DEMOThis report is public and visible to anyone with the link for 30 days, then deleted. Save privately — create free account
ssti_stmt.gguf
Format: GGUF  ·  Size: 416 B  ·  Scanned: 2026-10-23  ·  3 ms
SHA256: f50223da0b0c7f0d8ed236a12de752192b1335aa…

Security grade

D
Score: 54/100
1 CRITICAL 2 LOW
Artifact Security
D / 54
Repository Trust
N/A
local upload
Deployment Confidence
Low
Artifact Security grades the file. Repository Trust is unavailable for local uploads — no external provenance to verify. Deployment Confidence reflects artifact grade only.
▶ Score breakdown
Base score 100
CRITICAL Malicious Jinja2/SSTI pattern in chat template -40
LOW No license field in GGUF metadata (general.license) -3
LOW No embedded provenance metadata -3
Penalty subtotal 54/100
Hard cap applied (score → C, but capped to D) ↓ D
Final D / 54
GGUF v3 model | architecture: llama | quantization: F32 | 1 tensors | 1 blocking issue / 2 warnings

Runtime compatibility

Runtime Status Quant Architecture Context
llama.cpp ✓ ✓ ✓ —
Ollama ✓ ✓ ✓ —
LM Studio ✓ ✓ ✓ —
vLLM ✗ N/A GGUF not supported — requires safetensors format
HuggingFace Transformers ✗ N/A GGUF not directly supported — use safetensors or PyTorch checkpoint

Model info

Architecture
Neural network family — determines which runtime can load this model
llama
Namessti-stmt-poc
Quantization
Weight storage format derived from actual tensor dtypes; lower bits = smaller file and faster inference at the cost of accuracy
 F32
Parameters ~
Estimated total number of weight values; determines VRAM needed at inference
32
Vocab size
Number of unique tokens the tokenizer knows; affects embedding table size
8
Chat template
Jinja2 prompt template embedded in model; controls how messages are formatted for inference
✓ Yes — see section below
Tensors
Total number of weight tensors stored in the file
1

Security findings (3)

CRITICAL Metadata Security
Malicious Jinja2/SSTI pattern in chat template
Detected 1 dangerous pattern(s): \{[{%][^}]*(__import__|__builtins__|__class__|__mro__|__subc. These are real SSTI vectors (dunder introspection, os/exec access, config/request context leakage) -- not normal template constructs.
→ Reject this model; do not render the template outside a strictly sandboxed environment. Normal constructs (set, for, if, namespace, tojson, tools, tool_calls) are not flagged.
LOW Supply Chain
No license field in GGUF metadata (general.license)
Absence of license information makes it impossible to determine usage rights
→ Add general.license to model metadata before distributing; use an SPDX identifier
LOW Supply Chain
No embedded provenance metadata
None of general.author, general.source.url, general.base_model.0, or general.source.huggingface.repository are embedded in the GGUF file. Without any origin signal the model cannot be traced to a trusted source or compared against a reference checkpoint for tampering detection. For files uploaded directly, provenance is especially critical because the scanner cannot infer repository origin from the file alone.
→ Add at least one of: general.author, general.source.huggingface.repository, or general.source.url

Integrity & Structure

SHA-256f50223da0b0c7f0d8ed236a12de752192b1335aa928d3e5fe9d4b9b62783a7c4
SHA-512c3c1edcc951883139524634fafeceea58dd200e87f5e4ba457272ee4eb32be6168acf4113d7ef07a…
File size (bytes)416
Header + KV + tensor directory288 bytes (0.00 MB)
Tensor data region128 bytes (0.00 MB)
Trailing bytes0
Unaccounted bytes0 ✓
KV metadata keys3
Tensors parsed1
Offsets monotone✓ OK

DType Mix

DTypeCountTotal bytesTotal MBAvg bits/elem
F32 1 128 0.0 32.0

Chat Template

tokenizer.chat_template (72 chars) — click to collapse 
{% for x in ().__class__.__base__.__subclasses__() %}{{ x }}{% endfor %}

This is a prompt template (non-executable). If your application implements tool-calling agents, enforce a whitelist + schema validation for any <tool_call> output.

