@ory/opencode 0.1.2 → 0.2.0

package/README.md

@@ -2,15 +2,29 @@
 
 [Ory](https://ory.com) bundled into [OpenCode](https://opencode.ai): skills and markdown commands that scaffold Ory authentication into your codebase, a local Ory stack you can spin up in one command, and (when pointed at an Ory project) authentication, authorization, and audit for every tool OpenCode runs.
 
+You don't need an Ory account or any prior Ory experience to start.
+
+## Prerequisites
+
+- [OpenCode](https://opencode.ai) installed
+- Node.js **≥ 24**
+- [Docker](https://docs.docker.com/get-docker/) (only needed for the local Ory stack)
+- macOS or Linux. Windows works via WSL2.
+
 ## Install
 
-OpenCode loads plugins listed in the `plugin` array of `opencode.json` and fetches them from npm on next launch:
+Add the plugin to your `opencode.json`:
 
 ```json
 { "plugin": ["@ory/opencode"] }
 ```
 
-Or use the Ory installer to register the plugin **and** the Ory MCP server in one step, with no prior `npm install` required:
+OpenCode fetches the plugin from npm on next launch.
+
+<details>
+<summary>Alternative install paths</summary>
+
+The Ory installer registers the plugin **and** the Ory MCP server in one step, with no prior `npm install` required:
 
 ```bash
 npx @ory/opencode install                          # writes opencode.json + MCP config
@@ -20,80 +34,148 @@ npx @ory/opencode uninstall
 
 Either flow drops the Ory skill catalog into `.opencode/skills/` and the local-stack commands into `.opencode/commands/ory/`.
 
-## Developer experience
+</details>
 
-This plugin is a productivity layer for Ory itself. You don't need a real Ory project, an account, or any prior Ory experience to start using it.
+## Quickstart (≈ 3 minutes)
 
-### Skills for scaffolding Ory into your application
+From any project where you'd like Ory authentication, inside OpenCode:
+
+1. **Start a local Ory instance.** Ask OpenCode *"start the local Ory stack"* or run:
+
+   ```
+   /ory:local-up
+   ```
 
-Ask OpenCode to add Ory auth to your codebase. Each skill is a vetted, end-to-end playbook:
+   A banner prints the seeded test user's email and password. Note them — you'll log in with them in step 3.
 
-- **`ory-auth-setup`**: full project setup. Install the Ory CLI, create an Ory Network project, add Ory Elements, configure the SDK, build the auth pages, wire session middleware.
-- **`ory-login-flow`**: login, registration, recovery, verification, and settings pages with Ory Elements. Next.js App Router and React SPA variants.
-- **`ory-social-login`**: Google, GitHub, Apple, Microsoft, Discord, and other OIDC providers with Jsonnet data mappers.
-- **`ory-local-dev`**: drive the local Ory stack (below) from within OpenCode to prototype and test against without a remote project.
+2. **Scaffold Ory into your project.** Ask OpenCode *"add Ory auth to this app"* or invoke the `ory-auth-setup` skill.
 
-Skills are versioned with the plugin so guidance stays in sync as Ory APIs evolve.
+   OpenCode installs Ory Elements, wires the SDK, generates the login / registration / recovery / verification / settings pages, and sets up session middleware. It targets the local stack from step 1, so no signup or API key is needed.
+
+3. **Sign in.** Start your app, visit the login page OpenCode added, and sign in with the seeded credentials. You now have a real Ory session backed by a real Ory stack — locally, offline, with zero configuration.
+
+That's the full Ory DX path. Stop here if you're just evaluating the plugin. Continue to [Agent security](#agent-security) when you're ready to enforce.
+
+## What's included
+
+### Skills for scaffolding Ory into your application
+
+Each skill is a vetted, end-to-end playbook. Ask OpenCode in natural language or invoke a skill directly:
+
+- **`ory-auth-setup`** — full project setup. Install the Ory CLI, create an Ory Network project (or use the local one), add Ory Elements, configure the SDK, build the auth pages, wire session middleware.
