Skills
skills/ahgraber/skills/spec-kit-tasks/Gen Agent Trust Hub

spec-kit-tasks

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill is highly vulnerable to instructions embedded in plan.md or spec.md (untrusted data). Ingestion points: plan.md and spec.md are loaded into the agent context in workflow steps 2 and 3. Boundary markers: None; the agent is not instructed to ignore content within these files. Capability inventory: Script execution (check-prerequisites.sh) and filesystem write (tasks.md). Sanitization: None; external content is directly transformed into executable tasks. An attacker could influence the generated tasks or the prerequisite check phase.
  • [Command Execution] (MEDIUM): The skill requires executing scripts/check-prerequisites.sh, which contains a pointer to an external script located at ../../spec-kit/scripts/check-prerequisites.sh. Executing code from paths outside the skill's controlled environment is a significant security risk.
  • [External Downloads] (LOW): References to the github/spec-kit repository are present. Since the github organization is not on the trusted sources list, these references are noted as unverified dependencies.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 06:40 AM