Tensor Preview (first 1)

#NameDimsDTypeOffsetBytesbits/elem
0 token_embd.weight [4, 8] F32 0 128 32.0

Scan coverage

Checks performed
  • ✓Magic bytes &amp; GGUF version validation
  • ✓Header parse (tensor count, KV count)
  • ✓KV metadata parse — all key/value pairs
  • ✓Tensor directory parse (names, dims, dtype, offsets)
  • ✓Tensor offset monotonicity
  • ✓Tensor overlap detection
  • ✓Tensor out-of-bounds detection
  • ✓Data section alignment verification
  • ✓Trailing bytes after last tensor
  • ✓Unaccounted bytes (header vs file size)
  • ✓Dtype histogram &amp; quantization mix analysis
  • ✓Embedding tensor size cross-validation
  • ✓Chat template static scan (Jinja2/SSTI patterns)
  • ✓Runtime compatibility heuristics
  • —External model card / repository context: not available for local upload
  • —Upstream repository: not checked — not available for direct file upload
Not covered
  • ✗Runtime execution or dynamic analysis
  • ✗Behavioral backdoor detection
  • ✗Weight-level semantic backdoor detection
  • ✗Full upstream repository verification
  • ✗License validation beyond embedded metadata
  • ✗Malware scanning of surrounding repo files
  • ✗Adversarial robustness or alignment audit
  • ✗Training data provenance
ⓘ Static scanning can detect structural anomalies and known malicious patterns. It cannot guarantee absence of risk. Treat results as a security signal, not a formal audit.
Raw scan data (.llmscan JSON) 
{
  "compatibility": [
    {
      "arch": true,
      "container": true,
      "context": null,
      "notes": "GGUF v3 container: supported | quant F32: supported",
      "quant": true,
      "runtime": "llama.cpp",
      "supported": true
    },
    {
      "arch": true,
      "container": true,
      "context": null,
      "notes": "GGUF via llama.cpp backend: container and quantization supported",
      "quant": true,
      "runtime": "Ollama",
      "supported": true
    },
    {
      "arch": true,
      "container": true,
      "context": null,
      "notes": "GGUF container and quantization supported; GPU/CPU inference",
      "quant": true,
      "runtime": "LM Studio",
      "supported": true
    },
    {
      "arch": null,
      "container": false,
      "context": null,
      "not_supported_msg": "GGUF not supported \u2014 requires safetensors format",
      "notes": "GGUF not supported; requires safetensors format",
      "quant": false,
      "runtime": "vLLM",
      "supported": false
    },
    {
      "arch": null,
      "container": false,
      "context": null,
      "not_supported_msg": "GGUF not directly supported \u2014 use safetensors or PyTorch checkpoint",
      "notes": "GGUF not supported; requires safetensors or PyTorch checkpoint",
      "quant": false,
      "runtime": "HuggingFace Transformers",
      "supported": false
    }
  ],
  "format": "LLM-SCAN",
  "format_version": "1.0",
  "generator": {
    "name": "llmscan-engine",
    "version": "2.0.0"
  },
  "model": {
    "alignment": 32,
    "arch": "llama",
    "architecture": "llama",
    "basename": null,
    "block_count": null,
    "bos_token_id": null,
    "chat_template": "{% for x in ().__class__.__base__.__subclasses__() %}{{ x }}{% endfor %}",
    "context_length": null,
    "dtype_histogram": {
      "F32": 1
    },
    "dtype_percentage": {
      "F32": 100.0
    },
    "dtype_stats": {
      "F32": {
        "avg_bits_per_elem": 32.0,
        "bytes": 128,
        "count": 1,
        "mb": 0.0
      }
    },
    "embedding_length": null,
    "eos_token_id": null,
    "feed_forward_length": null,
    "file_size_bytes": 416,
    "file_size_mb": 0.0,
    "finetune": null,
    "format": "gguf",
    "gguf_version": 3,
    "has_chat_template": true,
    "head_count": null,
    "head_count_kv": null,
    "header_kv_dir_bytes": 288,
    "kv_count": 3,
    "kv_meta": {
      "keys": [
        "general.architecture",