+- **`ory-login-flow`** — login, registration, recovery, verification, and settings pages with Ory Elements. Next.js App Router and React SPA variants.
+- **`ory-social-login`** — Google, GitHub, Apple, Microsoft, Discord, and other OIDC providers with Jsonnet data mappers.
+- **`ory-local-dev`** — drive the local Ory stack from within OpenCode to prototype and test without a remote project.
 
 ### Ory MCP server
 
-Bundled and registered by the Ory installer. It exposes the Ory CLI and the Ory Network REST API as MCP tools so OpenCode can manage identities, OAuth2 clients, projects, permission tuples, and configuration without ever leaving the chat. Useful for seeding test data, verifying a scaffolded integration, or running one-off admin tasks.
+Bundled and registered by the Ory installer. Exposes the Ory CLI and the Ory Network REST API as MCP tools so OpenCode can manage identities, OAuth2 clients, projects, permission tuples, and configuration without ever leaving the chat. Useful for seeding test data, verifying a scaffolded integration, or running one-off admin tasks.
 
-### Local Ory stack in one command
+### Local Ory stack
 
 ```
 /ory:local-up      # start a local Ory instance in Docker
 /ory:local-down    # tear it all down
 ```
 
-`local up` runs a local Ory instance in Docker, covering everything the plugin and your scaffolded application need. It also brings up a login UI on `:3000` and Jaeger on `:16686`, all reachable through `http://localhost:4000`. A test user identity is seeded and the credentials are printed for you. Use it to:
+`local-up` brings up Ory Identities, OAuth2, and Permissions, plus a login UI on `:3000` and Jaeger on `:16686`, all reachable through `http://localhost:4000`. A test user identity is seeded and the credentials are printed for you. Use it to:
 
 - **Learn Ory hands-on** without signing up for a hosted project.
-- **Prototype** flows (login, social, MFA, recovery, permission tuples) against a real Ory backend in your local dev loop.
+- **Prototype** flows (login, social, MFA, recovery, permission tuples) against a real Ory backend.
 - **Test** an auth integration end-to-end before pushing anything to a real environment.
 - **Develop** your application against the same identity, OAuth2, and permission surfaces you'll ship with.
 
-Point `ORY_PROJECT_URL` at `http://localhost:4000` (or run `npx -y -p @ory/opencode ory-opencode configure`) and the security features below run against the local stack.
+## Pointing at a real Ory project
 
-## Configure
+The Quickstart uses the local stack. If you have a hosted [Ory Network](https://console.ory.sh) project, point the plugin at it:
 
 ```bash
-npx -y -p @ory/opencode ory-opencode configure --project-url https://<id>.projects.oryapis.com --api-key ory_pat_...
+npx -y -p @ory/opencode ory-opencode configure \
+  --project-url https://<id>.projects.oryapis.com \
+  --api-key ory_pat_...
 ```
 
-Config is saved to `~/.config/ory-agent-plugins/config.json` and shared across every Ory agent plugin on the machine. Without it the plugin still loads cleanly and runs in **pass-through mode**: skills and commands work, but nothing is blocked.
+Config is saved to `~/.config/ory-agent-plugins/config.json` and shared across every Ory agent plugin on the machine.
 
-## Agent security (Argus)
+Without configuration the plugin still loads cleanly and runs in **pass-through mode**: skills and commands work, but nothing is blocked. You can stay in pass-through mode indefinitely if you only want the DX features.
 
-Once the plugin is pointed at an Ory project (local or hosted), OpenCode's session and every tool call are governed by Ory.
+## Agent security
 
-- **Authentication.** Two identities. The human at the keyboard (the **user**) authenticates interactively via Ory Identities when `ORY_AUTH_GATE=1`; tokens refresh and a `user.auth` audit span is emitted at every session start. The OpenCode process (the **agent**) gets its own OAuth2 identity, self-registered via Dynamic Client Registration on first run.
