Back to skill

Security audit

predge-settlement-risk

Security checks for vulnerabilities and agentic risk

Overview

This skill is an observe-only market risk checker whose network calls and optional API-key use are disclosed and aligned with its purpose.

Install only if you are comfortable sending Polymarket market identifiers to Predge and relying on its signed API output for a trade/skip gate. Keep PREDGE_API_BASE unset unless you intentionally point it at a trusted endpoint, and use SIMMER_API_KEY only when resolving Simmer market ids.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (14)

Tainted flow: 'req' from os.environ.get (line 92, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
92% confidence
Finding

The request target is partially controlled by the PREDGE_API_BASE environment variable, so the skill can be redirected to an arbitrary endpoint. In this observe-only skill that mainly risks data exfiltration of queried market identifiers and, more importantly, trust decisions being made from attacker-controlled responses, especially because attestation verification is only used for display and not enforced before gating.

Content

Scanner excerpt · settlement_risk_check.py (reported line 94)May include surrounding context.

python
url = f"{base}/v1/settlement-risk/{urllib.parse.quote(market.strip(), safe='')}"
    req = urllib.request.Request(url, headers={"accept": "application/json", "user-agent": f"simmer-skill/{SKILL_SLUG}"})
    try:
        with urllib.request.urlopen(req, timeout=timeout, context=_ssl_context()) as resp:
            return json.loads(resp.read().decode("utf-8"))
    except urllib.error.HTTPError as e:
        try:

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The README instructs users to run npx clawhub@latest, which pulls and executes the newest package version at runtime. This creates a supply-chain risk because a compromised or breaking upstream release could execute unintended code on the user's machine during login or publish operations.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

This command again uses npx clawhub@latest, causing execution of whatever version is currently published rather than a reviewed dependency version. If the package or its dependencies are compromised, users following the publish instructions may run attacker-controlled code.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

Referencing npx clawhub@latest for inspection has the same unpinned execution risk as the other commands. Even though this is documentation and not direct code execution by the skill itself, it still encourages unsafe operator behavior with a mutable remote package.

Content

No source excerpt is available for this finding.

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 · README.md (reported line 36)May include surrounding context.

npx clawhub@latest inspect predge-settlement-risk # expect Moderation: CLEAN

text

Simmer lists ClawHub skills that require `simmer-sdk` within about 6 hours: check `https://api.simmer.markets/api/sdk/skills`.

## License

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill advertises observable behavior that uses network access, may read local files such as DISCLAIMER.md, and may use environment variables (SIMMER_API_KEY), but it does not declare any explicit tool scope or permissions boundary. That mismatch weakens reviewability and can lead to over-granted execution in hosts that rely on manifest-declared capabilities for sandboxing or user consent.

Content

No source excerpt is available for this finding.

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 12)May include surrounding context.

md
attribution: "Vendor-authored by Predge. The data is UMA's own on-chain state; Predge reads, labels and signs it."
  simmer:
    credit: {name: "Predge", url: "https://predge.io", label: "by"}
    links: [https://api.predge.io/v1/settlement-risk/4424722, https://data.predge.io/samples/settlement-risk, https://api.predge.io/.well-known/predge-keys.json]
tags: [polymarket, uma, risk, pre-trade]
---
# Settlement-Risk Check (UMA on-chain)

External Transmission

Medium
Category
Data Exfiltration
Confidence
76% confidence
Finding

The skill instructs callers to send market identifiers to a third-party API endpoint. While described as observe-only and no-wallet/no-key, this still transmits user strategy interest and query patterns to an external service, creating privacy, availability, and trust risks if the API is logged, manipulated, or unavailable.

Content

Scanner excerpt · SKILL.md (reported line 24)May include surrounding context.

Before your strategy enters a Polymarket market, ask whether the market's resolution is contested on chain right now. One free call, no API key, no wallet:

text
GET https://api.predge.io/v1/settlement-risk/<market>
    <market> = Polymarket condition id (0x + 64 hex), numeric market id, or slug

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 · DEMO.md (reported line 92)May include surrounding context.

md
It answers one question before a strategy enters a Polymarket market: is the
market's resolution contested on chain right now?

    GET https://api.predge.io/v1/settlement-risk/<market>
        <market> = Polymarket condition id (0x + 64 hex), numeric market id or slug

The response says, from UMA's Optimistic Oracle and Polymarket's UMA CTF

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 · README.md (reported line 5)May include surrounding context.

md
It answers one question before a strategy enters a Polymarket market: is the
market's resolution contested on chain right now?

    GET https://api.predge.io/v1/settlement-risk/<market>
        <market> = Polymarket condition id (0x + 64 hex), numeric market id or slug

The response says, from UMA's Optimistic Oracle and Polymarket's UMA CTF

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 12)May include surrounding context.

md
It answers one question before a strategy enters a Polymarket market: is the
market's resolution contested on chain right now?

    GET https://api.predge.io/v1/settlement-risk/<market>
        <market> = Polymarket condition id (0x + 64 hex), numeric market id or slug

The response says, from UMA's Optimistic Oracle and Polymarket's UMA CTF

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 · settlement_risk_check.py (reported line 12)May include surrounding context.

python
It answers one question before a strategy enters a Polymarket market: is the
market's resolution contested on chain right now?

    GET https://api.predge.io/v1/settlement-risk/<market>
        <market> = Polymarket condition id (0x + 64 hex), numeric market id or slug

The response says, from UMA's Optimistic Oracle and Polymarket's UMA CTF

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 · settlement_risk_check.py (reported line 57)May include surrounding context.

python
It answers one question before a strategy enters a Polymarket market: is the
market's resolution contested on chain right now?

    GET https://api.predge.io/v1/settlement-risk/<market>
        <market> = Polymarket condition id (0x + 64 hex), numeric market id or slug

The response says, from UMA's Optimistic Oracle and Polymarket's UMA CTF

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

This JSON file includes natural-language instructions in the verify field, including an absolute requirement phrased as 'you MUST reproduce canonical'. Although not a spoken-language setting, this is the only plausible natural-language policy concern present because the file embeds imperative language without any opt-in or alternative phrasing.

Content

No source excerpt is available for this finding.