pentagonal（pentagonal 搜索）

Pentagonal Clawd — Smart Contract Forge & Token Intelligence

You have access to Pentagonal, a sovereign smart contract creation, security, and token intelligence platform. It generates production-quality contracts, runs 8-agent security pen tests, auto-fixes vulnerabilities, compiles to deployment-ready ABI + bytecode, and fetches comprehensive live market and security data for any token.

Operation Mode

This skill operates in two modes. Check which tools are available and use the appropriate mode:

Mode A — MCP Tools (preferred)

If pentagonal_lookup, pentagonal_audit, etc. are available as registered tools, use them directly:

Mode B — Direct API (when MCP is not connected)

If MCP tools are not available, use bash to call the Pentagonal API directly:

Token lookup — public, no auth required (rate limited: 1 request/minute):

curl -s -X POST https://www.pentagonal.ai/api/fetch-contract \
  -H "Content-Type: application/json" \
  -d '{"address":"<CONTRACT_ADDRESS>","chainId":"<CHAIN>"}'

Token lookup — with API key (no rate limit):

curl -s -X POST https://www.pentagonal.ai/api/fetch-contract \
  -H "Content-Type: application/json" \
  -H "x-pentagonal-api-key: $PENTAGONAL_API_KEY" \
  -d '{"address":"<CONTRACT_ADDRESS>","chainId":"<CHAIN>"}'

Supported chainId values: ethereum, polygon, base, arbitrum, optimism, bsc, solana

Rate limit handling: If you receive a 429, read the retryAfter field and wait that many seconds before retrying. The error message will say exactly how long: "Rate limited. Please wait 47 more seconds before trying again."

# Parse retryAfter and wait automatically
RESPONSE=$(curl -s -X POST https://www.pentagonal.ai/api/fetch-contract \
  -H "Content-Type: application/json" \
  -d '{"address":"0x...","chainId":"ethereum"}')
RETRY=$(echo "$RESPONSE" | python3 -c "import sys,json; d=json.load(sys.stdin); print(d.get('retryAfter',0))" 2>/dev/null)
if [ "${RETRY:-0}" -gt 0 ] 2>/dev/null; then sleep "$RETRY"; fi

Workflows

Workflow A: Research an Existing Token

Use when the user asks about a token, coin, or contract address.

1. LOOKUP  →  2. AUDIT if user wants security analysis

Mode A (MCP):

pentagonal_lookup(address, chain, fields=["all"])
  → if verified source code returned:
      pentagonal_audit(code=<source from lookup>, chain=<chain>)

Mode B (bash):

curl -s -X POST https://www.pentagonal.ai/api/fetch-contract \
  -H "Content-Type: application/json" \
  -d '{"address":"0x...","chainId":"ethereum"}' | python3 -m json.tool

Field filtering — use fields param in Mode A to keep responses focused:

• Market data only: fields: ["price", "market"]
• Security check only: fields: ["security"]
• Source code for audit: fields: ["code"]

Workflow B: Create a New Contract

Use when the user asks to build, generate, or write a smart contract.

1. GENERATE  →  2. AUDIT  →  3. FIX (if needed)  →  4. RE-AUDIT  →  5. COMPILE

Mode A (MCP):

1. pentagonal_generate(prompt, chain, use_learned_rules=true)
2. pentagonal_audit(code, chain, use_learned_rules=true) — group by Critical → High → Medium → Low
3. pentagonal_fix(code, finding_title, finding_description) — one call per finding, critical first
4. Re-audit to confirm resolution
5. pentagonal_compile(code) — present ABI, bytecode, constructor args

Mode B (bash): Inform the user that contract generation and auditing require the MCP server or a Pentagonal account at pentagonal.ai. The public API only supports token lookups via bash.

pentagonal_lookup — Field Reference (Mode A)

Chain Selection Guide

Important Rules

1. NEVER skip the audit step when creating contracts — every generated contract must be audited before presenting to the user
2. ALWAYS fix critical and high findings before compiling
3. Use pentagonal_lookup first when researching existing tokens — don't speculate about market data
4. Default to use_learned_rules: true — the self-learning system is Pentagonal's core advantage
5. NEVER handle private keys — output deployment commands/scripts instead
6. Present findings clearly — group by severity, include line numbers, explain the risk
7. For Solana, always clarify: program (Anchor/Rust) or token (SPL config)?
8. Respect rate limits — if you hit a 429, read the countdown and wait. Do not retry immediately.

Self-Learning System

Pentagonal accumulates security knowledge with every audit. Each audit extracts generalized security rules from findings and injects them into future prompts. Always keep use_learned_rules: true.

Use pentagonal_rules (Mode A) to inspect the current knowledge base.

Installation

Claude Code / OpenClaw (CLI):

# Global install (available in all sessions)
mkdir -p ~/.claude/skills/pentagonal-clawd
cp SKILL.md ~/.claude/skills/pentagonal-clawd/
cp -r references ~/.claude/skills/pentagonal-clawd/

# Project-level
mkdir -p .claude/skills/pentagonal-clawd
cp SKILL.md .claude/skills/pentagonal-clawd/

Invoke with /pentagonal-clawd in your terminal session.

Claude.ai: Upload pentagonal-clawd-skill.zip via Settings → Customize → Skills.

Deployment & Examples

See references/deployment.md for deployment commands (Foundry, Hardhat, Anchor) and references/examples.md for conversation flow examples.