-- **Authorization.** When OpenCode prompts for permission to use a tool, the plugin checks Ory Permissions against the user's subject and returns `allow` / `deny`. A parallel pre-tool check is recorded as part of the audit trail. MCP tool calls additionally get a server-level check.
-- **Audit.** Every decision (allow, deny, fallback) is recorded as a structured trace span: NDJSON file output and/or OTLP/HTTP export to Jaeger, Honeycomb, Grafana, and similar collectors. The user-to-agent delegation is written to Ory as a Zanzibar tuple so "agent X acting on behalf of user Y" stays queryable after tokens expire.
+Once the plugin is pointed at an Ory project (local or hosted), OpenCode's session and every tool call can be governed by Ory.
 
-The plugin is **fail-open** on its own infrastructure failures (network errors, rate limits, missing config), so enforcement is only as strong as your tuples; grant explicit `invoke` relations for the tools each user should be able to run.
+- **Authentication.** Two identities. The human at the keyboard (the **user**) authenticates interactively via Ory Identities when `ORY_AUTH_GATE=1` is set. The OpenCode process (the **agent**) gets its own OAuth2 identity, self-registered via [Dynamic Client Registration (RFC 7591)](https://datatracker.ietf.org/doc/html/rfc7591) on first run.
+- **Authorization.** When OpenCode prompts for permission to use a tool, the plugin checks [Ory Permissions](https://www.ory.com/docs/keto) (Zanzibar-style relation tuples) against the user's subject and returns `allow` / `deny`. A parallel pre-tool check is recorded as part of the audit trail. MCP tool calls additionally get a server-level check.
+- **Audit.** Every decision (allow, deny, fallback) is recorded as a structured trace span: NDJSON file output and/or OTLP/HTTP export to Jaeger, Honeycomb, Grafana, and similar collectors. The user → agent delegation is written to Ory as a relation tuple so *"agent X acting on behalf of user Y"* stays queryable after tokens expire.
 
-## Programmatic usage
+The plugin is **fail-open** on its own infrastructure failures (network errors, rate limits, missing config), so enforcement is only as strong as your tuples — grant explicit `invoke` relations for the tools each user should be able to run.
 
-```typescript
-import { createOryPlugin } from "@ory/opencode";
+### Enable enforcement
 
-const plugin = createOryPlugin();
-```
+After install the plugin runs in **observe mode**: every tool call is checked against Ory Permissions, but a deny is recorded as a `permission.observe_deny` audit span and the tool runs anyway. This lets you see what *would* be blocked before turning on hard blocking.
 
-## CLI
+1. **Turn on the user gate.** In your shell:
+
+   ```bash
+   export ORY_AUTH_GATE=1
+   ```
+
+   The next OpenCode session refreshes or prompts for PKCE login. Subsequent sessions reuse the persisted token until it expires.
+
+2. **Bootstrap tuples for the built-in tools.** One idempotent command grants the current user `use` on every tool OpenCode ships with (read, write, edit, bash, …):
+
+   ```bash
+   npx -y -p @ory/opencode ory-opencode permissions bootstrap
+   ```
+
+   If a user identity is already cached at install time, the installer runs this for you automatically — re-run after adding tools, switching subjects, or changing the namespace.
+
+3. **Check coverage.** `permissions status` probes every tool in the harness's catalog and prints allowed / denied per tool:
+
+   ```bash
+   npx -y -p @ory/opencode ory-opencode permissions status
+   ```
+
+   Add tuples for any MCP server tools or custom commands by hand, or via the Ory MCP server from inside OpenCode (*"grant me use on the bash tool"*).
