Tinman - AI Failure Mode Research, Prompt Injection & Tool Exfil Detection

⬇ 2,715 downloads · ⭐ 2

High Risk

AI security scanner with active prevention - 168 detection patterns, 288 attack probes, safer/risky/yolo modes, agent self-protection via /tinman check, local Oilcan event streaming, and plain-language dashboard setup via /tinman oilcan

H:4 D:4 A:0 C:1

⚠️ Hazard Flags

PRIVILEGED PERSISTENCE FS_READ_WORKSPACE FS_READ_USER FS_DELETE NET_EGRESS_ANY CREDS_FILES CREDS_BROWSER CREDS_KEYCHAIN PI_WEB PI_DOCUMENTS

📋 Capabilities

Execution

❌ Shell execution
❌ Code execution
❌ Install dependencies
✅ Persistence
Privilege: admin

Filesystem

✅ Read workspace
❌ Write workspace
✅ Read home
❌ Write home
❌ Read system
✅ Delete

Network

Egress: any
❌ Ingress

Credentials

❌ Environment vars
✅ Credential files
✅ Browser data
✅ Keychain

Actions

❌ send messages❌ post public❌ purchase❌ transfer money❌ deploy❌ delete external

🔒 Containment

Level: maximum

Recommended:

LOG_ACTIONS: Audit trail for all actions

⚡ Risks

Base64 decode + execute pattern in: tinman_runner.py critical

Mitigation: Decode and review obfuscated content before use.

SSH key access patterns in: SKILL.md, tinman_runner.py high

Mitigation: Verify legitimate need for SSH access. Consider sandboxing.

Cloud credential access patterns in: tinman_runner.py high

Mitigation: Verify legitimate need for cloud credential access.

Keychain/credential store access in: tinman_runner.py high

Mitigation: Block keychain access or use isolated environment.

Prompt injection patterns detected in: SKILL.md high

Mitigation: Review SKILL.md for hidden instructions. Do not use with untrusted input.

Unauthorized tool use: MCP_SYS_FILE_DESTRUCTION, MCP_SYS_PERMISSION_MANIPULATION critical

Mitigation: Remove destructive file operations.

Command injection risk: MCP_CMD_DANGEROUS_SYSTEM, MCP_SQL_BLIND critical

Mitigation: Remove dangerous system commands.

Want a deeper analysis?

This report was generated by static analysis. Get an LLM-powered deep review with behavioral reasoning and attack surface mapping.

