new-website
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONOBFUSCATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses various scripts to automate development workflows:
- Node.js scripts in
scripts/(e.g.,verify.mjs,wire-hooks.mjs,build-marker.mjs,anchor-ids.mjs) usespawnSyncandexecSyncto managenpm,git, andastrobuild processes. - Shell scripts (
ship.sh,check_external_links.sh,check_internal_links.sh) usegit,curl, andawkfor deployment, external link auditing, and internal link graph analysis. - Python scripts (
generate_og_cards.py,set_pdf_title.py) perform specialized tasks like generating social share images and updating PDF metadata. - [OBFUSCATION]: The skill implements email address obfuscation in
templates/astro/src/lib/obfuscate.tsandtemplates/astro/src/components/EmailLink.astro. This uses Base64 encoding and string reversal to prevent harvest bots from scraping plaintext addresses from the built HTML. This is a common and safe web development practice. - [DYNAMIC_EXECUTION]: Shell commands and scripts are executed dynamically during the build and deployment phases. For example,
ship.shPromotes a preview build to production by pushing git refs, andverify.mjsexecutes the site's test suite vianpm test. These operations are transparent and consistent with the skill's purpose as a developer toolkit. - [INDIRECT_PROMPT_INJECTION]: The skill features capabilities to ingest and analyze external data, creating a surface for indirect injection if that data is malicious.
- Ingestion points:
scripts/cloudflare-check.mjs(fetches and parses user-provided domains) andscripts/check_external_links.sh(crawls external URLs to check for dead links). - Boundary markers: Absent, as the analysis is structural (e.g., checking HTTP headers or link status) rather than linguistic.
- Capability inventory: Extensive system capabilities including file writes, network requests, and subprocess execution are present, as required for a build and audit tool.
- Sanitization: The skill uses standard libraries (e.g., Node's
new URL()andhttpmodules) to validate and sanitize domain inputs and URLs before processing.
Audit Metadata