example-counter

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires generating runnable files that embed plaintext secrets/credentials (e.g., the genesis wallet seed, APP__INFRA__SECRET hex, and various storage/password env values) directly into code and Docker env/config, forcing the model to output secret values verbatim.

---

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most listed endpoints are local services or official midnight.network API endpoints (low risk), but the direct GitHub releases link to compact-installer.sh combined with the provided curl | sh install instruction is a risky downloadable executable pattern that could deliver malware unless you verify the GitHub repo and inspect the script first.

---

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The prerequisites explicitly instruct running curl https://github.com/midnightntwrk/compact/releases/latest/download/compact-installer.sh | sh which fetches and immediately executes remote install script (the Compact compiler) as a required dependency for the skill, creating a clear remote-code-execution risk.

---

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I scanned the skill for literal values that look like real credentials.

Flagged:

• APP__INFRA__SECRET = "303132333435363738393031323334353637383930313233343536373839303132"
• This exact hex string appears in the root compose.yml (indexer environment) and in counter-cli/standalone.env.example. It is presented as an infrastructure secret (APP__INFRA__SECRET) — a literal secret value used by the service — so it meets the criteria for a secret (a concrete credential value present in the docs).

Ignored (not flagged) and why:

• APP__INFRA__STORAGE__PASSWORD, APP__INFRA__PUB_SUB__PASSWORD, APP__INFRA__LEDGER_STATE_STORAGE__PASSWORD = "indexer"
• Simple, low-entropy setup password / demo value. Per rules, simple/setup passwords are ignored.
• GENESIS_MINT_WALLET_SEED = "0000000000000000000000000000000000000000000000000000000000000001"
• Known/test genesis seed used for standalone mode; a test/demo value (explicitly documented as the genesis wallet). Treated as non-sensitive per the "Test Agent / Quick Start" guidance.
• SIDECHAIN_BLOCK_BENEFICIARY = "04bcf7ad3be7a5c7..."
• This is a public key / beneficiary address (public data), not a secret.

---

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). This skill is explicitly a blockchain/web3 CLI and DApp template that creates and manages wallets, derives keys, signs transactions, constructs/balances transactions, and submits them to a Midnight network. It contains concrete crypto APIs and calls: HDWallet/seed derivation, ledger.ZswapSecretKeys/DustSecretKey, createKeystore/unshieldedKeystore.signData, signTransactionIntents, wallet.balanceUnboundTransaction / wallet.finalizeRecipe / wallet.submitTransaction, deployContract/findDeployedContract, and increment via counterContract.callTx.increment(). It also includes faucet/genesis mint usage and functions to register UTXOs for DUST generation. These are direct crypto/blockchain wallet and transaction execution capabilities (signing and sending funds/txs), which match the "Crypto/Blockchain (Wallets, Swaps, Signing)" criterion for Direct Financial Execution.