Kilo CLI Coding Agent

Security checks across malware telemetry and agentic risk

Overview

The skill is coherent for running Kilo CLI, but it asks for broad GitHub write access and encourages autonomous background agents that can change code, push branches, and post to GitHub.

Install only if you trust Kilo CLI and are comfortable supervising autonomous coding agents. Use a temporary worktree or container, provide the narrowest possible GitHub token, and require manual review before pushes, PR creation, PR comments, or workflow-affecting actions.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI09: Human-Agent Trust Exploitation
Medium
What this means

A user may believe private files outside the project are protected when the skill only documents changing the process working directory.

Why it was flagged

The text frames the working directory as a containment boundary, but the artifacts do not show an enforced sandbox or filesystem restriction; a normal CLI process can still access other paths if permitted.

Skill content
# Start agent in target directory ("little box" - only sees relevant files)
**Why workdir matters:** Agent wakes up in a focused directory, doesn't wander off reading unrelated files
Recommendation

Treat workdir as a convenience, not a sandbox; run in a container, temporary clone, or restricted account when handling sensitive files.

#
ASI02: Tool Misuse and Exploitation
Medium
What this means

The agent could push code or project changes to GitHub under the user's credentials before the user has reviewed the result.

Why it was flagged

The workflow tells the agent to run Kilo in a way that can install dependencies, modify code, commit, and push changes, with no explicit approval checkpoint before the remote mutation.

Skill content
tmux -S "$SOCKET" send-keys -t fix-78 "cd /tmp/issue-78 && npm install && kilo run 'Fix issue #78: <description>. Commit and push.'" Enter
Recommendation

Require explicit user confirmation before any git push, PR creation, PR comment, dependency install, or other remote/account-changing action.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

If misused, the token could allow repository changes, workflow-related actions, or public GitHub activity beyond the immediate coding task.

Why it was flagged

The skill requires a powerful GitHub token with repository and Actions write authority, but the artifacts do not clearly limit it to one repository, branch, or operation.

Skill content
"description": "GitHub Personal Access Token with repo and actions:write permissions", "required": true
Recommendation

Use a fine-grained, least-privilege GitHub token scoped to the specific repository and task, avoid actions:write unless truly needed, and rotate/revoke it after use.

#
ASI10: Rogue Agents
Low
What this means

Multiple Kilo sessions may keep running, consume resources, or continue working on repositories after the user has moved on.

Why it was flagged

Background and parallel agent sessions are disclosed and central to the skill, but they create long-running activity that needs monitoring and cleanup.

Skill content
Use **bash background mode** for non-interactive coding work.
Can run 5+ parallel fixes!
Recommendation

Monitor sessions with the documented process/tmux commands and kill or clean up all background sessions when the task is complete.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

The behavior of the external Kilo CLI package is outside this skill's static scan and must be trusted separately.

Why it was flagged

The skill depends on a globally installed external CLI package that is not pinned or included in the reviewed artifacts.

Skill content
npm install -g @kilocode/cli
Recommendation

Install Kilo CLI only from a trusted source, consider pinning a known-good version, and review its permissions and configuration before use.