Skills specs-creator Audit History
📦

Audit History

specs-creator - 8 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v8 LatestJul 9, 2026, 09:00 AM No confirmed findings0No capability change
v7 Jul 9, 2026, 09:00 AM No confirmed findings0No capability change
v6 Jul 9, 2026, 04:08 AM No confirmed findings1No capability change
v5 Jul 5, 2026, 03:32 PM No confirmed findings0No capability change
v4 Jul 5, 2026, 03:32 PM No confirmed findings0 Contains scriptsFilesystem access
v3 Jun 30, 2026, 12:29 AM No confirmed findings4Contains scriptsFilesystem accessExternal commandsEnv variables
v2 Jan 23, 2026, 02:18 AM No confirmed findings0No capability change
v1 Jan 23, 2026, 02:29 AM No confirmed findings0Baseline

Jul 9, 2026, 09:00 AM

All static findings were false positives after context review. The warnings match Markdown formatting, validation-script keywords, and template examples rather than shell execution, secret access, reconnaissance, or malware behavior.

9
Files scanned
3,183
Lines analyzed
2
Review items
0
False positives ignored
Audited by: codex

Jul 9, 2026, 09:00 AM

All static findings were false positives after context review. The warnings match Markdown formatting, validation-script keywords, and template examples rather than shell execution, secret access, reconnaissance, or malware behavior.

9
Files scanned
3,183
Lines analyzed
2
Review items
0
False positives ignored
Audited by: codex

Jul 9, 2026, 04:08 AM

Most static findings are false positives from markdown formatting, placeholder template examples, and validation code. One medium-risk finding remains because the skill instructs agents to run bundled community validation scripts, which creates external command execution risk despite the scripts appearing purpose-built for local spec checks.

9
Files scanned
3,183
Lines analyzed
3
Review items
0
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
Ruby/shell backtick execution
- T004: Run Validation Scripts in `.claude/skills/specs-creator/scripts/`
The workflow explicitly tells the agent to run validation scripts from the bundled scripts directory. Running community-provided scripts is a real external command execution risk, even if the scripts appear intended for local validation.
Audited by: claude

Jul 5, 2026, 03:32 PM

All reviewed static findings are false positives based on file context. The matches are markdown formatting, template placeholders, defensive validation code, or incident response examples.

9
Files scanned
3,183
Lines analyzed
2
Review items
3
False positives ignored
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.

Critical
Ransomware keywords
- Ransomware/security breach: [incident response]
Force-confirmed blocker/critical static finding; AI dismissal overridden.
Critical
Ransomware keywords
Ransomware/security breach:
Force-confirmed blocker/critical static finding; AI dismissal overridden.
High
LLM role tokens injection
SpecType.SYSTEM: self.SYSTEM_SECTIONS,
Force-confirmed blocker/high static finding; AI dismissal overridden.
Audited by: codex

Jul 5, 2026, 03:32 PM

All reviewed static findings are false positives based on file context. The matches are markdown formatting, template placeholders, defensive validation code, or incident response examples.

9
Files scanned
3,183
Lines analyzed
2
Review items
3
False positives ignored
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.

Critical
Ransomware keywords
- Ransomware/security breach: [incident response]
Force-confirmed blocker/critical static finding; AI dismissal overridden.
Critical
Ransomware keywords
Ransomware/security breach:
Force-confirmed blocker/critical static finding; AI dismissal overridden.
High
LLM role tokens injection
SpecType.SYSTEM: self.SYSTEM_SECTIONS,
Force-confirmed blocker/high static finding; AI dismissal overridden.
Audited by: codex

Jun 30, 2026, 12:29 AM

Static analysis reported 241 issues and a 100/100 risk score, but review found the high and critical alerts are mostly keyword false positives in templates and validators. No prompt injection, network exfiltration, credential theft, destructive behavior, or confirmed malicious intent was found. The real risk is that the skill writes local spec files and asks agents to run bundled Python validation scripts, so it should publish with a filesystem and script execution warning.

9
Files scanned
3,183
Lines analyzed
8
Review items
1
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.

Medium
Bundled Validation Scripts Process Local Filesystem Paths
The skill workflow asks agents to run local Python validation scripts. The scripts read Markdown files, detect specs directories, and recursively inspect spec files with depth and result limits. This supports legitimate documentation validation but should be disclosed because it operates on local project files.
There is direct evidence of local file reads and recursive Markdown discovery. Confidence is below very high because the scripts include size, extension, symlink, depth, and result limits and no malicious intent was found.
Medium
Skill Writes Specification Files in the Workspace
The task files direct the agent to save generated PRD, technical specification, and UX specification documents under specs/. This is expected behavior for the skill, but it modifies the user workspace and relies on the overwrite safeguards in the instructions.
The file write destinations and overwrite restrictions are explicit in the task and skill instructions. This is a real workspace modification risk, but it is bounded to documentation output.
Low
Static Malware Keyword Alerts Are Template Context
Critical and high alerts for ransomware, C2, API keys, and database URLs occur in template examples for disaster recovery, deployment, and environment variable documentation. These lines are not executable and do not exfiltrate secrets.
The flagged text is clearly inside documentation templates and example sections. It names incident response and deployment concepts rather than implementing malware behavior.
Low
Backtick Shell Execution Alerts Are Markdown Formatting
The static external command findings in Markdown files are mostly inline paths, fenced examples, and task wording. They do not execute by themselves, although the workflow does separately instruct agents to run validation scripts.
The evidence is Markdown syntax and example command text, not runtime shell execution. Confidence is high, with the caveat that the skill also has an intended script-running workflow.
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.

Low
Weak Cryptography and Sensitive File Alerts Are Substring False Positives
Many high and critical findings point to words such as design, system, md, and template labels. No weak cryptographic API, Windows SAM access, certificate file read, or secret file handling was found at these reviewed locations.
The reviewed lines show enum names, file-extension patterns, section names, and UX template wording. There is no cryptographic operation or sensitive database access at those locations.

Detected Patterns

Local Python Script Execution WorkflowLocal Markdown Discovery and Validation
Audited by: codex

Jan 23, 2026, 02:18 AM

Documentation generation skill with no security concerns. All 241 static findings are false positives triggered by pattern matching in markdown templates and documentation files. The skill contains Python validation scripts and markdown templates for specification creation. No actual code execution, sensitive data access, or malicious patterns present.

9
Files scanned
3,183
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.
Audited by: claude

Jan 23, 2026, 02:29 AM

All 161 static findings evaluated as false positives. The skill consists of documentation, templates, and validation scripts for specification creation. No malicious code patterns, command injection vectors, or credential exfiltration detected. The flagged backticks are markdown code formatting, not shell execution. References to cryptographic terms are standard industry practices in documentation examples.

7
Files scanned
3,278
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.
Audited by: claude