daisyui-blueprint 1.6.2 → 1.6.4-beta
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/README.md +45 -1
- package/chunks/chunk-1.js +1 -1
- package/chunks/chunk-10.js +1 -1
- package/chunks/chunk-11.js +1 -1
- package/chunks/chunk-12.js +1 -1
- package/chunks/chunk-13.js +1 -1
- package/chunks/chunk-14.js +1 -1
- package/chunks/chunk-15.js +1 -1
- package/chunks/chunk-16.js +1 -1
- package/chunks/chunk-17.js +1 -1
- package/chunks/chunk-18.js +1 -1
- package/chunks/chunk-19.js +1 -1
- package/chunks/chunk-2.js +1 -1
- package/chunks/chunk-20.js +1 -1
- package/chunks/chunk-21.js +1 -1
- package/chunks/chunk-22.js +1 -1
- package/chunks/chunk-3.js +1 -1
- package/chunks/chunk-4.js +1 -1
- package/chunks/chunk-5.js +1 -1
- package/chunks/chunk-6.js +1 -1
- package/chunks/chunk-7.js +1 -1
- package/chunks/chunk-8.js +1 -1
- package/chunks/chunk-9.js +1 -1
- package/package.json +2 -1
- package/server.js +1 -1
package/README.md
CHANGED
|
@@ -46,7 +46,7 @@ these optional runtime settings:
|
|
|
46
46
|
| --- | ---: | --- |
|
|
47
47
|
| `MAX_PAGES` | `5` | Maximum Page Architect page IDs returned or accepted per call. |
|
|
48
48
|
| `MAX_COMPONENT_RESPONSE_CHARS` | `24000` | Maximum serialized Component Syntax Expert response size in characters. The first requested entry is always returned even if it alone exceeds the budget. |
|
|
49
|
-
| `MAX_COMPONENT_SKIPS` | `7` | Maximum Page Architect
|
|
49
|
+
| `MAX_COMPONENT_SKIPS` | `7` | Maximum components from selected Page Architect pages that may omit syntax retrieval. |
|
|
50
50
|
| `QA_MAX_FILES` | `300` | Maximum files Quality Inspector may audit per call. |
|
|
51
51
|
| `QA_MAX_MB` | `2` | Maximum total megabytes Quality Inspector may audit per call. |
|
|
52
52
|
| `FIGMA_TIMEOUT` | `15` | Figma API request timeout in seconds. |
|
|
@@ -64,12 +64,45 @@ next entry would exceed the serialized response budget. When its next step lists
|
|
|
64
64
|
batch advances to implementation and, for non-conversion workflows, Quality Inspector. A single
|
|
65
65
|
entry is always returned so an unusually large component cannot stall the workflow.
|
|
66
66
|
|
|
67
|
+
Creative Director and Page Architect are mandatory when the agent creates a new complete page or
|
|
68
|
+
screen. Call Creative Director first, then Page Architect. Both tools are optional for other tasks.
|
|
69
|
+
Conversion workflows skip both tools. Component Syntax Expert remains mandatory before markup.
|
|
70
|
+
|
|
71
|
+
A Page Architect call with only `context` returns page options. It preserves the active page
|
|
72
|
+
selection and retrieved syntax. Call Page Architect again with the chosen `PageIDs` before
|
|
73
|
+
Component Syntax Expert. Only the explicitly selected pages create component requirements.
|
|
74
|
+
|
|
75
|
+
Inspector checks static component use only in the supplied files. A component that is absent from
|
|
76
|
+
those files can exist elsewhere in the page. These absence findings are advisory and do not require
|
|
77
|
+
extra components or a wider edit scope. Missing required syntax calls remain blocking findings.
|
|
78
|
+
|
|
67
79
|
For Quality Inspector, use `auditIntent: "fix_changes"` after Blueprint edits UI code; findings
|
|
68
80
|
may then be fixed only within the supplied changed scope. Use `auditIntent: "report_only"` only
|
|
69
81
|
when the user explicitly requests a read-only audit, and report findings without editing files.
|
|
70
82
|
Every `files[].path` must be relative to the `projectRoot` stored by Setup Expert, such as
|
|
71
83
|
`src/App.tsx` or `.`; never pass an absolute path.
|
|
72
84
|
|
|
85
|
+
Blueprint creates one design by default. It creates variants only when the user explicitly asks for
|
|
86
|
+
them. Use separate sub-agents when the client supports them and capacity is available. Otherwise,
|
|
87
|
+
create the variants sequentially in the same agent. Each variant must use a distinct `workflowId`
|
|
88
|
+
and start with Setup Expert. Give parallel variants separate files. Assign one coordinator to
|
|
89
|
+
shared files.
|
|
90
|
+
|
|
91
|
+
When Creative Director supplies guidance, record four planning decisions before implementation.
|
|
92
|
+
Pass the current `concept`, `domainAnchor`, `signatureElement`, and `compositionRule` values to each
|
|
93
|
+
`fix_changes` audit in `originalityEvidence`. Update the values if the design decisions change.
|
|
94
|
+
Include an intentional monochromatic, tone-on-tone, or material palette in the relevant planning
|
|
95
|
+
decision. Inspector uses this evidence and the original `userRequest` to check repeated anchor
|
|
96
|
+
colors. Keep `userRequest` faithful to the user's request; put your own design decisions in
|
|
97
|
+
`originalityEvidence`.
|
|
98
|
+
Quality Inspector validates the four fields and returns a workflow finding when they are missing.
|
|
99
|
+
`SKIP_CREATIVE_DIRECTOR=true` and `report_only` audits do not require these fields. When Creative
|
|
100
|
+
Director supplied guidance, the fields remain required if a browser is unavailable.
|
|
101
|
+
|
|
102
|
+
These caller-provided decisions are planning records, not proof of originality. Quality Inspector
|
|
103
|
+
does not verify when they were written, enforce unchanged values, or confirm that the UI implements
|
|
104
|
+
them. Rendered review remains a separate check.
|
|
105
|
+
|
|
73
106
|
Then ask your AI agent to use daisyUI Blueprint:
|
|
74
107
|
|
|
75
108
|
- "Build a product management dashboard. Use daisyUI Blueprint."
|
|
@@ -77,3 +110,14 @@ Then ask your AI agent to use daisyUI Blueprint:
|
|
|
77
110
|
- "Convert this Bootstrap form to daisyUI."
|
|
78
111
|
- "Create a daisyUI theme from this image."
|
|
79
112
|
- "Check my UI with daisyUI Blueprint and fix the issues."
|
|
113
|
+
|
|
114
|
+
## Development tests
|
|
115
|
+
|
|
116
|
+
Run `bun test --no-env-file` for the default suite. Built-server integration uses synthetic
|
|
117
|
+
credentials and a strict test-only license-response mock. It also checks that missing and rejected
|
|
118
|
+
credentials stop startup. The production server still validates every license.
|
|
119
|
+
|
|
120
|
+
For the live license-service check, set `LICENSE` and `EMAIL` in the process environment, then run
|
|
121
|
+
`bun run --no-env-file test:integration:live`. This sets `CHECK_LIVE_MCP_INTEGRATION=true`.
|
|
122
|
+
The live check fails with a clear error when the credentials are missing. Remote image checks
|
|
123
|
+
remain separate and require `CHECK_REMOTE_IMAGES=true`.
|