🧠 Deep Analysis — $5.00

🚨 Incident Response

Kill switch: Stop the agent process

Containment: Review logs for unexpected actions

Recovery: Depends on skill capabilities

📄 Raw SSDS JSON click to expand

{
  "meta": {
    "document_id": "ssds:auto:tinman:0.6.4",
    "ssds_version": "0.2.0",
    "scanner_version": "0.4.0+fe6fd9123d50",
    "created_at": "2026-03-05T01:32:18.422Z",
    "created_by": {
      "agent": "safeagentskills-cli/generate-ssds"
    },
    "language": "en",
    "notes": "Auto-generated SSDS. Manual review recommended."
  },
  "skill": {
    "name": "Tinman -  AI Failure Mode Research, Prompt Injection & Tool Exfil Detection",
    "version": "0.6.4",
    "format": "agent_skill",
    "description": "AI security scanner with active prevention - 168 detection patterns, 288 attack probes, safer/risky/yolo modes, agent self-protection via /tinman check, local Oilcan event streaming, and plain-language dashboard setup via /tinman oilcan",
    "publisher": "ClawHub",
    "source": {
      "channel": "clawhub",
      "slug": "agent-tinman",
      "downloads": 2715,
      "stars": 2
    },
    "artifact": {
      "sha256": "15fb7cc2f2857be302e3254d7e00a53e07fd876460ce234e159c4dc677878082",
      "hash_method": "files_sorted"
    }
  },
  "capabilities": {
    "execution": {
      "can_exec_shell": false,
      "can_exec_code": false,
      "privilege_level": "admin",
      "can_install_deps": false,
      "can_persist": true
    },
    "filesystem": {
      "reads_workspace": true,
      "reads_user_home": true,
      "reads_system": false,
      "writes_workspace": false,
      "writes_user_home": false,
      "writes_system": false,
      "can_delete": true
    },
    "network": {
      "egress": "any",
      "ingress": false
    },
    "credentials": {
      "reads_env_vars": false,
      "reads_credential_files": true,
      "reads_browser_data": true,
      "reads_keychain": true
    },
    "services": [],
    "actions": {
      "can_send_messages": false,
      "can_post_public": false,
      "can_purchase": false,
      "can_transfer_money": false,
      "can_deploy": false,
      "can_delete_external": false
    },
    "prompt_injection_surfaces": [
      "web",
      "documents"
    ],
    "content_types": [
      "general"
    ]
  },
  "hazards": {
    "hdac": {
      "H": 4,
      "D": 4,
      "A": 0,
      "C": 1
    },
    "flags": [
      "PRIVILEGED",
      "PERSISTENCE",
      "FS_READ_WORKSPACE",
      "FS_READ_USER",
      "FS_DELETE",
      "NET_EGRESS_ANY",
      "CREDS_FILES",
      "CREDS_BROWSER",
      "CREDS_KEYCHAIN",
      "PI_WEB",
      "PI_DOCUMENTS"
    ],
    "custom_flags": [
      {
        "code": "BASE64_EXEC",
        "name": "Base64 Execute",
        "description": "Decodes and executes obfuscated code in: tinman_runner.py"
      },
      {
        "code": "CRED_SSH",
        "name": "SSH Key Access",
        "description": "Accesses SSH keys in: SKILL.md, tinman_runner.py"
      },
      {
        "code": "CRED_CLOUD",
        "name": "Cloud Credential Access",
        "description": "Accesses cloud credentials (AWS/GCP/Azure) in: tinman_runner.py"
      },
      {
        "code": "CRED_BROWSER",
        "name": "Browser Profile Access",
        "description": "Accesses browser cookies/profiles in: tinman_runner.py"
      },
      {
        "code": "CRED_KEYCHAIN",
        "name": "Keychain Access",
        "description": "Accesses system keychain in: tinman_runner.py"
      },
      {
        "code": "PROMPT_INJECTION",
        "name": "Prompt Injection Risk",
        "description": "Contains prompt injection patterns in: SKILL.md"
      },
      {
        "code": "PERSISTENCE",
        "name": "Persistence Mechanism",
        "description": "Creates scheduled tasks or startup entries in: tinman_runner.py"
      },
      {
        "code": "PRIVILEGE_ESCALATION",
        "name": "Privilege Escalation",
        "description": "Uses elevated privileges (sudo/root) in: tinman_runner.py"
      },
      {
        "code": "FILE_DELETE",
        "name": "File Deletion",
        "description": "Can delete files in: tinman_runner.py"
      },
      {
        "code": "TOOL_ABUSE",
        "name": "Unauthorized Tool Use",
        "description": "MCP_SYS_FILE_DESTRUCTION, MCP_SYS_PERMISSION_MANIPULATION: File destruction or wiping commands"
      },
      {
        "code": "COMMAND_INJECTION",
        "name": "Command Injection Risk",
