wolfstar-project · kodiakhq · Aug 9, 2026 · Aug 9, 2026 · Aug 9, 2026 · Aug 9, 2026
diff --git a/.agent-zero.example.yml b/.agent-zero.example.yml
@@ -57,5 +57,11 @@ runner:
  maxOutputBytes: 200000
 
 model:
+ # ai-gateway, anthropic, google, openai, or openai-compatible.
+ # Credentials are read only from the provider's documented environment variable; they are
+ # never accepted from repository configuration or persisted with task evidence.
  provider: openai-compatible
  name: gpt-5
+ # Optional explicit pricing enables deterministic cost accounting.
+ # inputCostPerMillionTokens: 1.25
+ # outputCostPerMillionTokens: 10
diff --git a/.agents/skills/orpc-server b/.agents/skills/orpc-server
diff --git a/.github/ISSUE_TEMPLATE/bug_report.yml b/.github/ISSUE_TEMPLATE/bug_report.yml
@@ -31,7 +31,7 @@ body:
  - Agent runtime
  - Runner
  - CLI
- - oRPC server
+ - Dashboard
  - GitHub adapter
  - Configuration
  - Build or CI

diff --git a/.github/ISSUE_TEMPLATE/feature_request.yml b/.github/ISSUE_TEMPLATE/feature_request.yml
@@ -32,7 +32,7 @@ body:
  - Agent runtime
  - Runner
  - CLI
- - oRPC server
+ - Dashboard
  - GitHub adapter
  - Model provider
  - Developer experience

diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md
@@ -7,6 +7,6 @@ Read `/AGENTS.md` before editing and load the relevant skill from `/.agents/skil
 - Keep runtime command execution and target-repository mutation inside `packages/runner`.
 - Keep `observe` mode read-only.
 - Add deterministic tests for behavior, state transitions, and safety-sensitive changes.
-- Use oRPC in `apps/server`, `@bomb.sh/args` plus `@clack/prompts` in the CLI, tsdown for builds, and Oxlint/Oxfmt for code quality.
+- Keep `apps/dashboard` frontend-only, use `@bomb.sh/args` plus `@clack/prompts` in the CLI, tsdown for package builds, and Oxlint/Oxfmt for code quality.
 - Do not introduce Hono, ESLint, Prettier, npm, Yarn, or Bun without an accepted architectural proposal.
 - Run `aube run check:repo`, `aube run lint:ci`, `aube run typecheck`, `aube test`, and `aube run build` before handing off a complete change.
diff --git a/.github/workflows/ci.yaml b/.github/workflows/ci.yaml
@@ -87,4 +87,5 @@ jobs:
  run: aube run build
 
  - name: 📦 Verify package version
- run: node --input-type=module --eval "import('./packages/shared/dist/index.mjs').then(({ version }) => { if (version !== '0.1.0') process.exit(1) })"
+ run: |
+ node --input-type=module --eval "Promise.all([import('./packages/shared/dist/index.mjs'), import('./packages/shared/package.json', { with: { type: 'json' } })]).then(([built, pkg]) => { if (built.version !== pkg.default.version) { console.error('Injected version', built.version, 'does not match package.json version', pkg.default.version); process.exit(1) } })"
diff --git a/.github/workflows/release.yaml b/.github/workflows/release.yaml
@@ -32,7 +32,8 @@ jobs:
  run: aube run build
 
  - name: Verify injected version
- run: node --input-type=module --eval "import('./packages/shared/dist/index.mjs').then(({ version }) => { if (version !== '0.1.0') process.exit(1) })"
+ run: |
+ node --input-type=module --eval "Promise.all([import('./packages/shared/dist/index.mjs'), import('./packages/shared/package.json', { with: { type: 'json' } })]).then(([built, pkg]) => { if (built.version !== pkg.default.version) { console.error('Injected version', built.version, 'does not match package.json version', pkg.default.version); process.exit(1) } })"
 
  # Publishing remains intentionally absent until the @agent-zero package
  # scope and npm trusted publishing policy are configured.
diff --git a/.gitignore b/.gitignore
@@ -2,14 +2,16 @@ node_modules/
 .turbo/
 dist/
 .output/
+.nuxt/
 *.tsbuildinfo
 coverage/
+playwright-report/
+test-results/
+test-report.junit.xml
 .env
 .env.*
 !.env.example
 .agent-zero/
+.data/
 *.log
 .DS_Store
-
-# Nitro build output
-.output/
diff --git a/.skills/agent-zero-architecture/SKILL.md b/.skills/agent-zero-architecture/SKILL.md
@@ -16,7 +16,7 @@ Keep dependency direction explicit while changing the monorepo.
 - `runner`: command execution and checkout mutation boundary, plus the policy-to-boundary factory.
 - `agent`: orchestration, the lifecycle machine, and the validation policy.
 - `cli`: argument parsing and terminal presentation.
-- `apps/server`: composition root for webhook ingestion, task execution, and evidence publication.
+- `apps/dashboard`: frontend-only Nuxt operational dashboard with no runtime-package dependencies.
 
