A setup wizard for checking which of four coding-agent command-line tools—Codex, Cursor, Antigravity, and OpenCode—are installed, then configuring Exa and Context7 connections for them.

Detecting supported CLIs, configuring Exa and Context7 MCPs, previewing setup changes, and enabling or disabling the review gate.

It avoids checking each tool and configuration separately, while also supporting a dry run and a review-gate toggle.

setup — Command by greenpolo

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/greenpolo/cc-multi-cli-plugin/setup

Clone the repo

git clone --depth 1 https://github.com/greenpolo/cc-multi-cli-plugin

Per session 20 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 9,304 The whole file, excluding the scripts and references it only reads on demand.

Security scan F 6 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00020	$0.09304
Opus 5	$0.00010	$0.04652
Sonnet 5	$0.00004	$0.01861
Haiku 4.5	$0.00002	$0.00930

Measured yesterday against content hash 486443d1124d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade F, and why

setup scanned grade F with 6 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Harvests environment variableshighData exfiltration

Enumerating or grepping the environment for keys collects credentials unrelated to what the mod says it does.

## Step 3 — Collect API keys (Exa required, Context7 optional)

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

Create the directory with `Bash` if it doesn't exist. **File permissions:** on Unix, `chmod 600` the file. On Windows, skip the chmod — NTFS default ACLs on user-profile paths already restrict to that user. Note in your

Downloads and executes remote codehighSupply chain

curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.

- Cursor: `curl https://cursor.com/install -fsS | bash` (Unix) or `irm 'https://cursor.com/install?win32=true' | iex` (Windows PowerShell). After install, the binary lives at `$LOCALAPPDATA/cursor-agent/agent.cmd` on Win

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- `~/.codex/config.toml` — TOML parse; look for `mcp_servers.<any-name>.env.EXA_API_KEY` (any exa-flavored server) and `.CONTEXT7_API_KEY`.

Reads MCP configurationmediumAgent snooping

mcp.json carries server URLs and auth tokens; reading it lets a mod discover and abuse other integrations.

2. **Claude Code's project-level MCP config:** `~/.claude/.mcp.json` (parse JSON; look for any `mcpServers.<name>.env.EXA_API_KEY` and `.CONTEXT7_API_KEY`).

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- Cursor: `curl https://cursor.com/install -fsS | bash` (Unix) or `irm 'https://cursor.com/install?win32=true' | iex` (Windows PowerShell). After install, the binary lives at `$LOCALAPPDATA/cursor-agent/agent.cmd` on Win

plugins/multi/commands/setup.md · 489 lines

How it starts

The opening of the file, as written. The whole thing — 489 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Review-gate toggle (short-circuit)

If $ARGUMENTS contains --enable-review-gate or --disable-review-gate, this is a review-gate toggle, NOT a full setup run. Do ONLY this, then stop:

node "${CLAUDE_PLUGIN_ROOT}/scripts/multi-cli-companion.mjs" setup --json $ARGUMENTS

Present the resulting status (it reflects the new review-gate state) and do not run any of the wizard steps below.

You are running the setup wizard for cc-multi-cli-plugin. Work through these steps in order, using your native tools. Do NOT invoke subagents. Do NOT run the plugin's companion scripts (except the review-gate short-circuit above and the Antigravity detection probe noted in Step 1).

If the user passed --dry-run anywhere in $ARGUMENTS, enumerate changes but make NO writes.

Step 1 — Detect installed CLIs

This plugin supports four CLIs: Codex, Cursor, Antigravity, and OpenCode.

Run each probe via Bash:

codex --version
Cursor: the binary is named agent (not cursor-agent). Try agent --version first. On Windows the installer does NOT add it to PATH — always fall back to $LOCALAPPDATA/cursor-agent/agent.cmd --version (a.k.a. C:/Users/<name>/AppData/Local/cursor-agent/agent.cmd). Remember whichever path works; use it throughout the rest of setup.
Antigravity: the binary is agy (Google's Antigravity CLI). Try agy --version. On Windows the installer drops it at $LOCALAPPDATA/agy/bin/agy.exe and may not add it to PATH. The companion's setup probe (node "${CLAUDE_PLUGIN_ROOT}/scripts/multi-cli-companion.mjs" setup --json) reports whether agy is detected. (EXPERIMENTAL; read-only research/explore only.)
OpenCode: the binary is opencode. Try opencode --version. On Windows, npm installs a .cmd shim; a stale bun opencode.exe in PATH may shadow it — the adapter resolves only the .cmd shim, never the .exe. Install via npm install -g opencode-ai.

Tabulate which succeed. For each failure, tell the user the install command:

Read the full file on GitHub · 489 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 489 lines · 20 tokens per session scan F 486443d1124d

Subscribe to this mod's changes

setup is a command published in the GitHub repository greenpolo/cc-multi-cli-plugin (89 stars, last pushed 15d ago), licensed Apache-2.0. It adds 20 tokens to every session and 9,304 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it F with 6 findings (harvests environment variables, asks for root, downloads and executes remote code). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Code review acp agent-client-protocol ai-coding anthropic claude-code claude-code-plugin

setup

Getting it into your agent

What it costs to keep this loaded

Grade F, and why

Harvests environment variableshighData exfiltration

Asks for rootmediumPrivilege escalation

Downloads and executes remote codehighSupply chain

Reads agent configuration directoriesmediumAgent snooping

Reads MCP configurationmediumAgent snooping

Makes network callslowCapability

How it starts

Review-gate toggle (short-circuit)

Step 1 — Detect installed CLIs

What this file has done since we first saw it

Other commands, from other repositories

delegate

icpg-bootstrap

version

go-review

release-harn

esp-harden