        "description": "MCP_CMD_DANGEROUS_SYSTEM, MCP_SQL_BLIND: Dangerous system commands or Windows exploitation"
      }
    ],
    "confidence": {
      "level": "medium",
      "basis": [
        "static_analysis"
      ],
      "notes": "Detected 11 security patterns (6 vendored rule hits). Review recommended."
    },
    "rationale": {
      "H": "H4: Critical: Privilege escalation or malware detected",
      "D": "D4: Critical: Credential theft or data exfiltration",
      "A": "A0: No side effects detected",
      "C": "C1: General content"
    }
  },
  "containment": {
    "level": "maximum",
    "required": [],
    "recommended": [
      {
        "control": "LOG_ACTIONS",
        "reason": "Audit trail for all actions"
      }
    ],
    "uncontained_risk": "Risk level depends on manual review of actual capabilities."
  },
  "risks": {
    "risks": [
      {
        "risk": "Base64 decode + execute pattern in: tinman_runner.py",
        "severity": "critical",
        "mitigation": "Decode and review obfuscated content before use."
      },
      {
        "risk": "SSH key access patterns in: SKILL.md, tinman_runner.py",
        "severity": "high",
        "mitigation": "Verify legitimate need for SSH access. Consider sandboxing."
      },
      {
        "risk": "Cloud credential access patterns in: tinman_runner.py",
        "severity": "high",
        "mitigation": "Verify legitimate need for cloud credential access."
      },
      {
        "risk": "Keychain/credential store access in: tinman_runner.py",
        "severity": "high",
        "mitigation": "Block keychain access or use isolated environment."
      },
      {
        "risk": "Prompt injection patterns detected in: SKILL.md",
        "severity": "high",
        "mitigation": "Review SKILL.md for hidden instructions. Do not use with untrusted input."
      },
      {
        "risk": "Unauthorized tool use: MCP_SYS_FILE_DESTRUCTION, MCP_SYS_PERMISSION_MANIPULATION",
        "severity": "critical",
        "mitigation": "Remove destructive file operations."
      },
      {
        "risk": "Command injection risk: MCP_CMD_DANGEROUS_SYSTEM, MCP_SQL_BLIND",
        "severity": "critical",
        "mitigation": "Remove dangerous system commands."
      }
    ],
    "limitations": [
      "Static analysis only - runtime behavior not verified"
    ]
  },
  "incident_response": {
    "kill_switch": [
      "Stop the agent process"
    ],
    "containment": [
      "Review logs for unexpected actions"
    ],
    "recovery": [
      "Depends on skill capabilities"
    ]
  },
  "evidence": [
    {
      "evidence_id": "EV:file-1",
      "type": "file_excerpt",
      "title": "_meta.json",
      "file_path": "_meta.json"
    },
    {
      "evidence_id": "EV:file-2",
      "type": "file_excerpt",
      "title": "SKILL.md",
      "file_path": "SKILL.md"
    },
    {
      "evidence_id": "EV:file-3",
      "type": "file_excerpt",
      "title": "tinman_runner.py",
      "file_path": "tinman_runner.py"
    },
    {
      "evidence_id": "EV:cisco-1",
      "type": "file_excerpt",
      "title": "PROMPT_INJECTION_IGNORE_INSTRUCTIONS [HIGH] SKILL.md:269: **Evidence:** \"Ignore previous instructions and...\"",
      "file_path": "SKILL.md"
    },
    {
      "evidence_id": "EV:cisco-2",
      "type": "file_excerpt",
      "title": "MCP_PI_PRIVILEGE_ESCALATION [HIGH] SKILL.md:111: Add patterns to the allowlist (bypass security checks for trusted items).",
      "file_path": "SKILL.md"
    },
    {
      "evidence_id": "EV:cisco-3",
      "type": "file_excerpt",
      "title": "MCP_CMD_DANGEROUS_SYSTEM [CRITICAL] tinman_runner.py:545: # PowerShell",
      "file_path": "tinman_runner.py"
    },
    {
      "evidence_id": "EV:cisco-4",
      "type": "file_excerpt",
      "title": "MCP_SYS_FILE_DESTRUCTION [CRITICAL] tinman_runner.py:630: \"destructive\": [\"rm -rf\", \"rm -r\", \"mkfs\", \"dd if=\", \"> /dev/\", \"shred\"],",
      "file_path": "tinman_runner.py"
    },
    {
      "evidence_id": "EV:cisco-5",
      "type": "file_excerpt",
      "title": "MCP_SYS_PERMISSION_MANIPULATION [CRITICAL] tinman_runner.py:631: \"privilege_escalation\": [\"sudo\", \"chmod 777\", \"chmod +x\", \"chown\", \"setuid\"],",
      "file_path": "tinman_runner.py"
    },
    {
      "evidence_id": "EV:cisco-6",
      "type": "file_excerpt",
      "title": "MCP_SQL_BLIND [HIGH] tinman_runner.py:1537: await asyncio.sleep(interval_minutes * 60)",
      "file_path": "tinman_runner.py"
    }
  ]
}