dsh-draw 0.1.0

package/THIRD_PARTY_NOTICES.md

@@ -0,0 +1,21 @@
+# Third-party notices
+
+`dsh-draw` bundles no third-party source code; all TypeScript/JavaScript
+sources in this repository are original works by the dsh-draw contributors,
+licensed under Apache-2.0 (see `LICENSE`).
+
+The package depends on the following software. The published tarball bundles
+only `zod` (the wire codec library inlined into the host and client bundles);
+everything else is install-time:
+
+| Package | Version range | License | Purpose |
+|---|---|---|---|
+| [zod](https://github.com/colinhacks/zod) | `^4.4.3` | MIT | Wire codecs shared by the host and client Typert faces (bundled into `lib/`) |
+| [tsdown](https://github.com/rolldown/tsdown) | `^0.22.14` | MIT | Build-time bundling of `lib/` (a regular dependency so the git-install channel's `prepare` script can build) |
+| [typescript](https://github.com/microsoft/TypeScript) | `^7.0.2` | Apache-2.0 | Build-time declaration emission (`lib/types/`) |
+| [@deepseek-ai/cordis](https://www.npmjs.com/package/@deepseek-ai/cordis) | `^4.0.1` (peer) | See package | The plugin runtime |
+| [@deepseek-ai/schemastery](https://www.npmjs.com/package/@deepseek-ai/schemastery) | `^3.18.0` (peer) | See package | Configuration schema |
+| `@deepseek-ai/dsh-*` peers | `0.1.0-rc.6` (peer) | See packages | Official harness seams (`dsh-tools`, `dsh-session`, `dsh-attachment`, `dsh-credentials`, `dsh-llm`, `dsh-brand`, `dsh-typert-protocol`) |
+
+At runtime the plugin only talks to the image-generation endpoints the user
+configures and to the harness services listed as peerDependencies.

package/cordis.patch.yml

@@ -0,0 +1,48 @@
+# dsh-draw bundle patch: mount the unified image-generation router.
+#
+# Every key below is a Config field (Schemastery schema); invalid values fail
+# the profile load loudly. See README.md "Configuration" for the full table.
+- insert:
+    - id: dsh-draw
+      name: dsh-draw
+      config:
+        # Ordered engine chain; the router walks it top-down and falls back on
+        # failure. apiKeyRef names an environment variable (credential
+        # reference) — never put a literal key here.
+        engines:
+          - id: openai
+            baseUrl: https://api.openai.com/v1
+            model: gpt-image-1
+            apiKeyRef: OPENAI_API_KEY
+            enabled: true
+            sizeMap: { square: 1024x1024, landscape: 1536x1024, portrait: 1024x1536, auto: auto }
+            qualitySupported: true
+            styleSupported: true
+            responseFormat: b64_json
+            imageMediaType: image/png
+          - id: cogview
+            baseUrl: https://open.bigmodel.cn/api/paas/v4
+            model: cogview-3-flash
+            apiKeyRef: ZHIPU_API_KEY
+            enabled: true
+            sizeMap: { square: 1024x1024, landscape: 1344x768, portrait: 768x1344, auto: 1024x1024 }
+            qualitySupported: false
+            styleSupported: false
+            responseFormat: url
+            imageMediaType: image/png
+        # Engine id the router prefers; must name a configured engine.
+        defaultEngine: openai
+        # Per-generation HTTP timeout in milliseconds (1000..600000).
+        requestTimeoutMs: 120000
+        # Cap on images one call may produce (1..10).
+        maxImagesPerCall: 4
+        # Prompt length cap in characters (1..32000).
+        maxPromptLength: 4000
+        # Per-session generation-call cap (1..100000).
+        maxGenerationsPerSession: 200
+        # Per-session generated-image byte cap (1048576..4294967296).
+        maxBytesPerSession: 209715200
+        # Consecutive failures before an engine enters cooldown (1..10).
+        failureThreshold: 2
+        # Engine cooldown in milliseconds after the threshold trips (1000..3600000).
+        cooldownMs: 60000