← sources

Postman desktop automatically installed a Claude Code plugin on my Mac. Logs, config and installer source evidence.

source: reddit (reddit:ClaudeAI)published 2026-10-25 19:07ingested 2026-10-26 02:36

retained text

A heads-up for Claude Code users who also have Postman installed: after I opened Postman on my Mac for the first time in a while, a local audit found that its desktop auto-installer had installed and enabled a Claude Code plugin at user scope. The same run installed a Codex plugin too.

I asked Codex to inspect the logs, configuration and cached installer source. I consider an API client changing unrelated coding-agent setups a serious consent and trust problem. I have removed the app and integrations.

## Direct Claude Code evidence

Postman's main.log at 2026-10-24T09:34:14.418Z records:

```text
PostmanAgentPluginService~install: script completed
{"source":"auto-existing-upgrade","results":["claude-code:installed","codex:installed"],"installer":"script"}
```

The next entry says autoInstall.plugin.existing-user finished with outcome installed, signedIn true, useScript true, and scriptResults claude-code:installed,codex:installed.

~/.claude/settings.json enabled `postman@claude-plugins-official`. Its entry in ~/.claude/plugins/installed_plugins.json recorded:

```json
{
  "scope": "user",
  "version": "2.0.1",
  "installedAt": "2026-10-24T09:34:02.610Z",
  "lastUpdated": "2026-10-24T09:34:02.610Z",
  "gitCommitSha": "c6352ef5605cd46d3f817e15c601eeb9a56c08f7"
}
```

Cache: ~/.claude/plugins/cache/claude-plugins-official/postman/2.0.1. The installed Postman installer source has a Claude route that installs/updates this plugin at user scope:
https://github.com/postmanlabs/postman-plugin/blob/6c8e452e0f4237f4bcc0901cb1ac442dec707e49/installer/src/hosts/claude-code.ts

## The accompanying Codex integration

Codex's global config enabled Postman's plugin and held hook trust hashes. Its startup hook tells the agent to use Postman's API skill for most non-trivial API engineering tasks. The installer attempts to approve that hook by reading its hash with hooks/list and writing trusted_hash through config/batchWrite:
https://github.com/postmanlabs/postman-plugin/blob/6c8e452e0f4237f4bcc0901cb1ac442dec707e49/installer/src/hosts/codex-hooks.ts

That approval routine is specific to Codex. It does not prove Claude's hook approval was bypassed, and the local trust record alone cannot establish who created it.

Full evidence, including both auto-install log entries, Codex config/trust hashes, startup instructions, MCP auth failure and pinned source links:
https://github.com/postmanlabs/postman-app-support/issues/14370

Environment: macOS, Postman desktop 12.33.4, Claude plugin 2.0.1, Codex plugin 1.1.0. A Postman CLI was also found with modification times matching the install window. This evidence establishes installation/configuration changes; it does not establish data exfiltration.

## Check your setup

Inspect ~/.claude/settings.json and ~/.claude/plugins/installed_plugins.json. For an unwanted user-scope plugin, the installer's own removal route uses:

```sh
claude plugin uninstall postman@claude-plugins-official --scope user --json
```

Restart Claude Code afterwards. Also check ~/.codex/config.toml and ~/.local/bin/postman. Deleting Postman.app alone leaves agent integrations behind.

Postman should stop automatic installation into coding agents, explain the exact opt-in and files affected, leave hook trust decisions to the user in the agent host, and provide complete rollback. I want integrations I explicitly choose, not global changes discovered after another tool starts behaving differently.

Credentials, account identifiers and unrelated private logs are omitted.

document fetched 2026-10-26 00:21 · canonical https://reddit.com/r/ClaudeAI/comments/1xfauxa/postman_desktop_automatically_installed_a_claude