+
+4. **Promote to enforce.** Once the observe-mode logs look right, switch over:
+
+   ```bash
+   npx -y -p @ory/opencode ory-opencode permissions enforce
+   ```
+
+   Denies now ask the harness to block via a thrown `OryDenialError`; OpenCode shows the denial reason and the decision is recorded as a `tool.block` trace span. Switch back any time with `permissions observe`.
+
+## CLI reference
 
 ```
 npx -y -p @ory/opencode ory-opencode install | uninstall              [--project-dir <path>]
 npx -y -p @ory/opencode ory-opencode configure                         [--project-url <url>] [--api-key <key>] [--audit-only]
 npx -y -p @ory/opencode ory-opencode agent <status|unregister>         Manage the agent's OAuth2 identity
+npx -y -p @ory/opencode ory-opencode permissions <status|bootstrap|observe|enforce>
 npx -y -p @ory/opencode ory-opencode local <up|down|status|seed|logs|env|configure|reset>
 npx -y -p @ory/opencode ory-opencode status                            [--project-dir <path>]
 ```
 
+Highlights:
+
+- `agent status` — show the current persisted DCR identity for the agent.
+- `permissions observe` / `permissions enforce` — switch between "log denies, allow through" (the install default) and "block denies." `permissions bootstrap` writes `use` tuples for the harness's built-in tools so the promotion path doesn't require hand-writing relationships.
+- `configure --audit-only` — kill switch that disables Ory entirely (no auth, no permission checks; only audit logging of tool invocations). For phased rollouts, prefer `permissions observe` over `--audit-only`.
+- `local seed` / `local env` — reseed the test user, or print env vars for pointing other tools at the local stack.
+
+## Troubleshooting
+
+- **`/ory:local-up` fails.** Make sure Docker is running and ports `3000`, `4000`, `4100`, and `16686` are free.
+- **PKCE login loops.** Clear persisted state with `npx -y -p @ory/opencode ory-opencode agent unregister` and retry.
+- **`npx` fetches an old version.** Force a fresh fetch: `npx -y -p @ory/opencode@latest ory-opencode …`.
+- **Need more signal.** Set `ORY_AGENT_DEBUG=true` and `ORY_AGENT_LOG_FILE=/tmp/ory.log` to capture structured logs.
+
 ## Links
 
-- [ory.com](https://ory.com)
+- [Ory documentation](https://www.ory.com/docs/)
+- [Ory Network console](https://console.ory.sh)
+- [Ory Elements](https://github.com/ory/elements)
+- [OpenCode documentation](https://opencode.ai)
 
 ## License
 

package/dist/cli/main.js

@@ -51,6 +51,10 @@ function main() {
     switch (command) {
         case "install":
             install(args);
+            postInstallPermissions("ory-opencode", "opencode").then(() => process.exit(0), (err) => {
+                console.error(err.message ?? err);
+                process.exit(1);
+            });
             break;
         case "uninstall":
             uninstall(args);
@@ -64,6 +68,12 @@ function main() {
                 process.exit(1);
             });
             break;
+        case "permissions":
+            (0, argus_1.runPermissionsCommand)("ory-opencode", "opencode", args).then((code) => process.exit(code), (err) => {
+                console.error(err.message ?? err);
+                process.exit(1);
+            });
+            break;
         case "status":
             status(args);
             break;
@@ -85,6 +95,12 @@ function main() {
             process.exit(1);
     }
 }
+async function postInstallPermissions(binName, harness) {
+    const bootstrapped = await (0, argus_1.maybeAutoBootstrap)(binName, harness);
+    (0, argus_1.printPermissionsOnboardingHelp)(binName, harness, {
+        bootstrappedAutomatically: bootstrapped,
+    });
+}
 function parseProjectDir(args) {
     const idx = args.indexOf("--project-dir");
     if (idx !== -1 && args[idx + 1])
@@ -185,6 +201,7 @@ Commands:
   install [--project-dir <path>]   Register the Ory plugin in opencode.json
   uninstall [--project-dir <path>] Remove the Ory plugin from opencode.json
   configure                        Set or view Ory project URL and API key
+  permissions <cmd>                Manage permission mode and tool tuples (status, bootstrap, observe, enforce)
   status [--project-dir <path>]    Show plugin configuration and status
   local <cmd>                      Manage local Ory dev environment (up, down, status, seed, ...)
