokens
+   with name, symbol, price, and progress toward graduation.
+2. **If address given**: Use \`get_token\` MCP tool for full token details
+   including price, market cap, holder count, and bonding curve progress.
+3. **Price preview**: Use \`calculate_buy\` or \`calculate_sell\` to show
+   what a specific FET amount would buy/sell.
+4. **Generate trade link**: Use \`get_trade_link\` to create a pre-filled
+   buy or sell URL the user can share.
+
+## Notes
+
+- Graduation happens at 30,000 FET liquidity (auto DEX listing).
+- Trading fee: 2% per trade, 100% to protocol treasury (no creator fee).`,
+    "status/SKILL.md": `# /status -- Check Agent and Token Status
+
+Check the deployment status of agents and tokens.
+
+## Usage
+
+\`\`\`
+/status [agent_address_or_token_address]
+\`\`\`
+
+## Steps
+
+1. **Detect address type**:
+   - \`agent1q...\` -> Agentverse agent, check via GET /v1/hosting/agents/{addr}
+   - \`0x...\` -> Token address, check via \`get_token\` MCP tool
+2. **For agents**: Show name, running status, compiled status, wallet address.
+   Optionally fetch recent logs via GET /v1/hosting/agents/{addr}/logs.
+3. **For tokens**: Show name, symbol, price, market cap, holder count,
+   bonding curve progress, and whether it has graduated to DEX.
+4. **Platform overview**: If no address given, use \`get_platform_stats\`
+   to show overall platform statistics (total tokens, volume, etc.).
+
+## Notes
+
+- Agent compilation takes 15-60s after starting.
+- Token graduation occurs at 30,000 FET liquidity.`,
+    "todo/SKILL.md": `# /todo -- Create TODO from Document
+
+Transform a strategy document, roadmap, or feature spec into a structured TODO.md.
+
+## Usage
+
+- \`/todo\` - Create TODO.md from the most recent doc discussed
+- \`/todo docs/organic-growth-strategy.md\` - Create TODO from a specific file
+- \`/todo "my feature spec"\` - Create TODO from inline requirements
+
+## Template
+
+Use \`docs/TODO-template.md\` as the format reference. Key elements:
+
+### YAML Frontmatter
+
+\`\`\`yaml
+---
+title: Feature Name
+type: roadmap
+version: 1.0.0
+priority: Phase order
+total_tasks: N
+completed: 0
+status: PENDING
+depends_on: any prerequisites
+---
+\`\`\`
+
+### Task Tables
+
+\`\`\`markdown
+| Status | ID | Task | How | KPI | Depends |
+|:---:|:---|:---|:---|:---|:---|
+| \`[ ]\` | L-1 | Task name | Implementation details | Success metric | — |
+| \`[ ]\` | L-2 | Next task | How to do it | What proves it's done | L-1 |
+\`\`\`
+
+### Status Markers
+
+- \`[ ]\` - Pending
+- \`[~]\` - In Progress
+- \`[x]\` - Complete
+- \`[!]\` - Blocked
+
+### Phase Structure
+
+Group tasks into phases with clear progression:
+
+1. **Phase 1: Setup** - Foundation tasks
+2. **Phase 2: Build** - Core implementation
+3. **Phase 3: Test** - Verification
+4. **Phase 4: Deploy** - Launch tasks
+
+### Gate Criteria
+
+After each phase, add a gate checklist:
+
+\`\`\`markdown
+### Phase N Gate
+
+\`\`\`
+  [ ] First success criterion
+  [ ] Second success criterion
+  [ ] Third success criterion
+\`\`\`
+\`\`\`
+
+### Dependency Graph
+
+Add ASCII or Mermaid diagram showing task dependencies:
+
+\`\`\`markdown
+## Dependency Graph
+
+\`\`\`
+L-1 ──► L-2 ──► L-3 ──► P-1
+         │              │
+         └──► L-4 ──────┘
+\`\`\`
+\`\`\`
+
+### Progress Overview
+
+Include visual progress tracking:
+
+\`\`\`markdown
+## Progress Overview
+
+\`\`\`
+╭─────────────────────────────────────────────╮
+│   Phase 1: Setup  [░░░░░░░░░░]  0/4    0%  │
+│   Phase 2: Build  [░░░░░░░░░░]  0/6    0%  │
+│   TOTAL           [░░░░░░░░░░]  0/10   0%  │
+╰─────────────────────────────────────────────╯
+\`\`\`
+\`\`\`
+
+## Output
+
+By default, write to \`docs/TODO.md\`.
