Back to skill

Security audit

MosOblEIRC

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because authenticated provider requests may follow redirects and expose a token to another origin; the scanners also report a possible missing confirmation check and a reference-image overwrite risk.

Treat this as a review warning rather than evidence of malicious intent. Before installing, consider whether you are comfortable with the possible token disclosure on redirects and the reported missing confirmation check; also verify the Pillow version and understand the skill's required access. The scan was partial because two local path references were unresolved.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
Findings (23)

Tainted flow: 'req' from os.environ.get (line 40, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
98% confidence
Finding

Sending the provider token to the configured provider is legitimate, but urllib.request.urlopen follows redirects by default and can forward the custom X-Auth-Tenant-Token header to a different origin. A malicious or compromised redirect response could therefore disclose the account token, making this a conditional credential-leak vulnerability rather than evidence of intentional exfiltration.

Content

Scanner excerpt · relay/yandex/index.py (reported line 42)May include surrounding context.

python
headers = {"User-Agent": USER_AGENT, "X-Auth-Tenant-Token": token, "Accept": "application/json"}
    req = urllib.request.Request(BASE_URL + path, headers=headers, method="GET")
    try:
        with urllib.request.urlopen(req, timeout=timeout) as response:
            return response.status, _decode_json(response.read())
    except urllib.error.HTTPError as exc:
        return exc.code, _decode_json(exc.read())

Tainted flow: 'req' from os.environ.get (line 54, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
98% confidence
Finding

The POST appropriately authenticates to the provider, but its default urllib redirect handling can turn a redirected POST into a GET while retaining the custom token header. An attacker-controlled redirect destination could receive the provider credential, and implicit redirection also undermines the intended tightly controlled write flow.

Content

Scanner excerpt · relay/yandex/index.py (reported line 56)May include surrounding context.

python
body=json.dumps(payload,ensure_ascii=False,separators=(",",":")).encode("utf-8")
    req=urllib.request.Request(BASE_URL+path,data=body,headers=headers,method="POST")
    try:
        with urllib.request.urlopen(req,timeout=timeout) as response:
            return response.status,_decode_json(response.read())
    except urllib.error.HTTPError as exc:
        return exc.code,_decode_json(exc.read())

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

DirectClient.call('submit', ...) invokes _submit, which sends a POST once an operation ID, valid identifiers, a nonnegative integer reading, and a matching inventory entry are present. Neither method requires or validates explicit owner approval; an operation ID is not evidence of consent, and write status is always enabled. Provider readback confirms that a reading was recorded, not that its submission was authorized. An upstream component might enforce confirmation, but that safeguard is not demonstrated in the supplied chunk. Token access, inventory checks, and provider requests otherwise support the declared transport purpose. The absence of photo recognition and local-history processing in this transport-only chunk is not independently treated as a mismatch.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

When ALLOW_WRITES is true, a submit request with an allowed item/meter pair, a nonnegative integer reading, and any nonempty operation_id can trigger an upstream POST. The operation_id is only required and echoed; it is not checked against an owner-approved operation. Inventory validation, platform authentication, and post-write verification do not establish explicit approval of a particular reading. Confirmation might be enforced elsewhere, but no such enforcement is demonstrated in this chunk. The API calls and inventory access otherwise support the declared purpose; missing photo recognition and history functionality alone are not treated as unrelated behavior in this relay component.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · tests/test_release_check.py (reported line 23)May include surrounding context.

python
def test_runtime_variant_and_env_production_are_forbidden(self):
        (self.root / "config").mkdir()
        (self.root / "config" / "runtime.prod.json").write_text("{}", encoding="utf-8")
        (self.root / ".env.production").write_text("X=1", encoding="utf-8")
        errors = release_check.scan_root(self.root)
        self.assertTrue(any("runtime.prod.json" in x and "forbidden runtime file" in x for x in errors))
        self.assertTrue(any(".env.production" in x and "forbidden runtime file" in x for x in errors))

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · tests/test_release_check.py (reported line 26)May include surrounding context.

python
(self.root / ".env.production").write_text("X=1", encoding="utf-8")
        errors = release_check.scan_root(self.root)
        self.assertTrue(any("runtime.prod.json" in x and "forbidden runtime file" in x for x in errors))
        self.assertTrue(any(".env.production" in x and "forbidden runtime file" in x for x in errors))

    def test_yandex_credentials_are_scanned_even_with_unknown_extension(self):
        secret = "AQV" + ("Z" * 28)

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · tools/release_check.py (reported line 13)May include surrounding context.

python
"config/inventory.json",
    "config/runtime.json",
    "relay/yandex/inventory.json",
    ".env",
}

REQUIRED_FILES = {

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · tools/release_check.py (reported line 67)May include surrounding context.

python
return True
    if name in {"runtime.json", "inventory.json", "runtime.env"}:
        return True
    if name.startswith(".env") and "example" not in name:
        return True
    if rel.parts and rel.parts[0] == "config":
        if ("runtime" in name or "inventory" in name) and "example" not in name:

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/meter_ocr.py (reported line 20)May include surrounding context.

python
def _tesseract(img: Image.Image, psm: int) -> str:
    with tempfile.NamedTemporaryFile(suffix='.png') as f:
        img.save(f.name)
        cp=subprocess.run(['tesseract',f.name,'stdout','--psm',str(psm),'-l','eng','-c','tessedit_char_whitelist=0123456789'],
                          text=True,stdout=subprocess.PIPE,stderr=subprocess.DEVNULL,timeout=20,check=False)
    return digits_only(cp.stdout)

Adding a reference image silently overwrites any existing reference for the meter.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

ReferenceStore.add unconditionally replaces the meter's existing reference image without an explicit replacement option, backup, or history. An accidental or adversarially induced repeat call can destroy the trusted visual baseline used for meter recognition, although this alone does not bypass the separate submission-confirmation requirement.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
70% confidence
Finding

Tool defaults are unsafe or overly permissive (e.g. disabled TLS verification, no authentication, world-writable permissions). Unsafe defaults widen the attack surface.

Content

Scanner excerpt · tests/test_core.py (reported line 92)May include surrounding context.

python
client.call("ping")
        self.assertNotIn(self.secret, str(ctx.exception))

    def test_secret_file_must_not_be_group_or_world_readable(self):
        os.chmod(self.secret_path, 0o644)
        client = core.RelayClient("https://functions.yandexcloud.net/example", self.secret_path, opener=lambda *_: None)
        with self.assertRaises(core.SecretPermissionError):

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · tests/test_direct_transport.py (reported line 206)May include surrounding context.

python
bundle / "scripts",
            ignore=shutil.ignore_patterns("__pycache__"),
        )
        result = subprocess.run(
            [
                sys.executable,
                str(bundle / "scripts" / "mosobleirc.py"),

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · LICENSE (reported line 7)May include surrounding context.

text
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Unverifiable Dependency: Pillow has 211 known advisory(ies) (CVE-2026-112390 (Pillow: Infinite loop in MIC n_frames when the embedded TIFF); CVE-2026-96638 (Pillow: BMP RLE decoder memory-amplification DoS via unbounded delta escape / ab); CVE-2026-112391 (Pillow: PdfParser can recurse without limit in get_value() and linearize_page_tr) +208 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
40% confidence
Finding

OSV returned advisories for this package, but its resolved version is unknown. A package-name lookup does not establish whether the installed release is affected by those advisories.

Content

No source excerpt is available for this finding.