Back to skill

Security audit

FarmDash Futures Strategist

Security checks for vulnerabilities and agentic risk

Overview

The skill’s trading, optional setup, and data-sharing features fit its stated purpose and include clear user authorization requirements.

Trading can lose funds despite local signing and risk checks. Review each order and the disclosed 0.001% builder fee; keep private keys local. Before enabling autonomy, review policy limits, expiry, and revocation. FarmDash receives account and order information; onboarding and installation analytics are optional. Server-side safeguards were not independently verified because their implementation is unbundled; check the [publisher’s deployment status](https://www.farmdash.one/api/v1/agent/status) before execution.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (10)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The supplied chunk performs external agent onboarding, which is not disclosed in the description or shown to support the declared Hyperliquid workflow. It contains no market analysis, sizing, signing, order management, or drawdown checks. Although onboarding could support a broader implementation, the supplied material does not establish that relationship. The literal payload contains a placeholder, not an actual wallet address.

Content

No source excerpt is available for this finding.

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · openapi.yaml (reported line 5)May include surrounding context.

yaml
openapi: 3.0.3
info:
  title: FarmDash Futures Strategist API
  version: 2.1.0
  description: |
    WARNING: Running trade executions and cancellations places real perpetual futures trades and alters active market exposure, carrying significant risk of financial loss. In USER_SIGNED, Immediate, explicit manual end-user confirmation is required for every execution or cancellation. In BOUNDED_AUTONOMOUS, an owner-signed futures policy and valid session, venue authority, exact local signature, scopes and quantitative checks replace per-order human approval. Read x-farmdash-capability-truth and the live status endpoint; policy registration or paid access never bypasses author

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · SKILL.json (reported line 567)May include surrounding context.

json
"links": {
    "website": "https://www.farmdash.one/",
    "agentHub": "https://www.farmdash.one/agents",
    "canonicalSkill": "https://www.farmdash.one/openclaw-skills/farmdash-futures-strategist/SKILL.md",
    "documentation": "https://www.farmdash.one/docs",
    "status": "https://www.farmdash.one/api/v1/agent/status",
    "openApi": "https://www.farmdash.one/agents/openapi.yaml",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · SKILL.md (reported line 718)May include surrounding context.

md
* Futures trading carries risk of loss.
* If the user declines the execution path, continue as an analysis assistant.

**Install:** Copy this file into your OpenClaw workspace, or fetch `https://www.farmdash.one/openclaw-skills/farmdash-futures-strategist/SKILL.md`.

### Companion skills:
* FarmDash Trail Intelligence — DeFi research, Trail Heat scoring, protocol risk factors

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · SKILL.md (reported line 767)May include surrounding context.

md
- [FarmDash DeFi intelligence website](https://www.farmdash.one/)
- [FarmDash Agent Hub](https://www.farmdash.one/agents)
- [Canonical FarmDash Futures Strategist skill manual](https://www.farmdash.one/openclaw-skills/farmdash-futures-strategist/SKILL.md)
- [Agent integration documentation](https://www.farmdash.one/docs)
- [Live agent capability status](https://www.farmdash.one/api/v1/agent/status)
- [OpenAPI contract](https://www.farmdash.one/agents/openapi.yaml)

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 134)May include surrounding context.

md
* `https://www.farmdash.one/api/v1/agent/onboard`

**Hyperliquid upstreams:**
* `https://api.hyperliquid.xyz/info`
* `https://api.hyperliquid.xyz/exchange`
* `wss://api.hyperliquid.xyz/ws`

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 135)May include surrounding context.

md
**Hyperliquid upstreams:**
* `https://api.hyperliquid.xyz/info`
* `https://api.hyperliquid.xyz/exchange`
* `wss://api.hyperliquid.xyz/ws`

**Optional user-facing links:**

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 · SKILL.md (reported line 160)May include surrounding context.

md
* never skip the research step before non-reduce-only execution
* never accept a claimed signer/account relationship that Hyperliquid does not currently report
* never submit when the recovered L1 signer differs from `agentAddress`
* Never ask the user to paste a private key, seed phrase, or raw wallet export into the agent.


### EVM Spot Delegation vs Hyperliquid Perps Boundary

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 · SKILL.md (reported line 267)May include surrounding context.

md
must match the equity owner bound by the live research gate.
* **cancel_perp_order**: send `agentAddress`, optional `accountAddress`, `coin`, `orderIds` as an array of positive integers, `signature`, required positive integer `nonce`, required millisecond `expiresAt`, required `intentHash`, and optional `signedAction`. Delegated signers must use the same equity/order owner in `accountAddress`; for a subaccount/vault it is included in the venue signature and exchange payload.

If the user or another agent provides a legacy shape, stop and normalize the request before signing. Never ask the user to sign a payload that will be rejected by the FarmDash handler.

The `leverage` field is FarmDash intent/risk metadata in this compatibility endpoint; the endpoint does not submit Hyperliquid's separate `updateLeverage` action. Verify actual venue margin mode and leverage independently. Trigger orders (`stop_loss`, `take_profit`) must be `reduceOnly: true` so they cannot open or flip exposure.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 748)May include surrounding context.

Optional command after consent:

bash
curl -X POST https://www.farmdash.one/api/v1/agent/onboard \
  -H "Content-Type: application/json" \
  -H "X-ClawHub-Skill: farmdash-futures-strategist" \
  -d '{"agentAddress": "0xYOUR_AGENT_WALLET"}'