atlas-resume

Fail

Audited by Snyk on Nov 22, 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 documented session-resume procedure that reads on-disk ledgers, keeps the same run, and stops at the approval gates, with no concealment, pretext, or covert trigger; its one possible concern is the forced removal of possibly dirty worktrees in step 6g, which is a destructive step taken without a check, so it is flagged as medium under the recall-first rule.

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

  • Attempt to modify system services in skill instructions detected (low risk: 0.10). The skill directs the agent to rewrite the run's plan.dag.json state and force-remove dirty git worktrees (local task-local operational changes), but the evidence shows no explicit irreversible loss of meaningful state, since the worktrees are re-created at baseline and the DAG reset is recoverable.

Issues (2)

E004
CRITICAL

Prompt injection detected in skill instructions.

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 22, 2026, 01:13 AM
Issues
2