 ## Workflow
 
@@ -30,7 +30,7 @@ Keep dependency direction explicit while changing the monorepo.
 
 ## Reject these designs
 
-- Shell execution in the server, CLI presentation, GitHub adapter, model adapter, or agent state machine.
+- Shell execution in a transport adapter, CLI presentation, GitHub adapter, model adapter, or agent state machine.
 - HTTP request/response types inside the runtime.
 - GitHub SDK objects passed through shared contracts.
 - A generic `utils` package used to bypass ownership decisions.

diff --git a/.skills/agent-zero-safety/SKILL.md b/.skills/agent-zero-safety/SKILL.md
@@ -22,6 +22,9 @@ Safety properties are behavior, not documentation. Back every change with determ
 - A failed verification cannot be represented as success. `verified` is derived once, where the terminal result is built.
 - A run that cannot verify does not write. No checks means no change.
 - Isolation is never approximated. Requesting a sandbox that cannot be provided must fail.
+- Remote sandbox credentials stay private to the provider adapter. Requests, leases, snapshots, task state, and logs remain credential-free.
+- Runner pools enforce active, per-repository, and lease-duration ceilings before provisioning and stop expired leases.
+- Persistent task records omit review input and checkout paths, and recursively redact every string before storage.
 - A reviewer's claim is not evidence. Reject what the repository does not support, and keep the reasons.
 
 ## Review workflow

diff --git a/.skills/orpc-server/SKILL.md b/.skills/orpc-server/SKILL.md
diff --git a/.skills/tsdown/SKILL.md b/.skills/tsdown/SKILL.md
@@ -10,7 +10,7 @@ Use this skill when changing build entries, output formats, declarations, packag
 ## Repository rules
 
 - The base configuration is `scripts/tsdown.config.ts`.
-- Every package or app keeps a small `tsdown.config.ts` that imports the shared factory.
+- Every workspace built with tsdown keeps a small `tsdown.config.ts` that imports the shared factory. The Nuxt dashboard uses the Nuxt build pipeline and does not keep a tsdown config.
 - Publishable packages output ESM and CommonJS with declarations and source maps.
 - Apps and the CLI output ESM executables without publishable-library declarations unless explicitly needed.
 - Publishable package builds must keep `publint` and `attw` validation enabled.

diff --git a/AGENTS.md b/AGENTS.md
@@ -15,7 +15,7 @@ These instructions apply to humans and coding agents working in this repository.
 - Use aube for dependencies and scripts. It reads and writes `pnpm-lock.yaml` and `pnpm-workspace.yaml` in place; keep both files and do not create npm, Yarn, or Bun lockfiles.
 - `typescript` is overridden to `typescript-native-bridge` in `pnpm-workspace.yaml`, so `tsc` and every Compiler API consumer type-check on tsgo. Keep the pin exact and reinstall after changing it.
 - Use Turborepo through the root scripts; do not duplicate orchestration in package scripts.
-- Use tsdown through each package's `tsdown.config.ts` and the shared `scripts/tsdown.config.ts`.
+- Use tsdown through each tsdown-built package's `tsdown.config.ts` and the shared `scripts/tsdown.config.ts`. The Nuxt dashboard uses the Nuxt build pipeline.
 - Use Oxlint with type-aware checks and Oxfmt. Do not add ESLint or Prettier.
 - Do not edit `dist/`, `.turbo/`, or generated declaration files.
 
@@ -28,7 +28,7 @@ These instructions apply to humans and coding agents working in this repository.
 - `packages/config`: configuration parsing and policy.
 - `packages/shared`: stable cross-package contracts.
 - `packages/cli`: argument parsing and terminal presentation.
-- `apps/server`: oRPC transport and control-plane composition.
+- `apps/dashboard`: frontend-only Nuxt operational dashboard.
 