+Use \`/todo output:path/to/file.md\` to specify a different output.
+
+## Task ID Conventions
+
+Use short, prefixed IDs:
+
+- \`L-N\` - Launch/setup tasks
+- \`B-N\` - Build tasks
+- \`T-N\` - Test tasks
+- \`D-N\` - Deploy tasks
+- \`P-N\` - Prove/validate tasks
+- \`G-N\` - Growth tasks
+- \`S-N\` - Sustain tasks
+- \`F-N\` - Final/graduation tasks
+
+## KPI Guidelines
+
+KPIs should be:
+- Measurable ("All 7 running" not "it works")
+- Specific ("≥10 queries/day" not "many queries")
+- Verifiable (can check with a command or tool)
+
+## Integration with /grow
+
+The TODO created by \`/todo\` is designed to work with \`/grow\`:
+
+1. \`/todo docs/plan.md\` - Create the TODO
+2. \`/grow\` - Execute tasks one by one
+3. \`/grow 5\` - Execute 5 tasks in sequence`,
+    "tokenize/SKILL.md": `# /tokenize -- Tokenize an Agent
+
+Create a tradeable token for an existing Agentverse agent.
+
+## Usage
+
+\`\`\`
+/tokenize [agent_address] --name "Name" --symbol "SYM"
+\`\`\`
+
+## Steps
+
+1. **Get agent address**: If not provided, list the user's agents via
+   GET /v1/hosting/agents and let them choose.
+2. **Collect details**: name, symbol (ticker), description. Prompt for
+   any missing values.
+3. **Create token record**: Use the \`create_token_record\` MCP tool or
+   POST /agents/tokenize with:
+   - agentAddress, name, symbol, description
+   - chainId (default: 97 for BSC Testnet)
+4. **Return handoff link**: Show the deploy link and instructions.
+5. **Explain next steps**:
+   - Human clicks the link
+   - Connects wallet (MetaMask, etc.)
+   - Signs transaction (pays 120 FET deployment fee)
+   - Token goes live on the bonding curve
+
+## Notes
+
+- The 120 FET deploy fee is paid by the human signer, not the agent.
+- Trading fee: 2% per trade, 100% to protocol treasury (no creator fee).
+- Token graduates to DEX at 30,000 FET liquidity.`,
+    "welcome/SKILL.md": `# Welcome — Show Status & Choose Your Path
+
+> Run automatically when Claude Code starts, or manually with \`/welcome\`
+
+---
+
+## What This Does
+
+1. **Shows everything the user already has** — agents, tokens, wallet, swarm status
+2. **Offers three paths** — single agent, marketing team, or alliance swarm
+3. **Guides them to the right choice**
+
+---
+
+## Execution
+
+### Step 1: Gather Current Status
+
+Run these checks silently:
+
+\`\`\`bash
+# Check wallet
+npx agentlaunch wallet balances 2>/dev/null || echo "No wallet configured"
+
+# List their agents (from Agentverse)
+npx agentlaunch list --mine 2>/dev/null || echo "No agents yet"
+
+# Check if .env exists
+test -f .env && echo "API key configured" || echo "No .env file"
+\`\`\`
+
+### Step 2: Display Status
+
+Show a clear summary:
+
+\`\`\`
+╭────────────────────────────────────────────────────────────────╮
+│                   AGENT LAUNCH TOOLKIT                         │
+│                   agent-launch.ai                              │
+├────────────────────────────────────────────────────────────────┤
+│                                                                │
+│  YOUR STATUS                                                   │
+│  ───────────                                                   │
+│  Wallet:     0x1234...5678 (150.5 FET, 0.02 BNB)              │
+│  Agents:     3 on Agentverse                                   │
+│  Tokens:     1 deployed ($MYAGENT at 1,200 FET raised)        │
+│  Swarm:      None active                                       │
+│                                                                │
+│  RECENT ACTIVITY                                               │
+│  ───────────────                                               │
+│  • joke-teller: 45 queries today, 0.9 FET earned              │
+│  • price-monitor: Running, last alert 2h ago                   │
+│                                                                │
+╰────────────────────────────────────────────────────────────────╯
+\`\`\`
+
+If no agents/wallet yet:
+\`\`\`
+╭────────────────────────────────────────────────────────────────╮
+│  YOUR STATUS                                                   │
+│  ───────────                                                   │
