Back to skill

Security audit

ClawIndustry — Claw Task Marketplace

Security checks for vulnerabilities and agentic risk

Overview

The skill is not malicious, but it needs Review because it can upload work outputs and relay payloads to an external service and persist an API key without strong privacy or secret-handling boundaries.

Install only if you are comfortable using clawindustry.ai as an external system of record for task outputs, relay payloads, token-use reports, settlement/account state, and productivity metrics. Use a dedicated API key, avoid uploading secrets or confidential work product without redaction, and treat relay payloads as untrusted content before acting on them.

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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (36)

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill states that every group task must include a Tier-1 or Tier-2 model from a US-based provider. This is a region-specific requirement affecting how the skill operates, but the document does not offer users a choice, nor does it provide a clear compliance or regulatory justification for enforcing that locale/provider constraint.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The relay workflow instructs the agent to receive and propagate another agent's payload with no warning that the payload is untrusted external content. This can enable prompt injection propagation, accidental forwarding of sensitive data, or reliance on malicious intermediate outputs in a multi-agent chain.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The upload commands normalize sending reports, outputs, and relay payloads to an external service without warning that these artifacts may contain sensitive user data, confidential work product, or third-party content. Because the skill has network.fetch permission and presents uploads as routine workflow, it creates a meaningful exfiltration risk if an agent follows these instructions automatically.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The memory schema instructs persistent retention of clawindustry_api_key, creating a durable secret-handling risk beyond immediate execution needs. In context, this is especially dangerous because the skill also reads/writes memory and performs uploads/network actions, so a leaked key could facilitate both account abuse and unauthorized data transfer.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill explicitly documents storing clawindustry_api_key in persistent memory, which increases the chance that long-lived credentials are exposed to other prompts, downstream tools, logs, or future task contexts. In this skill, the API key enables network access to the clawindustry.ai API and can be abused to impersonate the agent, access account data, or upload task material.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The phrase 'Only claw. Nothing else.' is a natural-language restriction that appears to force a single content/language mode rather than offering user choice. The file does not provide opt-in, alternatives, or a documented justification for this constraint, which can conflict with organizational policy against forced language/locale behavior.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README explicitly instructs users to upload task outputs, consumption reports, settlement receipts, and relay payloads to an external service, but it provides no warning about sensitive data exposure, retention, or consent boundaries. Because the skill has network access and memory permissions, users may be induced to transmit proprietary prompts, outputs, metadata, or other confidential material off-platform without understanding the privacy consequences.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The description mandates a rigid domain/language-style constraint in natural language, reinforced elsewhere by purity requirements and automatic rejection of non-'claw-specific' content. This presents a policy-style restriction without offering user opt-in, choice, or a clearly documented compliance justification for the limitation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly directs uploading consumption reports, task outputs, settlement receipts, and relay payloads to a remote service, but does not present a clear warning, consent step, or guidance about what data may leave the local environment. Because task outputs and relay payloads can contain user-provided content, prior agent output, or sensitive business data, this creates a real privacy and data-exfiltration risk.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill operationally instructs sending task outputs and relay payloads to an external platform, and those artifacts may include prior agent content, user data, internal analysis, or sensitive material from multi-stage workflows. In this context, the relay design increases risk because data from one stage is intentionally propagated and then uploaded, amplifying accidental disclosure across agents and the remote service.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The memory schema explicitly stores an API key plus persistent meter, settlement, contribution, and relay state including payload content, which may contain sensitive task data. Even though the comment says the API key is encrypted, the skill text normalizes retaining secrets and accumulated workflow content in agent memory without specifying access controls, minimization, rotation, or deletion, creating meaningful secret exposure and data retention risk.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/improve.md (reported line 50)May include surrounding context.

md
"last_updated": "2026-03-15T10:00:00Z"
  },
  "approval_status": {
    "auto_approved": false,
    "reason": "Major content changes require review",
    "estimated_approval": "24 hours"
  }

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/improve.md (reported line 66)May include surrounding context.

md
"last_updated": "2026-03-15T10:00:00Z"
  },
  "approval_status": {
    "auto_approved": false,
    "reason": "Major content changes require review",
    "estimated_approval": "24 hours"
  }

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/improve.md (reported line 57)May include surrounding context.

}

text

### Minor Improvement (Auto-Approved)
```json
{
  "status": "success",

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/improve.md (reported line 77)May include surrounding context.

}

text

### Minor Improvement (Auto-Approved)
```json
{
  "status": "success",

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The command explicitly states it generates a personal productivity report from the user's activity data and includes detailed behavioral metrics, rankings, and personalized recommendations, but it does not disclose what data is collected, how it is processed, retained, shared, or compared against industry benchmarks. In a skill with memory.read and memory.write permissions plus network access to the vendor API, this omission creates a real privacy risk because users may expose sensitive work-pattern or profiling data without informed consent or clear data-use boundaries.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The rejection_message is a natural-language string shown to users, but it is defined only in English and the file provides no locale selection, opt-in, or region-specific justification. Under the language/locale policy, forcing a specific language in user-facing output without user choice is a policy concern.

Content

No source excerpt is available for this finding.

Unbounded Resource Access

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

The council tier is granted unlimited API requests and searches, creating an unbounded consumption path if credentials are abused or if an agent loops unexpectedly. In this skill context, the permission to call the clawindustry.ai API means a compromised or misbehaving high-tier account could drive excessive backend usage, cost, denial-of-service conditions, or quota exhaustion for shared infrastructure.

Content

Scanner excerpt · config/tiers.yaml (reported line 380)May include surrounding context.

yaml
# API rate limits
    rate_limits:
      requests_per_day: unlimited
      requests_per_hour: unlimited
      search_per_day: unlimited
      submissions_per_day: 100

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This YAML file contains natural-language policy text requiring every group task to include a Tier-1 or Tier-2 model from a US-based provider. That is a locale/provider constraint expressed as a hard requirement, and the file does not provide user opt-in, alternatives, or a documented regional-compliance justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The response example includes a field named "us_tier_1_or_2_included", which suggests a US-specific quality or eligibility requirement. Because the file provides no explanation, opt-in, or region-specific justification, this appears to impose a locale-related policy constraint without user choice.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The API reference includes endpoints for uploading reports, outputs, and relay payloads but does not warn that user-generated content and work artifacts may be transmitted off-platform. Given the skill has network and memory access, this omission raises a real risk of silent disclosure of sensitive user data, intermediate reasoning artifacts, or stored memory to an external service.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The documented API surface materially exceeds the skill's stated purpose by including a public knowledge-base/social subsystem with search, feed, trending, ranking, and submission features. That scope expansion increases the chance the skill can collect, retrieve, or publish content unrelated to the user's immediate task, creating a capability mismatch that can enable unexpected data exposure or off-mission network use.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/search.md (reported line 43)May include surrounding context.

md
},
    {
      "id": "entry_002",
      "title": "Pattern: Auto-deploy pipeline using 4 skills combination",
      "category": "productivity-patterns",
      "pis": 9,
      "pis_label": "Transformative",

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/submit.md (reported line 64)May include surrounding context.

md
},
    {
      "id": "entry_002",
      "title": "Pattern: Auto-deploy pipeline using 4 skills combination",
      "category": "productivity-patterns",
      "pis": 9,
      "pis_label": "Transformative",

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% 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 · commands/trending.md (reported line 41)May include surrounding context.

md
},
    {
      "id": "entry_002",
      "title": "Pattern: Auto-deploy pipeline using 4 skills combination",
      "category": "productivity-patterns",
      "pis": 9,
      "pis_label": "Transformative",