OpenAI Developer Docs

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

The agent may run the included helper script to retrieve documentation results.

Why it was flagged

The skill directs the agent to use a local shell wrapper. This is disclosed and central to the documentation-query purpose, with no artifact evidence of destructive or unrelated commands.

Skill content
Query the MCP server via the CLI wrapper in `scripts/openai-docs-mcp.sh`
Recommendation

Allow the helper only for OpenAI documentation lookups and review returned source URLs when accuracy matters.

What this means

The skill may fail if curl or jq are missing, and users may not see the optional endpoint override in metadata.

Why it was flagged

The included script requires curl and jq and supports MCP_URL, so the registry metadata under-declares runtime dependencies/configuration. The behavior is still disclosed in SKILL.md/script and does not show hidden installation or remote code execution.

Skill content
Required binaries (all must exist): none; Required binaries (at least one): none; Env var declarations: none
Recommendation

Ensure curl and jq are installed before use, and only set MCP_URL to an endpoint you trust.

What this means

Your documentation queries and fetched doc URLs are sent to the configured MCP server.

Why it was flagged

The script sends JSON-RPC payloads to an external MCP endpoint, with an environment variable allowing endpoint override. This is expected for a docs MCP client, but it means search terms and requested URLs leave the local environment.

Skill content
MCP_URL="${MCP_URL:-https://developers.openai.com/mcp}" ... -X POST "$MCP_URL" ... -d "$payload"
Recommendation

Use the default OpenAI endpoint unless you trust an override, and do not include secrets or private project data in documentation search queries.