Security risk report for @forgecat/openai_skills_openai-docs v0.0.11
Source Integrity
Low
Profile is sourced from OpenAI's official repository (github.com/openai/skills) with Apache-2.0 license and a specific commit hash.
Metadata indicates the source is a known, reputable vendor (OpenAI) with clear provenance and version tracking.
Agent Intent
Low
The profile provides legitimate guidance for using OpenAI documentation and APIs; it does not instruct the agent to ignore instructions, exfiltrate credentials, or leak system prompts.
Reference files (gpt-5p4-prompting-guide.md, upgrading-to-gpt-5p4.md, latest-model.md) document prompt-engineering patterns and model-selection advice without injecting malicious knowledge or weakening security defaults.
The workflow instructions (search docs, fetch sections, verify against current docs) are transparent and encourage verification against authoritative sources rather than blind obedience to embedded guidance.
Details
Evidence
SKILL.md: 'Treat OpenAI docs as the source of truth; avoid speculation.' and 'Reference files are convenience guides only; for volatile guidance such as recommended models, upgrade instructions, or prompting advice, current OpenAI docs always win.'
gpt-5p4-prompting-guide.md: Prompt blocks are presented as optional, selective tools with clear use-case boundaries (e.g., 'Use when:' conditions), not mandatory injections.
upgrading-to-gpt-5p4.md: Recommends a 'narrowest safe change set' and explicitly marks certain upgrades as 'blocked' rather than forcing unsafe changes.
Permissions
Low
The profile declares no explicit tool permissions (no shell, file_write, file_delete, or other high-risk categories).
The only MCP server defined (openaiDeveloperDocs) is scoped to OpenAI's official documentation endpoint and is consistent with the stated purpose of providing up-to-date API guidance.
The workflow relies on MCP doc-search and doc-fetch tools, which are read-only and aligned with the skill's function.
SKILL.md: 'Always use MCP doc tools before any web search for OpenAI-related questions' and fallback is restricted to 'official OpenAI domains (developers.openai.com, platform.openai.com)'.
MCP Risk
Low
The MCP server URL points to OpenAI's official developer documentation endpoint, a known-safe, read-only resource.
No hidden instructions, arbitrary binary execution, or unrestricted network/filesystem access are present in the MCP definition.
The server is used only for documentation retrieval, which is consistent with the profile's stated purpose.
Details
Evidence
MCP definition: 'openaiDeveloperDocs' with URL 'https://developers.openai.com/mcp' (official OpenAI domain, no suspicious parameters or hidden payloads).
SKILL.md tooling notes: 'Always use MCP doc tools before any web search for OpenAI-related questions' — transparent, scoped usage.
ForgeCat