AI Agent Security Skills

Detects leaked API keys, tokens, and credentials in agent workspaces and logs.

Scans package manifests for known CVEs and license violations in real time.

Blocks prompt injection attacks targeting AI agent instruction files and tool calls.

Applies seccomp profiles, drops capabilities, and enforces read-only filesystems.

Tracks outbound connections from agent containers and flags anomalous C2 traffic.

Audits Model Context Protocol tool calls for over-permissioned or malicious actions.

Connects agents to GitHub repos for automated security scanning on every push.

Monitors GitHub Actions workflows for injection vulnerabilities and mutable tag abuse.

Sends real-time security alerts and incident summaries to Slack channels.

Escalates critical security incidents to PagerDuty with severity mapping.

Posts formatted security alerts to Discord servers with embed support.

Delivers security notifications to Microsoft Teams channels via webhooks.

Routes agent security alerts to OpsGenie with priority mapping and on-call scheduling.

Builds behavioral baselines for AI agents and flags deviations from normal patterns.

Detects slopsquatting, dependency confusion, and typosquatting in agent packages.

Parses agent runtime logs and surfaces security-relevant events with structured output.

Automatically applies security patches to agent dependencies and base images.

Generates SOC 2 and ISO 27001 compliance reports from agent security posture data.

Records tamper-proof audit trails for all agent actions, config changes, and access events.

Manages network firewall rules for agent containers with allowlist-only egress policies.

Schedules encrypted backups of agent state, configs, and credential vaults to R2.

Automatically rotates API keys, tokens, and secrets on a configurable schedule.