Skip to content- Sources: primary, discussion
- Summary: A GitHub issue filed 2026-10-25 shows Postman's log recording an auto-existing-upgrade that installed Codex and Claude Code plugins, a remote MCP server at mcp.postman.com and a Postman CLI. The plugin's installer source calls approvePostmanHook to write a trusted_hash for its SessionStart hook in the Codex config, and the hook tells the agent to load a Postman skill for any non-trivial API task. The reporter says the evidence shows unwanted integration and instruction changes but does not establish data exfiltration.
- Why it matters: Any desktop app you run can rewrite your coding agent's global config and hook trust, so audit ~/.codex/config.toml and ~/.claude settings after updates.
- Comments: HN commenters debate whether this crosses into malware and point to the uninstall path posted in the issue thread.