speckit-auto

Fail

Audited by Snyk on Nov 17, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill is a legitimate, heavily guarded delivery pipeline with no obfuscation, pretext, impersonation, or covert trigger, but its default-on unattended execution of ticket-derived code under --yolo is a behavior the user may not expect, so under recall-first I rate it medium.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (medium risk: 0.30). The required --issue path has Stage 01 pass a Jira ticket URL to jira-to-speckit, which fetches the ticket's summary and description, and the ticket text enters the LLM's context. Jira tickets can be filed by outsiders, so this is outsider-authored text that the workflow consumes without the agent selecting a specific item.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

  • Potentially malicious external URL detected (medium risk: 0.30). The decisive boundary is an inbound setup fetch: Copilot and Claude Code install the superpowers skills from the obra/superpowers-marketplace plugin marketplace, whose content is adopted as installed skills and cannot be pinned or hash-checked. The publisher is an individual maintainer's repository, so it is unofficial, but its content is attributable and widely used, which makes it common/reputable unofficial and therefore medium. The other fetches are not elevated: the Spec Kit install is pinned to an enforced commit on the official github/spec-kit repo, and the superpowers clone verifies a pinned commit SHA.

MEDIUM W013: Attempt to modify system services in skill instructions.

  • Attempt to modify system services in skill instructions detected (low risk: 0.30). The skill's only machine-level actions are auto-installing a known provider package, checking the Node runtime, and writing project-scoped run state, with no sudo, user creation, system service, or SSH config changes in this file; the auto-install is the closest thing to a machine-state change, and it is the part I'd want the install reference reviewed for.

Issues (4)

E004
CRITICAL

Prompt injection detected in skill instructions.

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W012
MEDIUM

Unverifiable external dependency detected (runtime URL that controls agent).

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 17, 2026, 01:57 PM
Issues
4