More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Opencode Agent Team

    1

    Turn any OpenCode project into a software company in one command — a project-grounded agent team (PM, architect, devs, testers, reviewer, triager, security, devops, writer) with pipeline slash commands. V1+V2 compatible.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    opencode-agent-team

    opencode-agent-team

    Example session with this skill installed

    **Use the opencode-agent-team skill:** init --roles core --lang English

    • Read your context and instructions
    • Compiled the opencode-agent-team

    Explores the repo, confirms the team with 4 questions, then writes: .opencode/agents/{product-manager,architect,backend-dev,test-engineer,code-reviewer,bug-triager}.md (mode: subagent, correct permissions), .opencode/commands/{feature,fix,test-all,prd,adr,review,standup,team}.md, .opencode/agent-team.json, per-role memory dirs, and docs/{PROJECT-CONTEXT.md,TEAM.md,LEARNINGS.md,product/BACKLOG.md}. Fills PROJECT-CONTEXT.md with verified facts (stack, test commands, architecture, red flags), runs --check (frontmatter clean), and reports the roles created, commands available, and the first two commands to try.

    Fills PROJECT-CONTEXT.md

    Connects securely to your tools. The creator never sees your data.

    About this skill

    **Turn any OpenCode project into a working software company — in one command.** This skill bootstraps a **project-grounded agent team** inside your repo: subagents (product manager, architect, backend/mobile/web devs, test engineers, code reviewer, bug triager, security, devops, tech writer), pipeline slash commands, a shared context file, and per-role memory. The team is grounded in **your project's real code** — its stack, its tests, its architecture, its conventions — not generic prompts. ### What you get - **17 roles** — pick `core` (PM, architect, one dev per stack, test-engineer, code-reviewer, bug-triager) or the `full` team. - **10 pipeline commands** — `/feature`, `/fix`, `/test-all`, `/prd`, `/adr`, `/security-audit`, `/release`, `/standup`, `/review`, `/team`. - **A verified context file** (`PROJECT-CONTEXT.md`) every agent reads first — the single source of the team's quality, filled from your actual code (stack, test commands, architecture, conventions, hygiene red flags) before any agent runs. - **Per-role memory** so agents get smarter over time, committed with your repo (no secrets). ### Why it's different The team's quality comes from one rich, verified context file — not from long agent prompts. The skill explores your real code and writes verified facts (not guesses). Every generated agent carries correct OpenCode permission rules: the code-reviewer can't edit, testers can't reach the network, docs-only roles can only touch docs and their own memory. ### Compatibility - **V1 + V2** — detects your OpenCode version and emits the right permission format (`permissions:` list for V2, legacy `permission:` map for V1 via `compat_v1`). - **Any provider** — subagents inherit your session model by default, so it works with whatever provider you run. Pin models per tier or per role if you want. ### How it works 1. Explores your repo (stack, test commands, architecture, conventions, red flags). 2. Asks four quick questions (profile, role set, report language, docs location). 3. Writes a config and runs a self-contained Python renderer that generates every agent, command and doc. 4. Fills `PROJECT-CONTEXT.md` with verified facts about *your* project. After that, mention an agent by name (`@code-reviewer review this diff`) or run a command (`/feature add JWT login`). Also supports `add <role>`, `status`, and `upgrade <role|all>` operations. MIT-licensed.

    Built specifically for OpenCode (not Claude Code, Cursor, or generic agents). OpenCode subagents cannot spawn subagents, so multi-role coordination runs through the slash commands in your main session rather than nested agents. Requires Python 3 on PATH to run the renderer. The skill fills PROJECT-CONTEXT.md from your repo during the exploration step; the depth of that context (and therefore the team's quality) depends on that step being run against a real codebase.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 16 days ago

    Requires

    OpenCode V1 or V2
    Python 3.8+
    Python
    Git

    Permissions

    File Scopes

    .opencode/
    (agents
    commands
    agent-team.json
    config
    per-role
    memory)
    your
    docs
    dir
    /
    (PROJECT-CONTEXT.md
    TEAM.md
    LEARNINGS.md
    product/BACKLOG.md)

    Runs python3 to render templates into your project. Writes only under .opencode/ and your chosen docs directory. Never reads or writes .env files or secrets, never runs git write commands or deploys, and makes no network calls.

    Listed16 days ago

    What's inside

    Frequently Asked Questions

    No — it generates OpenCode-native agents and commands (mode: subagent, OpenCode permission rules, .opencode/ layout). It is built for OpenCode V1 and V2.

    Any. By default every generated subagent inherits your session model, so it works with whatever provider you run. You can optionally pin models per effort tier (heavy/other/light) or per role.

    No. The renderer never overwrites an existing file without --force, and it keeps a .bak when it does. It reuses existing AGENTS.md / .opencode / docs rather than duplicating them.

    No. Generated agents never edit .env files, never print secret values, never run git write commands, and never call or deploy to production. The renderer makes no network calls.

    A single self-contained Python script (standard library only) with all templates embedded. It writes agents, commands and docs into your project and nothing else.

    New in Workflow Automation