Back to skill

Security audit

Atomcode Delegation

Security checks for vulnerabilities and agentic risk

Overview

This skill is a real AtomCode delegation guide, but it encourages auto-approved background code-editing agents and parallel runs in the same workspace without enough containment guidance.

Install only if you intentionally want an agent to delegate code edits to AtomCode. Use a clean branch or separate worktree per task, verify the `-C` workdir before running, avoid `-y` until you understand AtomCode’s permissions, and review `git status` and diffs before keeping any changes.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The README shows headless, auto-approved (`-y`) execution that can modify project files, but it does not warn users that the delegated agent may perform unattended code changes in the target workspace. In a skill specifically designed to delegate coding tasks, this omission increases the chance of users running destructive or unintended edits without review, especially when combined with broad prompts or incorrect working directories.

Missing User Warnings

High
Confidence
98% confidence
Finding
The batch example launches multiple background agent processes against the same `$WORKDIR` without warning about concurrent writes, race conditions, merge conflicts, or corruption of in-progress changes. Because these are autonomous code-modifying tasks running in parallel, the skill context makes this more dangerous than a normal shell concurrency example: several agents may edit overlapping files simultaneously and leave the repository in an inconsistent state.

Static analysis

No suspicious patterns detected.