Back to skill

Security audit

Cournot

Security checks for vulnerabilities and agentic risk

Overview

The skill’s Cournot queries, account management, and optional wallet payments fit its stated purpose and include clear user controls.

Your event text goes to Cournot, and configured prepaid calls may be consumed by queries. Review the network, asset, amount, recipient, and approval requirement before confirming payments; mainnet uses real funds and packs are non-refundable. Protect the local plaintext key files and never share wallet secrets in chat. Review the optional wallet dependency separately before installing it. This review covered all bundled files, but did not test live payments or inspect external wallet software.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (25)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The command installs a separate Binance wallet skill rather than querying Cournot or managing its prepaid usage and credentials. Nothing in the supplied chunk establishes this installation as a supporting implementation detail for Cournot. The chunk does not itself demonstrate wallet transactions or access to funds.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

md
Use [errors.md](references/errors.md) to interpret failures and uncertain outcomes under these same principles. Flow references define business facts and permit

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 60)May include surrounding context.

md
| `divergence [count]` (first word only) | Cached pricing-divergence ranking — [divergence-flow.md](references/divergence-flow.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 70)May include surrounding context.

md
ue lists, cards, code blocks or browser/file previews. See [divergence-flow.md](references/divergence-flow.md) for navigation and verification.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 72)May include surrounding context.

md
llow the output contract and row-verification procedure in [divergence-flow.md](references/divergence-flow.md).

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 90)May include surrounding context.

md
e generated data rows and verify the draft as described in [divergence-flow.md](references/divergence-flow.md). A changed or missing URL is an incorrect result;

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 61)May include surrounding context.

md
| `balance` | Read balance and masked key — [account.md](references/account.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 62)May include surrounding context.

md
| `key` | Show the full key, or recover it using the wallet — [account.md](references/account.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 63)May include surrounding context.

md
| `import <key>` | Validate and save an existing key — [account.md](references/account.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 64)May include surrounding context.

md
| `rotate` | Preview rotation and its effect on the old key; execute only after confirmation — [account.md](references/account.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 65)May include surrounding context.

md
| `topup` | Let the user select a pack, preview and confirm payment — [payment.md](references/payment.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 86)May include surrounding context.

md
lid precheck rejection, within the original confirmed payment; see [payment.md](references/payment.md). No other automatic retries, background queries, automati

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 66)May include surrounding context.

md
| Anything else | Event query — [query-flow.md](references/query-flow.md) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

md
Node.js 22.20 or newer is required. All API operations use `scripts/cournot-client.mjs`; credentials and wallet signatures stay inside the client. All productio

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 92)May include surrounding context.

md
ex, Grok, OpenClaw, and other Agent Skills hosts use this same folder. Install `SKILL.md`, `agents/`, `scripts/` and `references/` from `skills/cournot/`.

Instruction Override

High
Category
Prompt Injection
Confidence
70% confidence
Finding

This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.

Content

Scanner excerpt · references/account.md (reported line 61)May include surrounding context.

md
the user to confirm in the Binance App. Preserve the **new** returned intent id. Query it with `auth-status --intent '<intentId>'`; this retrieves the existing signature and does not sign again. Stop polling on rejected/expired results; do not loop automatically.
- `complete`: show balance, save status, and masked key, except the explicit `key` reveal. `hasKey=false` with zero balance means the wallet has no key record, not an authentication failure.
- `developer_mode_required`: ask the user to enable Developer Mode in the Binance App. Do not change wallet settings automatically.
- `wallet_required`: assist the selected Binance wallet's login; do not show a payment or funding prompt for authentication.
- `WALLET_351817`: wallet does not support the requested message; stop, do not substitute another signature scheme.
- `WALLET_351801`: Developer Mode is disabled. `WALLET_10003002`: session expired; sign in again. Other wallet errors: explain that the wallet operation could not complete

Instruction Override

High
Category
Prompt Injection
Confidence
70% confidence
Finding

This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.

Content

Scanner excerpt · references/errors.md (reported line 24)May include surrounding context.

md
e. If usage is unconfirmed, say remaining calls may need checking and offer balance lookup. Do not guarantee no deduction or automatically repeat the query. |
| Invalid query arguments | Explain the known input issue and ask for corrected input. Do not mislabel it an invalid key merely because both use `4100`. |
| `WALLET_351817` | The wallet does not support this authentication message. Stop without substituting a signing scheme. |
| `developer_mode_required`, `WALLET_351801` | Ask the user to enable Developer Mode in the Binance App. |
| `WALLET_10003002` | The wallet session expired; offer signing in again. |
| Account `8000` / `4400` / `22004` | Respectively: wallet authentication is invalid or expired; no key is registered for this wallet; this manually issued key cannot be self-rotated. Explain only the returned case; none means insufficient funds. |
| `account_result_unknown` or failed connection during rotation | The key change is unconfirmed. Offer account recovery to inspect

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Market-title dates are forced into an English locale without user opt-in.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The skill requires title timestamps to be converted to English Month D, YYYY dates, including for users requesting other languages. This imposes an English locale on part of the output without offering a choice or documenting a locale-specific justification.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/payment.md (reported line 100)May include surrounding context.

md
- `state=approval_pending`: show the approval transaction hash. After it confirms, prepare a fresh payment; display and confirm any changed terms before executing again.
- A command error consumes an intent once wallet authorization has begun. Do not automatically prepare or pay again. For purchases follow the recovery rules above; for probability calls report uncertainty and stop.

Never execute without confirmation, silently switch an option, pay for a different resource, or make a second paid attempt.

## Wallet unavailable or blocked

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/payment.md (reported line 132)May include surrounding context.

md
1. Show the exact wallet, network, and setup action; warn that creating or importing a signer can control real assets.
2. Ask for a separate explicit setup confirmation immediately before running any setup command. A prior request to use Cournot, choose a payment route, or pay is not this confirmation.
3. After confirmation, follow the selected wallet or SDK's official setup flow. It may generate a wallet or accept an existing key only through its secure, non-echoing credential prompt. Never ask the user to paste a private key, seed phrase, session token, or wallet credential into chat, and never pass one through a command argument, environment variable, or captured tool output. If secure hidden input is unavailable, provide the official local setup command for the user to run and stop.
4. Report only the public wallet address, selected network, supported asset, and next funding step. Do not expose credentials or raw wallet output.

Once the user chooses a wallet, preserve that select

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

The workflow sends the user's event text to an external service without a user-facing privacy disclosure.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The resolve request includes the user's event in their own words, and the probability request later sends the same text. Although the production endpoint is identified, the workflow does not instruct the agent to tell the user that their query text will be transmitted to Cournot.

Content

No source excerpt is available for this finding.

Market dates must use English month names regardless of the user's language.

Low
Category
Output Handling
Confidence
98% confidence
Finding

The rendering instruction explicitly requires English month names without user opt-in or a documented locale-specific justification. This imposes a language constraint on user-visible output even when the rest of the response uses another language.

Content

No source excerpt is available for this finding.

Output formatting forces English month names and US numeric conventions without a user locale choice.

Low
Category
Output Handling
Confidence
97% confidence
Finding

The title formatter replaces matching timestamps with hard-coded English month names, and the number formatter explicitly selects en-US. Although labels must use the user's language, these output components impose a locale without opt-in or a documented justification.

Content

No source excerpt is available for this finding.