Back to skill

Security audit

Speediance scan PR34 r1

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rated this skill suspicious because ordinary workout mentions may trigger access to Speediance workout data, and browsing the exercise catalog also saves a full local copy without an explicit export request.

Consider the privacy implications of having Speediance workout details retrieved after a general mention of exercise, and the full exercise catalog saved locally during browsing. The verdict is suspicious, not malicious. Skillspector’s analysis is marked partial because some referenced paths could not be resolved; another scanner completed with no findings, but that does not establish that the skill is safe.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (10)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

Catalog retrieval and export match the declared functionality. However, the description explicitly says the caller decides where to store structured data. This command always invokes writeJSONFile, with library.json as the default destination and no visible option to disable the write. Consequently, even a browsing or stdout-only invocation persists the full catalog locally. That storage side effect is not accurately represented by the description.

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 · SKILL.md (reported line 21)May include surrounding context.

md
- "Speediance cloud API (HTTPS) — authentication, workout history, exercise catalog, program creation"
      files.read:
        - "config.json — optional credential/config file (working directory, or your OS user-config dir)"
        - ".env — optional env file in the working directory; only the SPEEDIANCE_EMAIL, SPEEDIANCE_PASSWORD, SPEEDIANCE_REGION and SPEEDIANCE_DEVICE_TYPE keys are read (parsed to a map, never exported into the process environment); the path keys SPEEDIANCE_CONFIG and SPEEDIANCE_TOKEN_CACHE are ignored there"
        - "token cache — cached session token in your OS user-cache dir by default (non-roaming); path overridable via an exported SPEEDIANCE_TOKEN_CACHE, or the token_cache_path key of a config.json given by --config, SPEEDIANCE_CONFIG or the user-config dir (ignored in a working-directory config.json). A legacy .token.json in the working directory is read once to migrate it."
        - "plan JSON files passed to the push command"
      files.write:

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 · SKILL.md (reported line 47)May include surrounding context.

md
description: "Device type integer — 1 = Gym Monster v1 (default, tested). Gym Monster 2 is untested; try 2 if exercises look wrong."
        required: false
      - name: SPEEDIANCE_CONFIG
        description: "Path to config.json (overrides discovery: working dir, then OS user-config dir). Read from the exported environment only, never from .env"
        required: false
      - name: SPEEDIANCE_TOKEN_CACHE
        description: "Override the token cache file location (default: OS user-cache dir). Read from the exported environment only, never from .env"

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 · SKILL.md (reported line 50)May include surrounding context.

md
description: "Path to config.json (overrides discovery: working dir, then OS user-config dir). Read from the exported environment only, never from .env"
        required: false
      - name: SPEEDIANCE_TOKEN_CACHE
        description: "Override the token cache file location (default: OS user-cache dir). Read from the exported environment only, never from .env"
        required: false
---

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 · internal/config/config.go (reported line 134)May include surrounding context.

go
if _, exported := exportedOnly(key); exported {
			continue
		}
		warnings = append(warnings, fmt.Sprintf("%s in .env is ignored: path settings must be exported", key))
	}
	return warnings
}

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 · internal/config/config.go (reported line 235)May include surrounding context.

go
return nil
	}
	return fmt.Errorf(
		"%w: set %s and %s (exported env vars, or a .env in the working directory), "+
			"or add \"email\"/\"password\" to config.json. config.json is resolved from "+
			"--config / %s, then ./%s, then <user-config-dir>/%s/%s (resolved this run: %s)",
		ErrMissingCredentials, EnvEmail, EnvPassword,

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

An ordinary workout statement triggers account access without a Speediance-specific request.

Medium
Category
Trigger Abuse
Confidence
94% confidence
Finding

The instruction treats a generic statement about exercising as authorization to access a connected account and retrieve all of today's detailed sessions. In an agent context, this can unnecessarily expose sensitive fitness information to the conversation or downstream processing without a request to retrieve Speediance data.

Content

No source excerpt is available for this finding.

A generic report of exercising is treated as an instruction to retrieve Speediance sessions.

Medium
Category
Trigger Abuse
Confidence
94% confidence
Finding

The machine contract directs an agent to retrieve fully resolved daily sessions whenever the client merely says they exercised. That trigger does not establish authorization to access Speediance health-related records and can cause unnecessary disclosure into agent context.

Content

No source excerpt is available for this finding.

Catalog browsing unconditionally saves the full catalog to a default local file.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest says the skill reads and emits structured data and leaves storage decisions to the caller. In internal/cli/library.go, L35–L39 unconditionally persist the full catalog, and L72 defaults the destination to library.json even when the caller only requests a search or JSON output. Thus browsing has an undeclared automatic filesystem side effect rather than requiring an explicit export destination.

Content

No source excerpt is available for this finding.