Back to skill

Security audit

Resource Hunter

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its public-resource search purpose, but it has review-worthy network safety issues that could let untrusted search results drive requests to unexpected destinations.

Review before installing in environments that have access to private intranet, cloud metadata, or loopback-only services. Normal use is user-directed public search, but the network fetch logic should be tightened before use on sensitive networks, and users should treat returned pan/torrent/video links as untrusted external content.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
src/resource_hunter/adapters.py:742
Finding

Dalipan fallback disables TLS certificate verification

Content
View full analysis
dict[str, Any]: try: payload = http_client.get_json(url, timeout=12) if isinstance(payload, dict): return payload raise SchemaError(f"invalid dalipan payload type: {type(payload).__name__}", source=self.name, url=url) except Exception as primary_error: request = urllib.request.Request(url, headers=DEFAULT_HEADERS | {"Accept": "application/json"}) try: with urllib.request.urlopen(request, timeout=12, context=ssl._create_unverified_context()) as response: payload = json.loads( response.read().decode(response.headers.get_content_charset() or "utf-8", errors="replace") ) except json.JSONDecodeError as exc: raise SchemaError(f"invalid json from {url}: {exc}", source=self.name, url=url) from exc except Exception as insecure_error: message = str(insecure_error) or str(primary_error) or "request failed" raise NetworkError(message, source=self.name, url=url) from insecure_error if not isinstance(payload, dict): raise SchemaError( f"invalid dalipan payload type: {type(payload).__name__}", source=self.name, url=url ) return payload ``` An equivalent legacy implementation also exists in `scripts/resource_hunter/precision_core.py`. ### Technical Analysis The normal Dalipan request uses standard TLS validation. However, any exception from that request causes the fallback path to call `ssl._create_unverified_context()`. This context disables verification of the server certificate and hostname. Consequently, HTTPS no longer authenticates `api.dalipan.com` during the f ...[truncated 2043 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
src/resource_hunter/adapters.py:336
Finding

Search-derived URLs are fetched without private-network or redirect validation

Content
View full analysis
list[str]: html_text = http_client.get_text(url, timeout=12) texts: list[str] = [url] ``` ### Technical Analysis The alias resolver obtains result URLs from public search-engine HTML and then retrieves those URLs directly. Although parsed search results are initially required to begin with `http`, the implementation does not: - Restrict retrieval to an allowlist ...[truncated 2848 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (1307)

Context-Inappropriate Capability

Critical
Category
Not specified by scanner
Confidence
99% confidence
Finding

This code bootstraps build dependencies by invoking pip install into an overlay directory and later builds/installs the package, creating an explicit network-capable package installation and code execution path inside a skill unrelated to package management. In context, this is especially dangerous because the skill should only perform resource lookup, yet it can modify the environment and execute package build/install workflows on the host.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied file appears to overstate functionality that is not visible in the provided material, which is a manifest-integrity problem even if not intentionally deceptive. In an agent ecosystem, inaccurate manifests impair security review and could mask future addition of risky behaviors without obvious documentation changes.

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · artifacts/live-tests/bing-attack-on-titan.html (reported line 5)May include surrounding context.

html
si_ST=new Date

//]]></script><head><!--pc--><title>Attack on Titan magnet OR pan - 検索</title><meta content="text/html; charset=utf-8" http-equiv="content-type" /><meta name="referrer" content="origin-when-cross-origin" /><meta property="og:description" content="Bing のインテリジェント検索機能により、探しているものをより簡単にすばやく見つけられるようになり、リワードも得られます。" /><meta property="og:site_name" content="Bing" /><meta property="og:title" content="Attack on Titan magnet OR pan - Bing" /><meta property="og:url" content="https://www.bing.com/search?q=Attack on Titan magnet OR pan" /><meta property="fb:app_id" content="3732605936979161" /><meta property="og:image" content="http://www.bing.com/sa/simg/facebook_sharing_5.png" /><meta property="og:type" content="website" /><meta property="og:image:width" content="600" /><meta property="og:image:height" content="315" /><link rel="alternate"

                      href="/search?format=rss&amp;q=Attack%20on%20Titan%20magnet%20OR%20pan"title='XML'type='text/xml' /><link rel="alternate"

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 · artifacts/live-tests/bing-attack-on-titan.html (reported line 20)May include surrounding context.

html
�と設定" data-priority="2"><a class="id_button tooltip  linkBtn" data-priority="2" id="id_rh_w" dir="ltr" aria-label="Microsoft Rewards" aria-expanded="false" aria-controls="rewid-f" aria-haspopup="true" role="button" tabindex="0" href="javascript:void(0)" h="ID=SERP,5055.1"><div id="rh_rwm"

                 class="serp red-balance kumo_rewards " data-rewards-widget="medallion" data-content="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