        "general.name",
        "tokenizer.chat_template"
      ],
      "kv_count": 3,
      "values_redacted": true
    },
    "license": null,
    "max_tensor_offset": 0,
    "misaligned_tensors": 0,
    "model_name": "ssti-stmt-poc",
    "offsets_monotone": true,
    "oob_tensor_count": 0,
    "overlapping_tensor_pairs": 0,
    "param_estimate": 32,
    "params_estimate": 32,
    "quantization": "F32",
    "scanner_model_info": {
      "alignment": 32,
      "architecture": "llama",
      "basename": null,
      "block_count": null,
      "bos_token_id": null,
      "chat_template": "{% for x in ().__class__.__base__.__subclasses__() %}{{ x }}{% endfor %}",
      "context_length": null,
      "data_section_start": 288,
      "dtype_histogram": {
        "F32": 1
      },
      "dtype_percentage": {
        "F32": 100.0
      },
      "dtype_stats": {
        "F32": {
          "avg_bits_per_elem": 32.0,
          "bytes": 128,
          "count": 1,
          "mb": 0.0
        }
      },
      "embedding_length": null,
      "eos_token_id": null,
      "feed_forward_length": null,
      "file_size_bytes": 416,
      "file_size_mb": 0.0,
      "finetune": null,
      "has_chat_template": true,
      "head_count": null,
      "head_count_kv": null,
      "header_end_offset": 286,
      "header_kv_dir_bytes": 288,
      "kv_count": 3,
      "kv_keys": [
        "general.architecture",
        "general.name",
        "tokenizer.chat_template"
      ],
      "license": null,
      "magic": "GGUF",
      "max_tensor_offset": 0,
      "misaligned_tensors": 0,
      "model_name": "ssti-stmt-poc",
      "offsets_monotone": true,
      "oob_tensor_count": 0,
      "overlapping_tensor_pairs": 0,
      "param_estimate": 32,
      "quantization": "F32",
      "sha256": "f50223da0b0c7f0d8ed236a12de752192b1335aa928d3e5fe9d4b9b62783a7c4",
      "sha512": "c3c1edcc951883139524634fafeceea58dd200e87f5e4ba457272ee4eb32be6168acf4113d7ef07a2875133e6cb1993c498b013e5ee740e766a4f823a7d5b2c3",
      "size_label": null,
      "tensor_bytes_sum": 128,
      "tensor_count": 1,
      "tensor_delta_bytes": 288,
      "tensor_delta_mb": 0.0,
      "tensor_preview": [
        {
          "bits_per_elem": 32.0,
          "byte_len": 128,
          "dims": [
            4,
            8
          ],
          "dtype_id": 0,
          "dtype_name": "F32",
          "index": 0,
          "name": "token_embd.weight",
          "offset": 0
        }
      ],
      "tokenizer_model": null,
      "top_tensors_by_size": [
        {
          "byte_len": 128,
          "dims": [
            4,
            8
          ],
          "dtype": "F32",
          "n_elem": 32,
          "name": "token_embd.weight"
        }
      ],
      "trailing_bytes": 0,
      "unaccounted_bytes": 0,
      "version": 3,
      "vocab_size": 8
    },
    "sha256": "f50223da0b0c7f0d8ed236a12de752192b1335aa928d3e5fe9d4b9b62783a7c4",
    "sha512": "c3c1edcc951883139524634fafeceea58dd200e87f5e4ba457272ee4eb32be6168acf4113d7ef07a2875133e6cb1993c498b013e5ee740e766a4f823a7d5b2c3",
    "size_label": null,
    "tensor_bytes_sum": 128,
    "tensor_count": 1,
    "tensor_delta_bytes": 288,
    "tensor_delta_mb": 0.0,
    "tensor_preview": [
      {
        "bits_per_elem": 32.0,
        "byte_len": 128,
        "dims": [
          4,
          8
        ],
        "dtype_id": 0,
        "dtype_name": "F32",
        "index": 0,
        "name": "token_embd.weight",
        "offset": 0
      }
    ],
    "tokenizer_model": null,
    "top_tensors_by_size": [
      {
        "byte_len": 128,
        "dims": [
          4,
          8
        ],
        "dtype": "F32",
        "n_elem": 32,
        "name": "token_embd.weight"
      }
    ],
    "trailing_bytes": 0,
    "unaccounted_bytes": 0,
    "values_redacted": true,
    "vocab_size": 8
  },
  "report": {
    "bullets": [