 The runtime must remain independent from HTTP, GitHub, terminal UI, and specific model providers. Adapters depend on the runtime; the runtime must not depend on adapters.
 

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -41,7 +41,7 @@ aube test
 | Configuration and policy | `packages/config` |
 | Shared contracts | `packages/shared` |
 | CLI parsing and presentation | `packages/cli` |
-| HTTP control-plane composition | `apps/server`  |
+| Dashboard frontend | `apps/dashboard` |
 
 Read [AGENTS.md](AGENTS.md) and the matching files in `.agents/skills/` before making architectural or safety-sensitive changes.
 

diff --git a/README.md b/README.md
@@ -33,10 +33,7 @@ Feedback is never treated as truth merely because it came from a human or an AI
 ## Architecture
 
 ```text
-GitHub webhook / CLI
- │
- ▼
- Server API ─── task events
+GitHub adapter / CLI
  │
  ▼
  Agent state machine
@@ -46,6 +43,8 @@ GitHub webhook / CLI
  │
  ▼
  Runner boundary ─── repository commands and file operations
+
+Nuxt dashboard ─── frontend-only operational interface
 ```
 
 | Package | Responsibility |
@@ -57,7 +56,7 @@ GitHub webhook / CLI
 | [`packages/config`](./packages/config) | Configuration parsing and policy |
 | [`packages/shared`](./packages/shared) | Stable cross-package contracts |
 | [`packages/cli`](./packages/cli) | Argument parsing and terminal presentation |
-| [`apps/server`](./apps/server)  | oRPC transport and control-plane composition |
+| [`apps/dashboard`](./apps/dashboard) | Frontend-only Nuxt operational dashboard  |
 
 Adapters depend on the runtime; the runtime never depends on adapters. See [docs/architecture.md](./docs/architecture.md) for the full dependency rules.
 
@@ -97,22 +96,40 @@ The CLI parses arguments with [`@bomb.sh/args`](https://github.com/bomb-sh/args)
 
 ---
 
-## Control plane
+## Dashboard
+
+`aube run dev` starts the frontend-only Nuxt dashboard on `http://localhost:3000`. It is an operational interface shell: it does not expose API or RPC routes, persist task data, import runtime packages, or execute repository work.
+
+`aube run test:e2e` builds the dashboard, starts the production preview on port 5678, and runs the Playwright smoke suite. Use `aube --filter @agent-zero/dashboard run test:e2e:ui` for Playwright UI mode.
+
+Hosted execution is available through the provider-neutral `RunnerPool`: every lease has a maximum lifetime, quota checks run before provisioning, expired sandboxes are stopped, and the agent receives only the ordinary `Runner` contract. See [the sandbox provider evaluation](./docs/sandbox-providers.md).
+
+Agent Zero supports native OpenAI, Anthropic, and Google Generative AI adapters, Vercel AI
+Gateway, and arbitrary OpenAI-compatible endpoints. Select the transport in repository policy and
+provide its credential through the environment:
 
-The type-safe oRPC API starts on `http://localhost:4040` and exposes `health`, `tasks.list`, `tasks.get`, and `tasks.create`. Call it from a dashboard or another service with `@orpc/client`:
+| `model.provider` | Credential environment variable | Model example |
+| ------------------- | -------------------------------------------------------- | ----------------------------- |
+| `ai-gateway` | `AI_GATEWAY_API_KEY` or Vercel OIDC | `anthropic/claude-sonnet-4.5` |
+| `anthropic` | `ANTHROPIC_API_KEY` | `claude-sonnet-4-5` |
+| `google` | `GOOGLE_GENERATIVE_AI_API_KEY` | `gemini-2.5-pro` |
+| `openai` | `OPENAI_API_KEY` | `gpt-5` |
+| `openai-compatible` | `OPENAI_COMPATIBLE_API_KEY` (or legacy `OPENAI_API_KEY`) | provider-specific |
 
-```ts
-import { createORPCClient } from '@orpc/client';
-import { RPCLink } from '@orpc/client/fetch';
-import type { RouterClient } from '@orpc/server';
-import type { AppRouter } from '@agent-zero/server/router';
+`AGENT_ZERO_MODEL_BASE_URL` is an optional operator environment variable for custom gateways and
+self-hosted endpoints. Endpoint URLs and credentials cannot be named or embedded in
+`.agent-zero.yml`, so untrusted repository policy cannot redirect a provider secret. The AI Gateway
+accepts `provider/model` identifiers and exposes the broader AI SDK provider catalog without adding
+provider-specific logic to the Agent Zero runtime.
 
