IGA Pages

• Outcome: Review
• Latest audit: Aug 12, 2026 · 7:25 AM UTC
• Version: 1.0.6

Overview

The skill is a coherent IGA Pages deployment helper, but it needs Review because it handles cloud credentials, remote deploys, local secrets, and tokenized preview links without enough safety scoping.

Review before installing. Use this skill only when you intentionally want IGA Pages deployment, avoid putting long-lived access keys directly in shell commands when possible, treat preview URLs with iga_token/iga_time as private, and only run env pull on trusted machines where .env.local is gitignored and protected.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The activation description uses broad everyday phrases such as 'deploy my app' and 'push this live', which can cause the skill to trigger in situations where the user did not explicitly request IGA Pages. In an agentic environment, over-broad routing can lead to unintended deployment actions, authentication prompts, or exposure of deployment URLs and project state.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill instructs users to pass access keys and secret keys directly on the command line, which exposes secrets through shell history, process listings, terminal logs, and agent traces. In a remote or shared environment, this creates a real credential leakage risk that could allow unauthorized access to the user's cloud account.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill documents 'iga pages env pull' as writing all project environment variables into .env.local but does not warn that this may materialize sensitive secrets onto disk. That increases the chance of accidental local disclosure through backups, editor tooling, file sharing, or mistaken commits if ignore rules are misconfigured.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill explicitly tells the agent to share full preview URLs containing 'iga_token' and 'iga_time' query parameters without warning that the tokenized URL may grant access. If such URLs are posted in chat logs, tickets, or shared channels, anyone with the link may access a protected preview deployment.

Autonomous Decision Making

Medium

• Category: Excessive Agency

Content

1. Ensure logged in with `iga whoami`; only run `iga login` if `whoami` fails.
2. If the directory is not linked yet, run `iga pages link` to create/associate the Pages project (this no longer deploys). If it is already linked, continue without relinking.
3. Inspect the project for supported integrations. If Supabase usage is detected, run `iga pages integration list` first. If the required Supabase binding already exists, continue to deploy without asking again. If no matching binding exists, ask before deploying:
   > "This project appears to use Supabase, which IGA Pages can connect as an integration. Would you like to connect it before deploy?"
4. If yes (Supabase): run `iga pages integration link supabase`. The synced env vars take effect on the next deploy automatically.
5. `iga pages deploy`.

• Confidence: 82% confidence
• Finding: The orchestration directs the agent to inspect the project, check existing integrations, and continue deployment without asking again when a matching binding exists. That enables autonomous progression through deployment-related actions based on inferred project state, which can cause unintended external changes or disclosures if the user's intent was ambiguous or the detected project context is wrong.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.