Factory AI Droid

Use Factory AI's droid CLI to interactively build, debug, refactor, review, and deploy code, with support for plugins, MCP servers, and multiple AI models.

Security Scan

The skill is an instruction-only wrapper around an external 'droid' CLI that references API keys, plugins, and MCP servers but fails to declare or justify the credentials and system access it will need, creating a mismatch between its claims and the actual operational surface.

Purpose & Capability

The SKILL.md describes using a third‑party 'droid' CLI to perform code changes, commits, PRs, and deployments. The top-level registry metadata lists no required binaries or env vars, but the SKILL.md metadata and examples require the 'droid' binary and mention FACTORY_API_KEY and deployment targets (fly.io). That mismatch (undeclared binary/env requirements and a missing description) is unexplained and disproportionate to an instruction-only wrapper.

Instruction Scope

Runtime instructions tell the agent to run 'droid' commands that operate on the codebase (multi-file edits, commit/PR generation, deploy) and to add MCP servers and plugins. The skill text does not constrain where data is sent or how sessions are stored; adding MCP servers or plugins could direct code and secrets to arbitrary external servers. The SKILL.md also contains a hard-coded local install path (/Users/mitchellbernstein/...), suggesting leftover local artifacts rather than a vetted, general instruction set.

Install Mechanism

There is no install spec and no code files — the skill is instruction-only, so it does not write files or pull arbitrary archives. This is lower-risk from an installation perspective. Note: the README's claim that 'droid' is already installed at a specific user path is a suspicious local artifact but not an install mechanism.

Credentials

The instructions explicitly reference FACTORY_API_KEY and a 'droid login' flow and imply other credentials may be required for deploy targets (e.g., fly.io), but the registry metadata declares no required environment variables or primary credential. Requesting API keys and enabling deployment without declaring or justifying those secrets is disproportionate and unexplained.

Persistence & Privilege

The skill does not request always:true and does not declare config path access. However, the ability it exposes to add MCP servers and plugins (via the 'droid' tool) could expand system reach or persistence outside the agent, depending on how the droid CLI manages plugins/servers. This is a capability-level risk rather than an explicit metadata claim by the skill.

What to consider before installing

This skill is basically an instruction sheet for a separate 'droid' CLI — it does not include or declare that CLI or the credentials it references. Before installing or using it: 1) Verify the provenance of the 'droid' CLI and inspect its binary/source — do not run it until you trust it. 2) Do not provide FACTORY_API_KEY or deploy credentials until the author explains what servers the CLI talks to and how data is handled. 3) Avoid adding MCP servers or plugins from unknown sources; those can forward code or secrets externally. 4) If you want to test, run the CLI in a sandbox or on a read-only copy of your repo and monitor network traffic. 5) Ask the skill author for a clear description, a declared list of required binaries/env vars, and where session data and network traffic are sent. These clarifications would raise confidence; until then treat the skill cautiously.

Factory AI Droid CLI

Use droid to build features, debug, refactor, and deploy code.

Installation

Already installed at: /Users/mitchellbernstein/.local/bin/droid

Authentication

droid login
# or set FACTORY_API_KEY env var
export FACTORY_API_KEY=your-api-key

Commands

Interactive Mode

droid                           # Start fresh session
droid "fix the login bug"       # Start with prompt
droid -r                        # Resume last session
droid -r session-id             # Resume specific session

Non-Interactive (Exec Mode)

droid exec "analyze this file"
droid exec "commit my changes with a good message"
droid exec "deploy to fly.io"
droid exec --help               # Show exec options

Options for Exec

droid exec --force "fix lint errors"    # Auto-apply without confirmation
droid exec --json "analyze code"        # JSON output
droid exec --model claude "task"        # Specify model

MCP Servers

droid mcp list                    # List installed MCP servers
droid mcp add server-name         # Add MCP server
droid mcp remove server-name      # Remove MCP server

Plugins

droid plugin list                 # List plugins
droid plugin add name             # Add plugin

Usage Patterns

Feature Development

droid exec "add a user settings page with dark mode toggle"

Debugging

droid exec "fix this error: [paste error]"

Code Review

droid exec "review the PR for security issues"

Git Operations

droid exec "create a PR for my changes"
droid exec "write a good commit message for the staged changes"

Deployment

droid exec "deploy to fly.io"

Multi-file Changes

droid
# Then in interactive mode:
@src/components/
@src/api/
Implement authentication flow

Notes

• Droid has deep codebase understanding across your org
• Supports model flexibility (OpenAI, Anthropic, xAI, etc.)
• MCP servers for extended capabilities
• Session-based memory for context continuity