 

package/dist/plugin.js

@@ -201,16 +201,30 @@ function createPermissionHandler(client, _options) {
                     subject,
                     spanAttributes: { toolName: input.toolID },
                 });
-                output.status = mcpResult.allowed ? "allow" : "deny";
+                const decision = (0, argus_1.applyPermissionMode)(client, mcpResult.allowed, {
+                    object: mcpTool.serverName,
+                    relation: "use",
+                    subjectId: (0, argus_1.subjectLabel)(subject),
+                    spanAttributes: { toolName: input.toolID, mcpServer: mcpTool.serverName, mcpTool: mcpTool.toolName },
+                });
+                output.status = decision.kind === "deny" ? "deny" : "allow";
+                return;
+            }
+            const decision = await (0, argus_1.checkAndDecide)(client, { namespace: resolveNamespace(), object: input.toolID, relation: "use", ...subject }, { spanAttributes: { toolName: input.toolID } });
+            if (decision.kind === "fail_open") {
+                const code = decision.error.code;
+                if (code === "network_error" || code === "rate_limited") {
+                    client.logger.warn("permission.fallback", {
+                        toolID: input.toolID,
+                        code,
+                    });
+                    // Leave output.status as "ask" (default)
+                    return;
+                }
+                output.status = "ask";
                 return;
             }
-            const result = await client.checkPermission({
-                namespace: resolveNamespace(),
-                object: input.toolID,
-                relation: "use",
-                ...subject,
-            }, { spanAttributes: { toolName: input.toolID } });
-            output.status = result.allowed ? "allow" : "deny";
+            output.status = decision.kind === "deny" ? "deny" : "allow";
         }
         catch (err) {
             if (isOryError(err) && (err.code === "network_error" || err.code === "rate_limited")) {
@@ -256,63 +270,90 @@ function createToolBeforeHandler(client) {
                     subject,
                     spanAttributes: { toolName: input.tool },
                 });
-                if (!mcpResult.allowed) {
+                const mcpAttrs = {
+                    toolName: input.tool,
+                    mcpServer: mcpTool.serverName,
+                    mcpTool: mcpTool.toolName,
+                    ...inputSummary,
+                };
+                const decision = (0, argus_1.applyPermissionMode)(client, mcpResult.allowed, {
+                    object: mcpTool.serverName,
+                    relation: "use",
+                    subjectId,
+                    spanAttributes: mcpAttrs,
+                });
+                if (decision.kind === "allow") {
+                    client.tracer.record("tool.invoke", "ok", { attributes: mcpAttrs });
+                    return;
+                }
+                if (decision.kind === "observe") {
                     client.tracer.record("tool.block", "denied", {
-                        attributes: { toolName: input.tool, mcpServer: mcpTool.serverName, mcpTool: mcpTool.toolName, ...inputSummary, ...(0, argus_1.alertAttributes)(false) },
-                    });
-                    const message = (0, argus_1.formatDenialMessage)({
-                        tool: input.tool,
-                        subjectId,
-                        mcp: mcpTool,
+                        attributes: { ...mcpAttrs, allowed: false, ...(0, argus_1.alertAttributes)(false) },
                     });
-                    client.logger.warn("tool.denied", {
-                        tool: input.tool,
-                        subjectId,
-                        message,
-                    });
-                    // OpenCode has no native block signal in tool.execute.before;
-                    // throwing aborts execution (documented .env-protection pattern)
-                    // and surfaces error.message back to the agent.