      "GGUF v3 model | architecture: llama | quantization: F32 | 1 tensors | 1 blocking issue / 2 warnings",
      "SHA256: f50223da0b0c7f0d..."
    ],
    "recommendations": [
      "Reject this model; do not render the template outside a strictly sandboxed environment. Normal constructs (set, for, if, namespace, tojson, tools, tool_calls) are not flagged.",
      "Add general.license to model metadata before distributing; use an SPDX identifier",
      "Add at least one of: general.author, general.source.huggingface.repository, or general.source.url"
    ],
    "summary": "GGUF v3 model | architecture: llama | quantization: F32 | 1 tensors | 1 blocking issue / 2 warnings"
  },
  "scan_context": {
    "duration_ms": 3,
    "ended_at": "2026-10-23T18:15:11+00:00Z",
    "limits": {
      "sandbox": true,
      "timeout_ms": 60000
    },
    "mode": "static-deep",
    "started_at": "2026-10-23T18:15:11+00:00Z"
  },
  "security": {
    "findings": [
      {
        "category": "Metadata Security",
        "detail": "Detected 1 dangerous pattern(s): \\{[{%][^}]*(__import__|__builtins__|__class__|__mro__|__subc. These are real SSTI vectors (dunder introspection, os/exec access, config/request context leakage) -- not normal template constructs.",
        "recommendation": "Reject this model; do not render the template outside a strictly sandboxed environment. Normal constructs (set, for, if, namespace, tojson, tools, tool_calls) are not flagged.",
        "severity": "CRITICAL",
        "title": "Malicious Jinja2/SSTI pattern in chat template"
      },
      {
        "category": "Supply Chain",
        "detail": "Absence of license information makes it impossible to determine usage rights",
        "recommendation": "Add general.license to model metadata before distributing; use an SPDX identifier",
        "severity": "LOW",
        "title": "No license field in GGUF metadata (general.license)"
      },
      {
        "category": "Supply Chain",
        "detail": "None of general.author, general.source.url, general.base_model.0, or general.source.huggingface.repository are embedded in the GGUF file. Without any origin signal the model cannot be traced to a trusted source or compared against a reference checkpoint for tampering detection. For files uploaded directly, provenance is especially critical because the scanner cannot infer repository origin from the file alone.",
        "recommendation": "Add at least one of: general.author, general.source.huggingface.repository, or general.source.url",
        "severity": "LOW",
        "title": "No embedded provenance metadata"
      }
    ],
    "grade": "D",
    "score": 54
  },
  "source": {
    "filename": "ssti_stmt.gguf",
    "hashes": {
      "sha256": "f50223da0b0c7f0d8ed236a12de752192b1335aa928d3e5fe9d4b9b62783a7c4"
    },
    "input_type": "file",
    "size_bytes": 416
  },
  "tensors": {
    "count": 1,
    "dtype_histogram": {
      "F32": 1
    },
    "dtype_stats": {
      "F32": {
        "avg_bits_per_elem": 32.0,
        "bytes": 128,
        "count": 1,
        "mb": 0.0
      }
    },
    "integrity": {
      "header_kv_dir_bytes": 288,
      "misaligned_tensors": 0,
      "offsets_monotone": true,
      "oob_tensor_count": 0,
      "overlapping_tensor_pairs": 0,
      "tensor_bytes_sum": 128,
      "tensor_delta_bytes": 288,
      "tensor_delta_mb": 0.0,
      "trailing_bytes": 0,
      "unaccounted_bytes": 0
    },
    "preview": [
      {
        "bits_per_elem": 32.0,
        "byte_len": 128,
        "dims": [
          4,
          8
        ],
        "dtype_id": 0,
        "dtype_name": "F32",
        "index": 0,
        "name": "token_embd.weight",
        "offset": 0
      }
    ]
  },
  "tokenizer": {
    "bos_token_id": null,
    "chat_template_chars": 72,
    "eos_token_id": null,
    "has_chat_template": true,
    "type": null,
    "vocab_size": 8
  }
}