docs: align every roster introduction on the eight Participant CLIs

AGENTS.md

@@ -4,10 +4,12 @@ This file provides guidance to coding agents when working with code in this repo
 
 ## What this is
 
-`crew` is a local coordination CLI for terminal coding agents (Claude Code, Codex CLI,
-Gemini CLI, Copilot CLI, Antigravity CLI, Pi CLI, opencode CLI). It gives independently running sessions a shared inbox, a
-reviewed task workflow, and an optional tmux launcher. **crew never calls a model
-provider itself** — it only coordinates via a workspace-local SQLite database.
+`crew` is a local coordination CLI for terminal coding agents — the eight Participant CLIs
+listed in `src/participants.ts` (Claude Code, Codex CLI, Gemini CLI, Copilot CLI,
+Antigravity CLI, Pi CLI, Little Coder, opencode CLI). It gives independently running
+sessions a shared inbox, a reviewed task workflow, and an optional tmux launcher. **crew
+never calls a model provider itself** — it only coordinates via a workspace-local SQLite
+database.
 
 The package publishes as `@dichovsky/crew` but the installed executable is `crew` (the
 unscoped npm name was unavailable — see DEC-7 in `docs/design/decisions.md`).

CLAUDE.md

@@ -4,10 +4,12 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 
 ## What this is
 
-`crew` is a local coordination CLI for terminal coding agents (Claude Code, Codex CLI,
-Gemini CLI, Copilot CLI, Antigravity CLI, Pi CLI, opencode CLI). It gives independently running sessions a shared inbox, a
-reviewed task workflow, and an optional tmux launcher. **crew never calls a model
-provider itself** — it only coordinates via a workspace-local SQLite database.
+`crew` is a local coordination CLI for terminal coding agents — the eight Participant CLIs
+listed in `src/participants.ts` (Claude Code, Codex CLI, Gemini CLI, Copilot CLI,
+Antigravity CLI, Pi CLI, Little Coder, opencode CLI). It gives independently running
+sessions a shared inbox, a reviewed task workflow, and an optional tmux launcher. **crew
+never calls a model provider itself** — it only coordinates via a workspace-local SQLite
+database.
 
 The package publishes as `@dichovsky/crew` but the installed executable is `crew` (the
 unscoped npm name was unavailable — see DEC-7 in `docs/design/decisions.md`).

README.md

@@ -5,11 +5,12 @@
 [![Node](https://img.shields.io/badge/node-%3E%3D24.15-brightgreen.svg)](https://nodejs.org)
 
 crew is a command-line tool that coordinates AI coding assistants running in a terminal.
-Tools such as Claude Code, Codex CLI, Gemini CLI, GitHub Copilot CLI, and Little Coder each
-run as their own independent terminal session; crew gives those sessions a shared inbox for
-notes, a task workflow where finished work must pass a separate review, and an optional
-launcher that opens all of them side by side. crew never contacts an AI model provider
-itself — it only passes coordination data between the sessions.
+The eight supported tools — Claude Code, Codex CLI, Gemini CLI, GitHub Copilot CLI,
+Antigravity CLI, Pi CLI, Little Coder, and opencode CLI — each run as their own independent
+terminal session; crew gives those sessions a shared inbox for notes, a task workflow where
+finished work must pass a separate review, and an optional launcher that opens all of them
+side by side. crew never contacts an AI model provider itself — it only passes coordination
+data between the sessions.
 
 **[Interactive documentation →](https://dichovsky.github.io/crew/)** — how crew works, what it is
 built from, and where its limits are, explained with diagrams you can step through.
@@ -22,9 +23,10 @@ built from, and where its limits are, explained with diagrams you can step throu
 - **A task isn't done because the agent says so.** The Worker (the agent assigned the work)
   finishing and the reviewer accepting the result are two separate, enforced steps
   (`submitted` then `completed`), so a Worker can't mark its own work accepted.
-- **Not locked to one vendor.** Claude Code, Codex CLI, Gemini CLI, Copilot CLI, Little
-  Coder, and locally hosted models can all join the same Crew — the group of agents sharing
-  one project's state. crew coordinates them; it doesn't replace them.
+- **Not locked to one vendor.** Claude Code, Codex CLI, Gemini CLI, Copilot CLI, Antigravity
+  CLI, Pi CLI, Little Coder, opencode CLI, and locally hosted models can all join the same
+  Crew — the group of agents sharing one project's state. crew coordinates them; it doesn't
+  replace them.
 - **Both a human and a script can drive it.** Every command has a plain-text form for
   reading and a `--json` form that prints one complete JSON object per line for scripts to
   parse — there is no separate API to learn.
@@ -188,7 +190,10 @@ flowchart TB
         A2["Codex CLI"]
         A3["Gemini CLI"]
         A4["Copilot CLI"]
-        A5["Little Coder"]
+        A5["Antigravity CLI"]
+        A6["Pi CLI"]
+        A7["Little Coder"]
+        A8["opencode CLI"]
     end
 
     subgraph crewcli["crew CLI - one-shot process, exits after every command"]
@@ -208,7 +213,7 @@ flowchart TB
     TMUX["tmux session<br/>one pane per Agent + Relay window"]
     BROWSER["Browser dashboard<br/>127.0.0.1 - token-auth - SSE"]
 