-const zero: RouterClient<AppRouter> = createORPCClient(
- new RPCLink({ url: 'http://localhost:4040' }),
-);
+To record cost, configure explicit rates; Agent Zero never guesses provider pricing:
 
-await zero.tasks.create({ repository: '.', feedback: 'Check error handling', mode: 'observe' });
-await zero.tasks.create({ repository: '.', trigger: 'proactive', mode: 'observe' });
+```yaml
+model:
+ provider: openai-compatible
+ name: gpt-5
+ inputCostPerMillionTokens: 1.25
+ outputCostPerMillionTokens: 10
 ```
 
 `observe` is the safe default and never writes files. Proactive pull-request webhooks are ignored until `proactive.enabled` is true. Automatic changes additionally require `mode: fix` or `autonomous`, `autofix.enabled`, sufficient confidence, an allowed change-risk class, repository-native checks, and (by default for proactive/autonomous work) an isolated runner. High-impact changes always require human approval.
@@ -124,7 +141,7 @@ await zero.tasks.create({ repository: '.', trigger: 'proactive', mode: 'observe'
 - **[aube](https://aube.jdx.dev)** &ndash; package manager, pinned through `packageManager`, reusing the pnpm lockfile and workspace files.
 - **[typescript-native-bridge](https://github.com/johnsoncodehk/typescript-native-bridge)** &ndash; overrides `typescript` repo-wide, so `tsc` keeps the classic package surface while the checker runs on tsgo in-process. The override lives in `pnpm-workspace.yaml` and is pinned exactly; the fork only publishes prerelease versions.
 - **[Turborepo](https://turborepo.dev)** &ndash; schedules workspace tasks in dependency order and caches tsdown build outputs.
-- **[tsdown](https://tsdown.dev)** &ndash; builds publishable packages as ESM and CommonJS with matching declarations and source maps, through the shared [tsdown configuration](./scripts/tsdown.config.ts). Apps stay ESM-only.
+- **[tsdown](https://tsdown.dev)** &ndash; builds publishable packages as ESM and CommonJS with matching declarations and source maps, through the shared [tsdown configuration](./scripts/tsdown.config.ts). The Nuxt dashboard uses the Nuxt build pipeline instead.
 - **[Oxlint](https://oxc.rs) + [Oxfmt](https://oxc.rs)** &ndash; type-aware linting and repository-wide formatting, extended with [`@e18e/eslint-plugin`](https://github.com/e18e/eslint-plugin) for modernization, module-replacement, and performance rules.
 - **[Knip](https://knip.dev)** &ndash; detects unused files, exports, and dependencies across the workspace as part of `lint:ci`.
 - **[`@arethetypeswrong/cli`](https://github.com/arethetypeswrong/arethetypeswrong.github.io) + [Publint](https://publint.dev)** &ndash; validate every package build.
@@ -136,15 +153,15 @@ GitHub Actions run typecheck, build/export validation, Oxlint, Oxfmt, tests, and
 
 ## Security model
 
-The included `LocalRunner` is intended for trusted local development. Production deployments must place it inside Docker, a microVM, or another ephemeral sandbox with CPU, memory, filesystem, and network policies. The server never invokes shell commands directly.
+The included `LocalRunner` is intended for trusted local development. Production deployments must place it inside Docker, a microVM, or another ephemeral sandbox with CPU, memory, filesystem, and network policies. Only `packages/runner` may invoke shell commands.
 
 Report vulnerabilities privately as described in [SECURITY.md](./SECURITY.md). Do not open a public issue.
 
 ---
 
 ## Agent Skills
 
-Task-specific Agent Skills live in `.skills/` and are exposed to coding agents through `.agents/skills/`. Skilld manages the versioned tsdown skill, and the same portable layout covers architecture, CLI, oRPC, Turborepo, and safety work:
+Task-specific Agent Skills live in `.skills/` and are exposed to coding agents through `.agents/skills/`. Skilld manages the versioned tsdown skill, and the same portable layout covers architecture, CLI, Turborepo, and safety work:
 
 ```bash
 aube run skills:list

diff --git a/apps/dashboard/app/app.vue b/apps/dashboard/app/app.vue
@@ -0,0 +1,7 @@
+<template>
+ <NuxtRouteAnnouncer />
+ <NuxtLoadingIndicator color="var(--az-accent)" />
+ <NuxtLayout>
+ <NuxtPage />
+ </NuxtLayout>
+</template>