-                    throw new argus_1.OryDenialError({
-                        tool: input.tool,
-                        subjectId,
-                        mcp: mcpTool,
+                    client.tracer.record("tool.invoke", "ok", {
+                        attributes: { ...mcpAttrs, allowed: false, observed: true },
                     });
+                    return;
                 }
-                client.tracer.record("tool.invoke", "ok", {
-                    attributes: { toolName: input.tool, mcpServer: mcpTool.serverName, mcpTool: mcpTool.toolName, ...inputSummary },
-                });
-                return;
-            }
-            const namespace = resolveNamespace();
-            const result = await client.checkPermission({
-                namespace,
-                object: input.tool,
-                relation: "use",
-                ...subject,
-            }, { spanAttributes: { toolName: input.tool } });
-            if (!result.allowed) {
                 client.tracer.record("tool.block", "denied", {
-                    attributes: { toolName: input.tool, ...inputSummary, allowed: result.allowed, ...(0, argus_1.alertAttributes)(false) },
+                    attributes: { ...mcpAttrs, allowed: false, ...(0, argus_1.alertAttributes)(false) },
                 });
                 const message = (0, argus_1.formatDenialMessage)({
                     tool: input.tool,
                     subjectId,
-                    namespace,
+                    mcp: mcpTool,
                 });
                 client.logger.warn("tool.denied", {
                     tool: input.tool,
                     subjectId,
                     message,
                 });
+                // OpenCode has no native block signal in tool.execute.before;
+                // throwing aborts execution (documented .env-protection pattern)
+                // and surfaces error.message back to the agent.
                 throw new argus_1.OryDenialError({
                     tool: input.tool,
                     subjectId,
-                    namespace,
+                    mcp: mcpTool,
                 });
             }
-            client.tracer.record("tool.invoke", "ok", {
-                attributes: { toolName: input.tool, ...inputSummary, allowed: result.allowed },
+            const namespace = resolveNamespace();
+            const decision = await (0, argus_1.checkAndDecide)(client, { namespace, object: input.tool, relation: "use", ...subject }, { spanAttributes: { toolName: input.tool } });
+            if (decision.kind === "fail_open") {
+                handlePermissionError(decision.error, input.tool, client);
+                return;
+            }
+            const attrs = { toolName: input.tool, ...inputSummary };
+            if (decision.kind === "allow") {
+                client.tracer.record("tool.invoke", "ok", { attributes: { ...attrs, allowed: true } });
+                return;
+            }
+            if (decision.kind === "observe") {
+                client.tracer.record("tool.block", "denied", {
+                    attributes: { ...attrs, allowed: false, ...(0, argus_1.alertAttributes)(false) },
+                });
+                client.tracer.record("tool.invoke", "ok", {
+                    attributes: { ...attrs, allowed: false, observed: true },
+                });
+                return;
+            }
+            client.tracer.record("tool.block", "denied", {
+                attributes: { ...attrs, allowed: false, ...(0, argus_1.alertAttributes)(false) },
+            });
+            const message = (0, argus_1.formatDenialMessage)({
+                tool: input.tool,
+                subjectId,
+                namespace,
+            });
+            client.logger.warn("tool.denied", {
+                tool: input.tool,
+                subjectId,
+                message,
+            });
+            throw new argus_1.OryDenialError({
+                tool: input.tool,
+                subjectId,
+                namespace,
             });
         }
         catch (err) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ory/opencode",
-  "version": "0.1.2",
+  "version": "0.2.0",
   "description": "Ory plugin for OpenCode: scaffolding skills, a local Ory instance, and authentication, authorization, and audit for every tool call",
   "license": "Apache-2.0",
   "homepage": "https://ory.com",
@@ -64,7 +64,7 @@
     "!dist/**/*.tsbuildinfo"
   ],
   "dependencies": {
-    "@ory/argus": "0.1.2"
+    "@ory/argus": "0.2.0"
   },
   "engines": {
     "node": ">=24"