+│  Wallet:     Not configured                                    │
+│  Agents:     None yet                                          │
+│  API Key:    Not set                                           │
+│                                                                │
+│  Let's get you started!                                        │
+╰────────────────────────────────────────────────────────────────╯
+\`\`\`
+
+### Step 3: Show the Three Paths
+
+\`\`\`
+╭────────────────────────────────────────────────────────────────╮
+│                   WHAT DO YOU WANT TO BUILD?                   │
+├────────────────────────────────────────────────────────────────┤
+│                                                                │
+│  1. SINGLE AGENT                                               │
+│     Chat + memory, deployed in 90 seconds                      │
+│     Cost: 120 FET to tokenize                                  │
+│     Best for: Getting started, specific use cases              │
+│     Command: npx agentlaunch                                   │
+│                                                                │
+│  2. MARKETING TEAM                                             │
+│     7 agents that grow your ecosystem                          │
+│     Writer, Social, Community, Analytics, Outreach, Ads,       │
+│     Strategy — all working together                            │
+│     Cost: 840 FET (7 × 120)                                    │
+│     Best for: Content, community, growth                       │
+│     Command: /build-swarm marketing                            │
+│                                                                │
+│  3. ALLIANCE SWARM (ASI Alliance Members)                      │
+│     27 agents for the full Fetch.ai + SingularityNET ecosystem│
+│     CEO, CTO, CFO, COO, CRO + 22 specialists                  │
+│     Cost: 3,240 FET (27 × 120)                                │
+│     Best for: ASI Alliance ecosystem builders                  │
+│     Command: /alliance                                         │
+│                                                                │
+├────────────────────────────────────────────────────────────────┤
+│  Type 1, 2, or 3 — or describe what you want to build         │
+╰────────────────────────────────────────────────────────────────╯
+\`\`\`
+
+### Step 4: Handle Choice
+
+**If they choose 1 (Single Agent):**
+\`\`\`bash
+npx agentlaunch
+\`\`\`
+This runs the interactive flow: name → deploy → open editor
+
+**If they choose 2 (Marketing Team):**
+Invoke \`/build-swarm\` with marketing preset
+
+**If they choose 3 (Alliance):**
+Invoke \`/alliance\` skill
+
+**If they describe something else:**
+Use your judgment to scaffold the right template
+
+---
+
+## First-Time Setup
+
+If no .env or API key:
+
+\`\`\`
+╭────────────────────────────────────────────────────────────────╮
+│  FIRST TIME SETUP                                              │
+├────────────────────────────────────────────────────────────────┤
+│                                                                │
+│  1. Get your Agentverse API key:                               │
+│     https://agentverse.ai/settings/api-keys                   │
+│                                                                │
+│  2. Set it up:                                                 │
+│     npx agentlaunch init                                       │
+│                                                                │
+│  3. Get testnet tokens (free):                                 │
+│     Message @gift on Agentverse: "claim 0x<your-wallet>"      │
+│     You'll receive 200 TFET + 0.001 tBNB instantly            │
+│                                                                │
+╰────────────────────────────────────────────────────────────────╯
+\`\`\`
+
+---
+
+## Tone
+
+- **Welcoming** — Make them feel at home
+- **Clear** — Show exactly what they have and what's possible
+- **Helpful** — Guide them to the right choice
+- **Elegant** — Clean formatting, no clutter
+
+This is their first impression. Make it great.`,
 };
 // ---------------------------------------------------------------------------
 // Docs — Full documentation