Add vigil: Onchain security scanner for DeFi traders on Base

vigil/SKILL.md

@@ -0,0 +1,166 @@
+---
+name: VIGIL Security Scanner
+description: Onchain security scanner on Base — scan token approvals, detect honeypots, analyze contracts for rugpull indicators, and score contract safety. Keyless read-only scanning via VIGIL API. Revoke actions require Bankr auth and are gated separately.
+var: ""
+tags: [crypto, security, base, defi]
+capabilities: [external_api, sends_notifications]
+---
+> **${var}** — Wallet address (`0x...`) or token contract address on Base to scan. Required. If empty, log `VIGIL_NO_TARGET` and exit cleanly (no notify).
+
+VIGIL is an onchain security scanner for DeFi traders on Base. It provides nine read-only scanning tools and one write action (revoke) that requires explicit Bankr authentication.
+
+**Read-only tools (this skill):**
+1. Approval Scanner — list all ERC-20/ERC-721 approvals, flag unlimited allowances
+2. Token Scanner — analyze contracts for rugpull indicators (hidden mint, proxy, tax manipulation, blacklist)
+3. Honeypot Detector — simulate buy/sell to detect trap tokens
+4. Safety Score — 0-100 composite rating based on code, ownership, liquidity, holders
+5. Wallet Report — full security posture assessment
+6. Wallet Monitor — real-time alerts for new approvals, risky interactions, and balance changes
+7. Token Market — price, liquidity, 24h volume, and pool age via DexScreener (no API key)
+8. Deployer Check — contract verification, name, and deployer reputation via Basescan
+9. Batch Scan — score multiple tokens in one call, ranked by risk
+
+**Write action (separate, not included here):**
+10. Approval Revoker — revoke dangerous approvals via Bankr transaction signing. This is a state-changing onchain transaction and is NOT part of this read-only skill.
+
+Read the last 2 days of `memory/logs/` so a repeat scan can note newly-granted or newly-revoked approvals.
+
+## Config
+
+- Target = `${var}`. Can be a wallet address or token contract address.
+- Chain = Base (`chainid=8453`, explorer `basescan.org`).
+- VIGIL API: `https://mcp.vigil.codes` (HTTPS, SSE transport)
+- GitHub: `https://github.com/vigilcodes/vigil-mcp`
+
+## Steps
+
+### 1. Validate input (strict — rejects injection)
+
+The target MUST be exactly `0x` followed by 40 hex characters. The regex
+below rejects any input containing quotes, spaces, or shell metacharacters,
+so it is safe to interpolate into the JSON payloads in later steps. Reject
+anything else and exit cleanly.
+
+```bash
+TARGET="${var}"
+
+# Strict allowlist: 0x + exactly 40 hex chars. Nothing else can pass.
+if ! printf '%s' "$TARGET" | grep -qiE '^0x[0-9a-f]{40}$'; then
+  echo "VIGIL_INVALID_TARGET: not a valid 0x address"
+  exit 0
+fi
+
+# Normalise to lowercase for consistent calls.
+TARGET="$(printf '%s' "$TARGET" | tr '[:upper:]' '[:lower:]')"
+```
+
+Because `$TARGET` is now guaranteed to match `^0x[0-9a-f]{40}$`, it contains
+no characters that could break the JSON body or the shell. A single address
+can be either a wallet or a token contract, so run the relevant tools and
+read each tool's own result — do not assume a type up front.
+
+### 1b. Safe call helper (checks errors before reading results)
+
+Every step below uses this helper. It fails loudly on a non-200 HTTP status
+or a JSON-RPC `error` body instead of silently passing `null` to `jq`, so a
+broken call is never reported as a clean scan.
+
+```bash
+VIGIL_API="https://mcp.vigil.codes/tools/call"
+
+vigil_call () {
+  # $1 = tool name, $2 = JSON arguments object
+  local name="$1" args="$2" body http code
+  body=$(curl -m 30 -s -w '\n%{http_code}' "$VIGIL_API" \
+    -H "Content-Type: application/json" \
+    -d "{\"jsonrpc\":\"2.0\",\"id\":1,\"method\":\"tools/call\",\"params\":{\"name\":\"$name\",\"arguments\":$args}}")
+  code=$(printf '%s' "$body" | tail -n1)
+  http=$(printf '%s' "$body" | sed '$d')
+
+  if [ "$code" != "200" ]; then
+    echo "VIGIL_HTTP_ERROR ($code) calling $name"; return 1
+  fi
+  if printf '%s' "$http" | jq -e '.error' >/dev/null 2>&1; then
+    echo "VIGIL_RPC_ERROR: $(printf '%s' "$http" | jq -c '.error')"; return 1
+  fi
+  printf '%s' "$http" | jq '.result'
+}
+```
+
+### 2. Scan approvals (wallet)
+
+```bash
+vigil_call scan_approvals '{"wallet": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 3. Scan token safety
+
+```bash
+vigil_call scan_token '{"token": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 4. Check honeypot
+
+```bash
+vigil_call detect_honeypot '{"token": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 5. Get safety score
+
+```bash
+vigil_call safety_score '{"contract": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 6. Generate wallet report
+
+```bash
+vigil_call wallet_report '{"wallet": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 7. Monitor wallet (real-time alerts)
+
+```bash
+vigil_call monitor_wallet '{"wallet": "'"$TARGET"'", "chain": "base", "lookback_blocks": 1000}'
+```
+
+### 8. Token market context (price + liquidity)
+
+```bash
+vigil_call token_market '{"token": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 9. Deployer reputation (verification + age)
+
+```bash
+vigil_call deployer_check '{"contract": "'"$TARGET"'", "chain": "base"}'
+```
+
+### 10. Batch scan multiple tokens
+
+```bash
+vigil_call batch_scan '{"tokens": ["'"$TARGET"'"], "chain": "base"}'
+```
+
+## Output Format
+
+VIGIL returns JSON with:
+
+- `approvals` — list of token approvals with risk levels
+- `safety_score` — 0-100 composite rating
+- `honeypot` — boolean + reason if detected
+- `rugpull_indicators` — list of suspicious patterns found
+- `recommendations` — action items
+
+## Risk Levels
+
+| Level | Icon | Meaning |
+|-------|------|---------|
+| CRITICAL | 🔴 | Active threat — revoke immediately |
+| HIGH | 🟠 | Dangerous pattern — likely exploit vector |
+| MEDIUM | 🟡 | Suspicious — proceed with caution |
+| LOW | 🟢 | Minor concern — monitor |
+| SAFE | ✅ | No issues detected |
+
+## Important: Revocation is NOT included
+
+The Approval Revoker tool performs state-changing onchain transactions via Bankr. It is intentionally excluded from this read-only skill. To revoke approvals, use the separate `vigil-revoke` skill (requires `BANKR_API_KEY` and explicit user confirmation).