registry / @trustbaseai/claude-collab-plugin / 0.21.2

@trustbaseai/claude-collab-plugin@0.21.2

AI Security Review

scanned 2h ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. An explicit install command persists package-owned Claude hooks in `~/.claude/settings.json`. The PostTool hook forwards selected activity summaries to the configured collaboration service; optional watch mode can send message previews to user-configured webhooks and auto-start a Claude sub-session.

Static reason

One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source

Trigger

User runs `claude-collab-plugin-install`; watch and auto-wake require explicit CLI/config activation.

Impact

Bash command fragments and edit metadata can leave the Claude session through `collab-send`; enabled webhooks receive message previews.

Mechanism

Persistent Claude hooks and opt-in collaboration-message forwarding.

Rationale

Source establishes a real persistent AI-agent extension that forwards selected activity, so it warrants a warning. It is explicitly user-invoked, package-aligned, and lacks install-time execution or a hardcoded external payload/exfiltration endpoint, so blocking is not justified.

Evidence

package.jsonsrc/settings.jssrc/hooks.jsscripts/claude-posttool-hook.jsbin/claude-collab-plugin-install.jsbin/claude-collab-plugin-watch.jssrc/webhook-store.js~/.claude/settings.json~/.claude/settings.json.bak.<timestamp>~/.claude-collab/config-watch.json~/.claude-collab/webhooks.json

@trustbaseai/claude-collab-plugin@0.21.2

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code