land-complex-change

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation references the installation of companion tools via npx skills add crissmoldovan/agent-skills. This command downloads and executes packages provided by the author from the NPM registry. This is documented as a standard extension mechanism for the author's toolset.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests untrusted data from repository diffs, blast maps, and external report specs.
  • Ingestion points: Processes repository diffs, blast maps, and report specs (found in SKILL.md and references/worked-landing.md).
  • Boundary markers: Implements a 'side-effect budget' with explicit ALLOWED and OUT OF BOUNDS definitions to delimit acceptable changes.
  • Capability inventory: Utilizes Read, Write, Edit, Grep, Glob, and Bash tools.
  • Sanitization: Employs human-in-the-loop consent gates for irreversible actions and a mandatory stop-and-remap procedure if the defined budget is breached.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 10:04 PM
Security Audit — agent-trust-hub — land-complex-change