📦

Audit History

tech-stack-evaluator - 8 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v8 LatestJul 9, 2026, 01:27 AM No confirmed findings0No capability change
v7 Jul 9, 2026, 01:27 AM No confirmed findings0No capability change
v6 Jul 6, 2026, 03:12 AM No confirmed findings1No capability change
v5 Jul 6, 2026, 03:12 AM No confirmed findings1Network accessExternal commands Contains scripts
v4 Jun 28, 2026, 09:09 AM 2 confirmed0No capability change
v3 Jan 16, 2026, 02:58 PM No confirmed findings0No capability change
v2 Jan 16, 2026, 02:58 PM No confirmed findings0Contains scripts
v1 Jan 15, 2026, 12:01 PM No confirmed findings0Baseline

Jul 6, 2026, 03:12 AM

No prompt injection or malicious intent was found in the reviewed files. Most static findings are false positives from dictionary keys, markdown fences, installation examples, API placeholders, and technology-evaluation wording. One medium-risk behavior remains because report_generator.py can export a report to an unconstrained caller-supplied filename.

15
Files scanned
5,083
Lines analyzed
5
Review items
1
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.

Medium
Python file write/append
with open(filename, 'w', encoding='utf-8') as f:
export_report writes generated output to the provided filename without visible path restriction. This legitimate export feature could overwrite arbitrary writable files if given an unsafe path.
Static false positives ignored (1)

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.

High
C2 keywords
"reasoning": "Scanner detected words like 'command', 'execute', and 'agent' in documentation. These
Force-confirmed blocker/high static finding; AI dismissal overridden.
Audited by: codex

Jul 6, 2026, 03:12 AM

No prompt injection or malicious intent was found in the reviewed files. Most static findings are false positives from dictionary keys, markdown fences, installation examples, API placeholders, and technology-evaluation wording. One medium-risk behavior remains because report_generator.py can export a report to an unconstrained caller-supplied filename.

15
Files scanned
5,083
Lines analyzed
5
Review items
1
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.

Medium
Python file write/append
with open(filename, 'w', encoding='utf-8') as f:
export_report writes generated output to the provided filename without visible path restriction. This legitimate export feature could overwrite arbitrary writable files if given an unsafe path.
Static false positives ignored (1)

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.

High
C2 keywords
"reasoning": "Scanner detected words like 'command', 'execute', and 'agent' in documentation. These
Force-confirmed blocker/high static finding; AI dismissal overridden.
Audited by: codex

Jun 28, 2026, 09:09 AM

The static analyzer reported a critical heuristic, but review found no command execution, network client usage, secret harvesting, obfuscation, or prompt injection attempt. Most high and medium matches are false positives from markdown examples, security terminology, URL parsing, and technology evaluation vocabulary. Residual risk is low because the skill contains Python scripts, checks one non-secret environment variable, and can write a report to a caller-provided filename.

15
Files scanned
5,083
Lines analyzed
5
Review items
3
False positives ignored

Confirmed security concerns (2)

Low
Non-Secret Environment Context Check
report_generator.py reads CLAUDE_DESKTOP to choose desktop or CLI report formatting. This is environment access, but it does not read API keys, tokens, passwords, or other secrets.
The exact environment variable name is visible and is not a credential. The value only changes output formatting behavior.
Low
User-Directed Report File Write
The report export helper writes generated markdown to the filename supplied by the caller. This is expected functionality, but callers should avoid untrusted or sensitive output paths.
The file write is explicit and limited to generated report output. Risk depends on how the caller supplies the filename, so it remains a low operational concern.
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.

Low
Static Critical Heuristic Dismissed
The scanner combined markdown backticks, documentation URLs, and an API key placeholder into a critical heuristic. Review found no subprocess, eval, exec, network client, credential read, or exfiltration behavior in the Python modules.
Direct code search found no command execution or network client usage. The cited locations are installation examples, URL input examples, and a non-secret context check.
Low
Security Terminology Misclassified as Dangerous Keywords
C2, weak cryptography, and certificate/key alerts are triggered by normal technology evaluation language such as compliance standards, encryption readiness, and scoring terms. No evidence found of malware control logic or cryptographic implementation.
The lines define compliance feature names and documentation for security scoring. The code does not open certificate files, implement weak crypto, or contact command infrastructure.
Low
Markdown Command Blocks Misclassified as Shell Execution
External command alerts are caused by markdown fences and example prompts in documentation. These examples are not executable code paths and no Python shell execution APIs were found.
The suspicious markers are markdown code fences and example text. Searches found no subprocess, os.system, popen, eval, or exec calls.
Audited by: codex

Jan 16, 2026, 02:58 PM

All 219 static findings are false positives. The scanner detected benign documentation patterns including security terminology, code block formatting, and technology names. No actual code execution, network requests, or credential access exists. Uses Python standard library only.

16
Files scanned
5,387
Lines analyzed
3
Review items
0
False positives ignored
Audited by: claude

Jan 16, 2026, 02:58 PM

All 219 static findings are false positives. The scanner detected benign documentation patterns including security terminology, code block formatting, and technology names. No actual code execution, network requests, or credential access exists. Uses Python standard library only.

16
Files scanned
5,387
Lines analyzed
3
Review items
0
False positives ignored
Audited by: claude

Jan 15, 2026, 12:01 PM

All 190 static findings are false positives. The skill is a legitimate technology evaluation tool. Flags were triggered by security/compliance documentation keywords, bash command examples in markdown, and URL references - not malicious code patterns.

14
Files scanned
4,847
Lines analyzed
2
Review items
0
False positives ignored

Risk Factors

📁 Filesystem access (1)
🔑 Env variables (1)
Audited by: claude