@acedatacloud/skills 2026.818.3 → 2026.819.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@acedatacloud/skills",
|
|
3
|
-
"version": "2026.
|
|
3
|
+
"version": "2026.819.1",
|
|
4
4
|
"description": "Agent Skills for AceDataCloud AI services — music, image, video generation, LLM chat, web search. Compatible with Claude Code, GitHub Copilot, Gemini CLI, OpenAI Codex, and 30+ AI coding agents.",
|
|
5
5
|
"keywords": [
|
|
6
6
|
"agent-skills",
|
|
@@ -116,7 +116,7 @@ For generated hero media, the first capability-specific call MUST be the provide
|
|
|
116
116
|
|
|
117
117
|
Poll according to the returned interval until terminal. For activation/source preparation, reserve up to two minutes (for example eight 15-second polls). If still pending, record one `ADC-SPOTLIGHT-SOURCE:v1` draft artifact with task ID and fingerprint, make no completion claim, and end without Maestro.
|
|
118
118
|
|
|
119
|
-
Inspect every selected asset at full resolution. Generated media MUST use a real accepted result. Product/Agent/deployment stories use faithful, sanitized screenshots or lifecycle evidence. Full request JSON is evidence-only; customer-facing frames never contain internal review or provenance language.
|
|
119
|
+
Inspect every selected asset at full resolution. Before Blueprint approval, perform a complete body download for every selected URL, require non-zero bytes, decode at full resolution, and inspect pixel-level semantic fit to `HERO_CASE` and its assigned material role. A metadata, filename, URL, or task status alone never proves availability, visual content, or semantic relevance. A failed download, decode, or semantic-fit check makes that asset ineligible: choose another asset or candidate and do not submit Maestro. Generated media MUST use a real accepted result. Product/Agent/deployment stories use faithful, sanitized screenshots or lifecycle evidence. Full request JSON is evidence-only; customer-facing frames never contain internal review or provenance language.
|
|
120
120
|
|
|
121
121
|
## 5. Require a rich material plan
|
|
122
122
|
|
|
@@ -214,10 +214,11 @@ Follow it with the complete Blueprint and role→asset provenance. The outer tas
|
|
|
214
214
|
The Blueprint defines archetype-specific gates plus these invariants:
|
|
215
215
|
|
|
216
216
|
- decoded frame 0 contains clear brand or product desire—not blank/loading/disclaimer;
|
|
217
|
-
-
|
|
217
|
+
- for a 30-second format, final-byte `ffprobe` duration is 29.95–30.05 seconds; any other requested format uses the same ±0.05-second tolerance;
|
|
218
|
+
- the selected real hero occupies a visually primary region for at least 40% of runtime, including one near-full-frame hold of at least 3 continuous seconds; authored type/UI may support it but never substitute for it;
|
|
218
219
|
- composition variety and pacing match this run's genome;
|
|
219
220
|
- no static material is dragged beyond the limit appropriate to the archetype;
|
|
220
221
|
- customer copy contains no internal review language, private sourcing facts or invented claims;
|
|
221
222
|
- narration/audio are complete and the CTA is readable.
|
|
222
223
|
|
|
223
|
-
Make an actual `Skill` call with `skill="visual-review"` against complete initial evidence. Fix blockers in one concentrated pass. Rebuild evidence from the exact final MP4 bytes and make a second actual confirmation call even when no initial blocker exists. The Reviewer receives the Blueprint, Creative Genome, role map, and evidence and judges whether this film fulfilled **its chosen director language**, not whether it resembles previous episodes.
|
|
224
|
+
Make an actual `Skill` call with `skill="visual-review"` against complete initial evidence. Fix blockers in one concentrated pass. Rebuild evidence from the exact final MP4 bytes, measure duration with `ffprobe`, verify hero runtime from the final contact sheet, and make a second actual confirmation call even when no initial blocker exists. If duration, hero-share, audio, CTA, or either review fails, Maestro MUST NOT return an accepted submission. The Reviewer receives the Blueprint, Creative Genome, role map, and evidence and judges whether this film fulfilled **its chosen director language**, not whether it resembles previous episodes.
|