Files

T

Yaowei Zheng d4faee3a1e Changelog, dev startup, README, AgentHub 0.4.0, model catalog, and landing site (#7 )

Branch-length batch covering tooling, the model layer, the Web App and the public
surfaces. Highlights:

- Changelog: a per-release `changelog/<version>/` tree, grouped by the surface each
  change touches, with a root CHANGELOG.md holding one line per release.
- Dev startup: `scripts/dev-prebuild.mjs` serializes the skills+core prebuild behind a
  lock and keeps `pnpm install` current; `pnpm dev` runs server+web together.
- AgentHub 0.3.3 -> 0.4.0: OmniMessage complete payloads carry one opaque `fidelity`
  object in place of item-level `signature`/`phase`, threaded verbatim through Trace,
  replay and resume; malformed classification adapted to the new error types.
- Model layer: a model is always referenced by an explicit `(provider, model_id)` pair.
  The provider is never inferred, guessed or defaulted -- both the catalog inference and
  the unique-match config resolution are gone, and CLI, SDK, server routes and
  run_subagent all require the complete pair. Catalog gains the Qwen Token Plan, Qwen
  Pay-As-You-Go and Fireworks AI gateways, plus an expanded OpenRouter group.
- Web App: catalog preset sync and per-group speed test on the Models page, positional
  slash commands, a markdown renderer, skill-library update reminders, and a vertically
  centred draft page whose upward menus size themselves to the room available.
- Public surfaces: restructured READMEs, the penguin.ooo landing site and blog, refreshed
  benchmark results for both suites, and the demo videos playing on the landing page.

Includes the fixes from a full review of the branch: 23 confirmed findings, among them a
provider-inference bug that could send one vendor's API key to another vendor's endpoint,
and an Escape handler that destroyed the composer's contents unrecoverably.

Verified on the branch head: pnpm test (1127 passing, 7 packages), pnpm typecheck and
pnpm format:check clean, Playwright e2e 14/14.

2026-07-21 17:43:31 +08:00

3.4 KiB

Raw Blame History

@prismshadow/penguin-server

The PenguinHarness Web backend — the Web implementation of the SDK's Human boundary. HTTP carries Prompt input, approvals and interrupts; Server-Sent Events stream the OmniMessage output. Adds multi-user auth, Project authorization, Session runtime, scheduling and usage accounting on top of @prismshadow/penguin-core.

Architecture

HTTP: Hono + @hono/node-server; createApp(deps) is pure assembly (no port bind — tests drive it via app.request()); index.ts is the startup entry (dotenv, graceful shutdown).
Storage: SQLite via Node's built-in node:sqlite (WAL) holds only indexes and aggregates (users, auth sessions, Project authorization, Agent/Session indexes, usage, UI prefs, error records). Agent State, Traces and Workspaces stay as files under ~/.penguin/data/<project>/agents/<agent>/, fully shared with the SDK and CLI.
Runtime: a session manager keeps active Sessions (get-or-resume-or-heal, per-Session mutex, run/compact driving); approvals surface over SSE as approval_request and decisions re-read the stored approval mode each time; interrupts converge pending approvals to deny before aborting; a scheduler fires agent_state/schedule/*.toml tasks while the service runs.
SSE: per-channel monotonic event ids with a bounded replay buffer (1000 events / 2MB); reconnects replay from Last-Event-ID or receive resync_required; heartbeat comment every 20s.
Usage: token_usage events are persisted row by row; costs are computed at query time from current per-model pricing.

The full route tables and the SSE protocol are documented in the Server API reference. DTO types are exported for type-only import via @prismshadow/penguin-server/api.

Environment

Variable	Meaning	Default
`PORT` / `HOST`	Listen port / address	`7364` / `127.0.0.1`
`PENGUIN_HOME`	Data root (shared with SDK/CLI)	`~/.penguin/data`
`PENGUIN_WEB_DB`	SQLite file path	`<root>/web.db`
`PENGUIN_WEB_DIST`	Front-end build dir (static hosting + SPA fallback when present)	`../web/dist`, or the bundled `web-dist/` in the npm package

.env in the process cwd is loaded automatically.

Running

pnpm --filter @prismshadow/penguin-server dev     # tsx watch (front end via the Vite dev proxy)
pnpm --filter @prismshadow/penguin-server build   # tsup → dist/
pnpm --filter @prismshadow/penguin-server start   # node dist/index.js

pnpm typecheck / test run tsc and vitest (tests use a temp root + in-memory DB; no ports, no live LLM calls).

Security notes (known MVP limits)

CSRF: session cookie is SameSite=Lax and writes accept only Content-Type: application/json; no CSRF token yet.
No login rate limiting: add throttling at a reverse proxy for public deployments.
Built-in admin starts as admin / penguin-2026: change it immediately (a banner keeps reminding until you do).
Passwords use node:crypto scrypt (scrypt$N$r$p$salt$hash, timingSafeEqual); login sessions renew on a 7-day sliding window; the DB stores only the token's sha256.
Model credentials live in the Project's hidden 0600 config file; the API always masks them.
Behind a reverse proxy, disable response buffering for SSE paths (the server already sends X-Accel-Buffering: no) and forward x-forwarded-proto to enable Secure cookies.

Part of PenguinHarness · Apache-2.0

3.4 KiB Raw Blame History