heating the forge...

@forgecat/openai_skills_security-best-practices — Security Risk Report | ForgeCat

openai_skills_security-best-practices

Security risk report for @forgecat/openai_skills_security-best-practices v0.0.10

SRC-010Medium

Profile claims authorship by 'OpenAI' with origin repository 'https://github.com/openai/skills' and a specific commit hash (0e7823cca07bc2cbf34718a383f9ae92525be6a5).
Static scanner flagged caution, likely reflecting inability to independently verify the claimed upstream source or commit authenticity without external network access.
No evidence of typosquatting, attacker-controlled dependencies, or supply-chain poisoning in the declared metadata or reference files.

Details

Findings

Data exfiltration
skills/security-best-practices/references/javascript-typescript-react-web-frontend-security.md · SKILL-002
Data exfiltration
skills/security-best-practices/references/javascript-typescript-vue-web-frontend-security.md · SKILL-002

Low

Profile is a legitimate security best-practices skill designed to guide AI agents in performing language/framework-specific security reviews and suggesting improvements.
Reference documentation (Go, Express, React, Next.js, jQuery, vanilla JS specs) are normative security guidance documents that describe secure patterns (MUST/SHOULD/MAY requirements) and insecure patterns (for detection/audit purposes)—they do not instruct the agent to ignore its instructions, exfiltrate credentials, leak system prompts, or install malicious payloads.
The 'Overrides' section in SKILL.md acknowledges that users may request bypassing best practices for project-specific reasons, but explicitly states 'do not fight with them'—this is transparent negotiation, not hidden manipulation or guidance poisoning.

Details

Evidence

Description: 'Perform language and framework specific security best-practice reviews and suggest improvements. Trigger only when the user explicitly requests security best practices guidance...'
SKILL.md Workflow: 'The initial step for this skill is to identify ALL languages and ALL frameworks... Then check this skill's references directory to see if there are any relevant documentation...'
Go spec §0: 'MUST NOT request, output, log, or commit secrets... MUST NOT "fix" security by disabling protections...' (descriptive constraints, not instructions to violate them)

Low

Profile declares no tools, hooks, or MCP servers—it is a pure markdown/documentation skill with no executable permissions.
The skill's function is to read and analyze code in the user's repository and provide guidance; it does not request file_write, file_delete, shell, or other high-risk categories.
Authority is minimal and matches the stated purpose: analyze code, load reference documentation, and report findings to the user.

Details

Evidence

Low

No MCP servers are defined in the profile.
No external binary execution, network access, or filesystem operations are declared.
The skill operates entirely within the scope of reading and analyzing code already present in the user's repository.

Details

Evidence