registry / @agentteams/runner / 0.0.98

@agentteams/runner@0.0.98

AgentRunner - Background runner that polls and executes AI agent tasks

AI Security Review

scanned 1h ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. After explicit initialization, this package persists a per-user daemon that polls the AgentTeams API and runs received AI-agent tasks. Its generated service configuration disables Codex sandboxing for spawned Codex runs.

Static reason

No blocking static signals were detected.; previous stored version diff introduced dangerous source

Trigger

User runs `agentrunner init --token <token>` without `--no-autostart`, then the registered service starts or restarts.

Impact

Remote task prompts can cause the selected local agent CLI to act in its working directory with reduced Codex safety controls.

Mechanism

First-party autostart daemon polling remote triggers and launching AI-agent CLIs with Codex sandbox bypass enabled.

Rationale

No unconsented npm install hook or concrete theft/destructive chain was found, so blocking is not warranted. The explicit first-party autostart plus default Codex sandbox bypass is a real dangerous capability that merits a warning.

Evidence

package.jsondist/commands/init.jsdist/commands/start.jsdist/autostart.jsdist/api-client.jsdist/handlers/trigger-handler.jsdist/runners/codex.js~/.agentteams/daemon.json~/Library/LaunchAgents/run.agentteams.runner.plist~/.config/systemd/user/agentrunner.service~/.agentteams/agentrunner-start.ps1~/.agentteams/agentrunner-task.xml

@agentteams/runner@0.0.98

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code