-    A1 & A2 & A3 & A4 -- "crew join / send / receive / task ..." --> BIN
+    A1 & A2 & A3 & A4 & A5 & A6 & A7 & A8 -- "crew join / send / receive / task ..." --> BIN
     BIN --> RUN --> CMD
     CMD --> FMT
     CMD --> WS

docs/design/product-spec.md

@@ -67,7 +67,8 @@ requires one process that stays alive for as long as the tmux session does.
 - Built-in Manager, Worker, and Inspector Roles, plus per-project overrides.
 - Team definitions and manual command rendering.
 - An optional tmux launch (all panes running the same Participant CLI) and a session Relay.
-- Explicit setup generators for Claude Code, Codex CLI, Gemini CLI, and Copilot CLI.
+- Explicit setup generators for each of the eight Participant CLIs (Claude Code, Codex CLI,
+  Gemini CLI, Copilot CLI, Antigravity CLI, Pi CLI, Little Coder, and opencode CLI).
 - Checked setup recipes for Ollama and LM Studio as Model Backends (inference servers that a
   Participant CLI may use; crew itself never contacts them).
 - Human-readable output and JSON output where each line is one complete JSON object, stable

docs/design/srs.md

@@ -49,12 +49,14 @@ is expanded here into consecutive ids, tracked in
 ### 1.1 Purpose (§9.5.1)
 
 crew is a command-line tool that runs on your own machine and coordinates terminal coding
-agents (Claude Code, Codex CLI, Gemini CLI, Copilot CLI). This SRS specifies **all v1 software
-requirements**, plus the post-v1 additions for the Console and for stopping and resuming a Team —
-functional and non-functional — in enough detail to design and test against. crew coordinates
-Participant CLIs (the AI command-line apps that act as Agents) that are already running; they
-share one State Store, an SQLite database file inside the project's Workspace. crew **never
-calls a model provider** and runs no AI model itself.
+agents (the eight Participant CLIs named in
+[Appendix A](#appendix-a--assumptions-dependencies-and-tbd-items), as recorded in the platform
+registry). This SRS specifies **all v1 software requirements**, plus the post-v1 additions for
+the Console and for stopping and resuming a Team — functional and non-functional — in enough
+detail to design and test against. crew coordinates Participant CLIs (the AI command-line apps
+that act as Agents) that are already running; they share one State Store, an SQLite database
+file inside the project's Workspace. crew **never calls a model provider** and runs no AI model
+itself.
 
 ### 1.2 Scope (§9.5.2)
 
@@ -82,7 +84,8 @@ calls a model provider** and runs no AI model itself.
 crew is a standalone command-line program that other software (Participant CLIs) and a human
 operator invoke. Its boundaries with the outside world:
 
-- **System interface** — Participant CLIs (Claude Code, Codex CLI, Gemini CLI, Copilot CLI)
+- **System interface** — Participant CLIs (the eight named in
+  [Appendix A](#appendix-a--assumptions-dependencies-and-tbd-items))
   invoke `crew` subcommands; crew never invokes a model provider.
 - **User interface** — a text CLI with two equally supported output formats (a human
   table/line format, and `--json` output where each line is one complete JSON object —
@@ -508,8 +511,8 @@ rather than restating them.
   or that someone edited since crew wrote it, shall require `--force` and a backup. *Verify:
   automated test — `tests/integration/commands/setup.test.ts`, `tests/unit/setup-fs.test.ts`.*
 - **FR-G07 — Actual invocation.** Setup output shall state the real invocation (Claude,
-  Gemini, Pi, Little Coder, and opencode `/crew`; Codex `$crew`; Copilot `/agent` selection
-  plus prompt). *Verify: automated test — `tests/unit/platforms.test.ts`.*
+  Gemini, Antigravity, Pi, Little Coder, and opencode `/crew`; Codex `$crew`; Copilot
+  `/agent` selection plus prompt). *Verify: automated test — `tests/unit/platforms.test.ts`.*
 - **FR-G08 — Scoped permission guidance.** Setup guidance shall prefer permission limited to the
   `crew` command. *Verify: inspection — setup-integration.md; `src/platforms/shared.ts`.*
 - **FR-G09 — No automatic bypass flags.** Blunt permission-bypass flags shall never be enabled

package.json

@@ -1,7 +1,7 @@
 {
   "name": "@dichovsky/crew",
   "version": "0.1.1",
-  "description": "Local coordination for terminal coding agents (Claude Code, Codex, Gemini, Copilot, Little Coder) via a workspace-local SQLite inbox and reviewed task workflow — no daemon, no model provider.",
+  "description": "Local coordination for terminal coding agents (Claude Code, Codex, Gemini, Copilot, Antigravity, Pi, Little Coder, opencode) via a workspace-local SQLite inbox and reviewed task workflow — no daemon, no model provider.",
   "type": "module",
   "license": "MIT",
   "publishConfig": {
@@ -17,7 +17,10 @@
     "codex",
     "gemini-cli",
     "copilot-cli",
+    "antigravity",
+    "pi",
     "little-coder",
+    "opencode",
     "inbox",
     "agents"
   ],