Back to skill

Security audit

简历诊断与成品导出·元简 yotta-resume

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious with medium confidence, mainly because the recommended install commands run an unpinned package and scanners also report maintenance capabilities not disclosed in the skill description.

Consider avoiding installation until the package version and source are reviewed, since the listed install commands can run an unpinned release. The scan is partial: 105 of 107 components were scanned, some static analysis was degraded, and two files were only partially inspected, so the evidence is incomplete. The suspicious verdict reflects reported security weaknesses and undisclosed capabilities; it is not a claim that the author is malicious.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (38)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Local source comparison and candidate generation are reasonable supporting functions for the declared sourced question bank. However, fetch_sources() also issues HTTPS GET requests to registered sources, selecting all active sources when source_ids is omitted. This external retrieval capability is not described. The mismatch is limited: fetching requires an explicit function call, so the code does not establish that networking happens by default or automatically. No resume upload, automatic application submission, or direct modification of the production question bank is shown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
87% confidence
Finding

Question-bank support is related to the declared interview functionality, so ordinary bank handling is not itself a mismatch. However, this chunk exposes a dedicated maintenance interface with network source acquisition and persistent bank update/rollback capabilities that the user-facing description does not disclose. Fetching requires explicit invocation, so the code does not demonstrate networking by default or automatic résumé transmission. The imported implementations are not supplied, which limits verification of their exact side effects.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
84% confidence
Finding

The supplied command runs sources fetch --all, initiating source collection rather than merely analyzing locally supplied resume materials. Although source maintenance could support the declared question-bank functionality, bulk retrieval is not disclosed in the description, and no explicit online opt-in is visible. The underlying script is not provided, so the exact sources and whether fetching necessarily uses the network cannot be confirmed.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding

The command explicitly requests source acquisition rather than résumé analysis, interview execution, or document export. Although maintaining interview sources could support the declared skill, external fetching is not described and appears inconsistent with its stated non-networked default. Confidence is limited because this is a documented invocation, not the fetch implementation; the snippet does not prove that network access occurs or that fetching runs automatically.

Content

No source excerpt is available for this finding.

Env Variable Harvesting

High
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code enumerates, copies, or searches environment variables for secrets. Bulk environment access can collect credentials unrelated to the skill's stated purpose.

Content

Scanner excerpt · test/tools/run_r2_4_e2e.py (reported line 37)May include surrounding context.

python
def run_cli(argv, stdin=None):
    environment = os.environ.copy()
    environment["PYTHONUTF8"] = "1"
    environment["PYTHONIOENCODING"] = "utf-8"
    return subprocess.run(

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The recommended installation command executes an unpinned npm package with automatic confirmation, allowing a future compromised release to run with the user's privileges. Although this is an installer rather than an MCP server, the supply-chain risk is real and could expose local resumes or modify agent skills.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

This installation variant also executes an unpinned npm package with automatic confirmation; specifying an installation directory does not constrain the installer's privileges. A compromised future release could access sensitive local files or install malicious skill content.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The recommended npx command automatically executes whichever package release the registry currently resolves, rather than a reviewed fixed version. A compromised future release could execute arbitrary code with access to local resumes and agent configuration; the scanner's MCP terminology does not change that risk.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The directory-based installation example executes an unpinned npm release without prompting. The destination argument is not a sandbox, so a malicious release could read sensitive files or alter other agent resources.

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.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The documented installation command executes an unpinned npm package with automatic confirmation. Although this is an installer rather than an MCP server and the comment does not execute itself, following it could run a compromised future release with access to local resumes and other user files.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The custom-directory example executes a floating npm release through npx with automatic confirmation. This is not an MCP configuration or automatic execution, but a user following the example could execute a compromised release with user-level filesystem access.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

This example automatically accepts execution of an unpinned npm release and requests installation across known agent directories. The comment is not itself executable, but following it exposes the user's files and multiple agent skill locations to a compromised release.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The default installation example uses npx to execute whichever release the registry currently resolves, without a version pin or confirmation. This is a documentation-level supply-chain weakness rather than an MCP reference; a compromised release could access sensitive local resume data.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

Even the directory-listing example executes an unpinned package before that package handles --list. The provided installer only lists directories for this option, but a compromised future release need not honor that behavior and would run with the user's permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The --dry-run option controls the installer, not the initial execution of the npm package selected by npx. Following this unpinned example could therefore execute a compromised release despite the apparent non-writing intent of the command.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The version-query example still executes the registry-selected, unpinned package with automatic confirmation. Although the supplied installer only prints a version for this option, that does not constrain a compromised release fetched later.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-resume'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

Requesting help through this command executes an unpinned npm package before any help handling occurs. This is not an MCP server reference, but following the documentation could run a compromised release with access to private local files.

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/test_contract_validate.py (reported line 314)May include surrounding context.

python
)

    def run_json(self, executable, *extra):
        completed = subprocess.run(
            [
                sys.executable,
                str(executable),

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/test_contract_validate.py (reported line 553)May include surrounding context.

python
self.assertNotIn("Traceback", completed.stderr)

    def test_invalid_today_is_a_controlled_error(self):
        completed = subprocess.run(
            [
                sys.executable,
                str(VALIDATOR),

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/test_contract_validate.py (reported line 578)May include surrounding context.

python
self.assertNotIn("Traceback", completed.stderr)

    def test_invalid_today_text_output_has_no_traceback(self):
        completed = subprocess.run(
            [
                sys.executable,
                str(VALIDATOR),

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/test_contract_validate.py (reported line 611)May include surrounding context.

python
class BundledContractTest(ContractFixture):
    def run_bundled(self):
        return subprocess.run(
            [
                sys.executable,
                str(VALIDATOR),

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/test_contract_validate.py (reported line 642)May include surrounding context.

python
self.assertEqual(payload["industry_count"], 5)

    def test_main_cli_contracts_command_succeeds(self):
        completed = subprocess.run(
            [
                sys.executable,
                str(CLI),

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/test_contract_validate.py (reported line 669)May include surrounding context.

python
contract = self.bundle["dual-role-contract.json"]
        contract["permissions"]["hr"]["private_evidence_ledger"] = ["read"]
        self.rewrite("dual-role-contract.json", contract)
        completed = subprocess.run(
            [
                sys.executable,
                str(CLI),

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/test_contract_validate.py (reported line 699)May include surrounding context.

python
def test_npm_package_includes_contract_files(self):
        npm = shutil.which("npm.cmd") or shutil.which("npm")
        self.assertTrue(npm, "npm executable not found")
        completed = subprocess.run(
            [npm, "pack", "--dry-run", "--json", "--ignore-scripts"],
            cwd=str(SKILL_DIR),
            stdout=subprocess.PIPE,