diff --git a/README.md b/README.md index 23c1a8c..9a1736d 100644 --- a/README.md +++ b/README.md @@ -47,6 +47,7 @@ For detailed documentation, see the [Further Reading](#further-reading) section │ Running Parallel Auto-delegates work; /fleet for explicit │ │ Tasks & /fleet parallel execution via subagents │ │ Autopilot Mode Hands-off autonomous task completion (CLI) │ +│ Critic Agent Dual-model plan & code review (experimental) │ │ Coding Agent Autonomous cloud agent — issues in, PRs out │ │ Agentic Workflows Automate repo tasks via GitHub Actions + AI │ │ Mission Control Dashboard to manage coding agents at scale │ @@ -94,6 +95,7 @@ For detailed documentation, see the [Further Reading](#further-reading) section - [Agentic Features](#agentic-features) - [Running Parallel Tasks](#running-parallel-tasks) - [Autopilot Mode](#autopilot-mode) + - [Critic Agent](#critic-agent) - [Copilot Coding Agent](#copilot-coding-agent) - [Agentic Workflows](#agentic-workflows) - [Mission Control](#mission-control) @@ -253,9 +255,12 @@ When creating or modifying database tables. **Effect:** When Copilot detects a task related to migrations, it loads this skill automatically. Works across CLI, VS Code, and the coding agent. +> 💡 **Built-in skills:** The CLI ships with built-in skills out of the box — no files required. The first is `customizing-copilot-cloud-agents-environment`, a guide for setting up the [Copilot Coding Agent's environment](https://docs.github.com/en/copilot/how-tos/use-copilot-agents/coding-agent/customizing-the-development-environment-for-copilot-coding-agent). More built-in skills will be added over time. Your own `.github/skills/` files work alongside built-in ones. + | | | |---|---| | **Scope** | Auto-loaded when the task domain matches | +| **Built-in skills** | Bundled with the CLI — no files required (added in v1.0.17) | | **Difference from prompts** | Skills are auto-detected; prompts are manually invoked | --- @@ -300,12 +305,24 @@ When creating or modifying database tables. | "The issue says..." (you paste it) | Agent reads the issue via GitHub MCP server | | "Upload to cloud storage" (you do it manually) | Agent calls cloud storage API through an MCP server | +#### MCP authentication & management (CLI) + +| Command | Description | +|---|---| +| `/mcp auth` | Authenticate or re-authenticate an MCP OAuth server; supports account switching | +| `mcp.config.list` | List all persistent MCP server configurations (server RPC) | +| `mcp.config.add` | Add a persistent MCP server configuration (server RPC) | +| `mcp.config.update` | Update an existing MCP server configuration (server RPC) | +| `mcp.config.remove` | Remove a persistent MCP server configuration (server RPC) | + +OAuth flows support HTTPS redirect URIs (via self-signed certificate fallback) and a device code flow (RFC 8628) as a fallback for headless and CI environments. + | | | |---|---| | **Scope** | Available to all agents in the session | | **Discovery** | VS Code has a built-in MCP server gallery (search `@mcp` in Extensions) | | **Security** | Servers run locally — your credentials stay on your machine | -| **OAuth / API keys** | MCP servers can request you to visit a URL for out-of-band auth flows (e.g. OAuth, API key entry) | +| **OAuth / API keys** | Use `/mcp auth` for OAuth flows; device code flow (RFC 8628) supported in headless/CI | --- @@ -332,6 +349,12 @@ Custom scripts that run automatically at specific lifecycle events — like pre- }, "startup": { "prompt": "/compact Summarize recent changes in RecipeShare" + }, + "postToolUseFailure": { + "command": "echo 'Tool ${tool} failed — check logs' >> .copilot/errors.log" + }, + "notification": { + "command": "osascript -e 'display notification \"Copilot finished\" with title \"RecipeShare\"'" } } } @@ -346,6 +369,11 @@ Custom scripts that run automatically at specific lifecycle events — like pre- | `post-edit` | After Copilot edits a file | | `pre-commit` | Before a git commit | | `startup` | When a CLI session starts — auto-submits a prompt or slash command | +| `preToolUse` | Before each tool call; return `permissionDecision: 'allow'` to auto-approve without prompting | +| `postToolUse` | After a **successful** tool call (does **not** fire on failures) | +| `postToolUseFailure` | After a **failed** tool call (new in v1.0.15) | +| `notification` | Asynchronously on shell completion, permission prompts, elicitation dialogs, and agent completion (new in v1.0.18) | +| `PermissionRequest` | When a tool requests permission — script can approve or deny programmatically (new in v1.0.16) | **Config notes:** @@ -599,10 +627,48 @@ Autopilot and `/fleet` are independent features that combine well. A common work | **Where** | Copilot CLI only (GA as of v1.0, March 2026) | | **Toggle** | Shift+Tab during a session, or `--autopilot` flag | | **Cost** | Each autonomous continuation step uses premium requests | +| **Session export** | Use `/share html` to export the session as a self-contained interactive HTML file (added in v1.0.15) | | **Docs** | [Autopilot mode](https://docs.github.com/en/copilot/concepts/agents/copilot-cli/autopilot) | --- +### Critic Agent + +An automatic **dual-model review** step that uses a complementary model to check plans and complex implementations for errors before they're finalized. + +> **When you need it:** You want an independent second opinion on a complex implementation — catching issues a single model might overlook because it produced the code in the first place. + +- Automatically activates during **plan mode** and on complex multi-step implementations +- Uses a **different model** from the one that authored the code, so it brings a fresh perspective +- Surfaces contradictions, missed edge cases, and logical errors +- Currently **experimental** — available for Claude models (enable via the `experimental` CLI setting) + +
+How to enable the Critic Agent (experimental) + +In `~/.copilot/settings.json`: + +```jsonc +{ + "experimental": { + "criticAgent": true + } +} +``` + +Once enabled, the CLI automatically invokes the Critic after generating a plan or complex implementation. You'll see a "Critic review" step appear in the timeline before Copilot proceeds. + +
+ +| | | +|---|---| +| **Where** | Copilot CLI only | +| **Status** | Experimental — Claude models only (added in v1.0.18) | +| **Cost** | Uses an additional LLM call with a complementary model | +| **Difference from code review** | Critic runs _before_ code is finalized, not after a PR is opened | + +--- + ### Copilot Coding Agent GitHub's autonomous agent that picks up issues, creates branches, and opens PRs — **without a chat session**. It runs on GitHub's cloud infrastructure, not your machine. @@ -868,7 +934,7 @@ squad > /status Key CLI commands: `squad init`, `squad status`, `squad triage` (auto-triage issues), `squad copilot` (add/remove @copilot), `squad doctor`, `squad nap` (context hygiene), `squad export`/`import`, `squad aspire` (observability dashboard). -#### What's new (v0.8.x) +#### What's new (v0.9.x) - **SubSquads** — break large teams into focused sub-groups (renamed from workstreams) - **Crash recovery** — sessions persist to disk; agents resume from checkpoint after failures @@ -971,6 +1037,7 @@ Based on [lessons from 2,500+ repositories](https://github.blog/ai-and-ml/github | **Plugins** | Bundled agent toolkits | `plugin.json` manifest | When sharing agent setups across repos | | **Running Parallel Tasks / `/fleet`** | Auto-delegates or explicitly fans out work to subagents | *(runtime capability)* / `/fleet` slash command | When your request has multiple independent jobs | | **Autopilot Mode** | Hands-off autonomous task completion | Copilot CLI (`--autopilot` / Shift+Tab) | When you want Copilot to work to completion without interaction | +| **Critic Agent** | Dual-model review of plans & implementations | Copilot CLI (experimental, `~/.copilot/settings.json`) | When you want an automatic second opinion before code is finalized | | **Coding Agent** | Autonomous cloud agent | GitHub infrastructure | When you want async issue automation | | **Agentic Workflows** | AI + GitHub Actions automation | GitHub Actions | When you want automated repo maintenance | | **Mission Control** | Multi-agent dashboard | GitHub.com / VS Code | When managing agents at scale |