Static command-execution alerts are false positives caused by Markdown backticks and fenced examples. The skill performs scoped, read-only access to a documented local Claude Router stats file and shows no network transfer, writes, prompt injection, or semantic abuse.
Static command-execution alerts are false positives caused by Markdown backticks and fenced examples. The skill performs scoped, read-only access to a documented local Claude Router stats file and shows no network transfer, writes, prompt injection, or semantic abuse.
Most external command findings are false positives caused by Markdown backticks and code fences in documentation. The meaningful risk is intentional local filesystem access to ~/.claude/router-stats.json, which may reveal router usage and cost history but shows no network exfiltration or prompt injection.
1
Files scanned
99
Lines analyzed
6
Review items
0
False positives ignored
Capability review items (4)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Read the stats file at `~/.claude/router-stats.json` and present the data in a clear, formatted way.
The skill explicitly instructs reading ~/.claude/router-stats.json from the user's home directory. This is scoped to a local stats file, but it is still hidden-file access that can expose usage history.
1. Use the Read tool to read `~/.claude/router-stats.json`
The step directs the Read tool to access ~/.claude/router-stats.json. Although intentional for the skill, it relies on local filesystem reads from a hidden directory.
Read the stats file at `~/.claude/router-stats.json` and present the data in a clear, formatted way.
The instruction targets a hidden Claude stats file in the user's home directory. The file may contain private usage and savings information, so the filesystem access is real.
1. Use the Read tool to read `~/.claude/router-stats.json`
The skill tells the assistant to read a hidden local stats path with the Read tool. This is expected behavior, but it remains a privacy-relevant filesystem access.
Most external command findings are false positives caused by Markdown backticks and code fences in documentation. The meaningful risk is intentional local filesystem access to ~/.claude/router-stats.json, which may reveal router usage and cost history but shows no network exfiltration or prompt injection.
1
Files scanned
99
Lines analyzed
6
Review items
0
False positives ignored
Capability review items (4)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Read the stats file at `~/.claude/router-stats.json` and present the data in a clear, formatted way.
The skill explicitly instructs reading ~/.claude/router-stats.json from the user's home directory. This is scoped to a local stats file, but it is still hidden-file access that can expose usage history.
1. Use the Read tool to read `~/.claude/router-stats.json`
The step directs the Read tool to access ~/.claude/router-stats.json. Although intentional for the skill, it relies on local filesystem reads from a hidden directory.
Read the stats file at `~/.claude/router-stats.json` and present the data in a clear, formatted way.
The instruction targets a hidden Claude stats file in the user's home directory. The file may contain private usage and savings information, so the filesystem access is real.
1. Use the Read tool to read `~/.claude/router-stats.json`
The skill tells the assistant to read a hidden local stats path with the Read tool. This is expected behavior, but it remains a privacy-relevant filesystem access.
The markdown code-fence and inline-backtick detections are false positives, with no external command execution present. The skill does instruct the assistant to read a hidden home-directory stats file, which is intended but still exposes local usage metadata.
1
Files scanned
99
Lines analyzed
6
Review items
0
False positives ignored
Capability review items (4)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Static findings for Ruby or shell backtick execution, weak cryptography, and system reconnaissance are false positives caused by Markdown fences, display text, and model names. The confirmed issue is intentional access to a hidden home-directory stats file, which is legitimate for this skill but should be disclosed to users.
1
Files scanned
99
Lines analyzed
2
Review items
3
False positives ignored
Capability review items (1)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
The skill instructs the assistant to read ~/.claude/router-stats.json. This is necessary for the stated purpose, but it accesses a hidden file under the user's home directory and may expose usage patterns or cost estimates.
The file path is explicitly named twice and the skill's steps direct the assistant to read it. The usage is semantically aligned with the skill, so the concern is disclosure and local-file scope rather than confirmed malicious intent.
Static false positives ignored (3)
These static matches were dismissed by semantic review or matched schema-only tokens, so they are shown for transparency but do not drive the quality score.
The static analyzer flagged Markdown backticks as Ruby or shell backtick execution. No executable shell command, Ruby code, or user-controlled command construction is present in the reviewed skill text.
The flagged lines are inline code formatting or Markdown code fences. They do not define or invoke a command execution primitive.
The static weak-cryptography hits do not identify a cryptographic function, algorithm use, key handling, or encoded payload. The flagged lines are metadata and plain output text.
The reviewed lines contain description and display text only. I did not find evidence of MD5, SHA1, DES, RC4, or any cryptographic operation.
The static reconnaissance hit is located inside a formatted cost-savings example. No system information command, host enumeration, process listing, or environment inspection is requested.
The flagged line is sample display output for delegation savings. It has no semantic connection to system reconnaissance behavior.
Pure prompt-based skill with no executable code. Only reads a single JSON file containing router statistics. No network access, no command execution, no sensitive data access. All 25 static findings are false positives caused by the scanner misidentifying markdown documentation examples and metadata fields as executable code.
Pure prompt-based skill with no executable code. Only reads a single JSON file containing router statistics. No network access, no command execution, no sensitive data access. All 25 static findings are false positives caused by the scanner misidentifying markdown documentation examples and metadata fields as executable code.
Pure prompt-based skill with no executable code. Only reads a single JSON file containing router statistics. No network access, no command execution, no sensitive data access.
1
Files scanned
99
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.