You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: CHANGELOG.md
+3-3Lines changed: 3 additions & 3 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -8,9 +8,9 @@ The format is based on Keep a Changelog, and this project currently tracks chang
8
8
9
9
### Added
10
10
11
-
- Added five original, school-neutral DOCX templates for classic, bordered, engineering, course-design, and modern-minimal report styles, together with a machine-readable catalog and reproducible builder.
11
+
- Added ten original, school-neutral DOCX templates covering classic, bordered, engineering, course-design, modern-minimal, compact-header, review-panel, code-notebook, data-analysis, and project-dossier report styles, together with a machine-readable catalog and reproducible builder.
12
12
- Added template selection and catalog-audit commands. User-supplied templates always win; built-ins are checked for institution names, example identities, embedded media, provenance, and license metadata.
13
-
- Added a persistent five-template integration test that builds every neutral template through the main report pipeline.
13
+
- Added a persistent ten-template integration test that builds every neutral template through the main report pipeline, plus a five-family uploaded-template fidelity corpus.
14
14
- Added the `universal_report` core package for cross-machine configuration, structured pipeline traces, content planning, template contracts, image manifests, format validation, and strict visual validation.
15
15
- Added `TemplateStyleContract` analysis with effective style inheritance, template-hash/version caching, page and section settings, role-based typography, table geometry, headers, footers, placeholders, and structural risks.
16
16
- Added ten generated compatibility fixtures, Python unit tests, core smoke tests, five universal end-to-end scenarios, and real-template regression coverage.
@@ -30,7 +30,7 @@ The format is based on Keep a Changelog, and this project currently tracks chang
30
30
31
31
### Changed
32
32
33
-
- The CLI, PowerShell wrappers, Web UI, documentation, and skill instructions now ask about a preferred template once and otherwise select from exactly five neutral built-in templates.
33
+
- The CLI, PowerShell wrappers, Web UI, documentation, and skill instructions now ask about a preferred template once and otherwise select from exactly ten neutral built-in templates.
34
34
- The former `experiment-report-template.docx` and `course-design-report-template.docx` files are now compatibility aliases for the classic laboratory and course-design templates.
35
35
- User-supplied templates now default to fidelity-preserving mode; repository style normalization and course-design enhancements run only when explicitly requested or when using a repository default template.
36
36
- The main wrappers and Web UI now expose structured progress, generation status, template/format/image/visual artifacts, exact image-count requests, and actionable quality recommendations.
Copy file name to clipboardExpand all lines: ROADMAP.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -35,7 +35,7 @@ Status: substantially implemented by the universal report refactor.
35
35
36
36
Before adding more document types, the current experiment-report flow should become easier to trust, debug, and reuse.
37
37
38
-
Completed foundations include user-template-first fidelity, five reproducible school-neutral fallback templates, catalog-based automatic selection, provenance/identity auditing, and a five-template pipeline regression.
38
+
Completed foundations include user-template-first fidelity, ten reproducible school-neutral fallback templates, catalog-based automatic selection, provenance/identity auditing, a ten-template pipeline regression, and a five-family uploaded-template fidelity corpus.
Copy file name to clipboardExpand all lines: SKILL.md
+5-4Lines changed: 5 additions & 4 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -21,11 +21,11 @@ description: Write Chinese university lab reports and course-design reports, or
21
21
- actual steps, code, screenshots, outputs, data, or conclusions
22
22
- whether the user wants a factual report or a clearly labeled sample version
23
23
- If the user already provides enough facts to write the report body, do not stop to ask for optional metadata such as name, class, date, or template files.
24
-
- For a DOCX or other template-controlled deliverable, if the user has not said whether a preferred template exists, ask once: `你有老师、学校或自己认可的优秀 DOCX/DOC 模板吗？有的话请上传，我会优先保持原格式；没有的话，我会从五套不含学校标识的内置模板中选择。`
24
+
- For a DOCX or other template-controlled deliverable, if the user has not said whether a preferred template exists, ask once: `你有老师、学校或自己认可的优秀 DOCX/DOC 模板吗？有的话请上传，我会优先保持原格式；没有的话，我会从十套不含学校标识的内置模板中选择。`
25
25
- Do not ask this template question when the user only wants report-body text. Do not ask it again after the user says no, chooses an internal template, or provides a template.
26
26
2. Write the full report content before touching template formatting.
27
27
3. If critical facts are missing, do not fabricate exact data, screenshots, or measurements.
28
-
4. If a local docx template exists and shell execution is available, run `scripts/analyze-docx-template.ps1 -TemplatePath <template.docx>` first. Use `scripts/extract-docx-template.ps1 -Path <template.docx>` when a readable outline is also useful. If the user has no template, run `python -m universal_report select-template --repo-root <repo> --report-type <type> --course <course>` or choose from the five entries in `examples/report-templates/catalog.json`.
28
+
4. If a local docx template exists and shell execution is available, run `scripts/analyze-docx-template.ps1 -TemplatePath <template.docx>` first. Use `scripts/extract-docx-template.ps1 -Path <template.docx>` when a readable outline is also useful. If the user has no template, run `python -m universal_report select-template --repo-root <repo> --report-type <type> --course <course>` or choose from the ten entries in `examples/report-templates/catalog.json`.
29
29
5. Treat the generated `TemplateStyleContract` and extracted structure as the source of truth. The contract includes effective style inheritance, page setup, title/body/caption roles, table geometry, headers, footers, sections, image placeholders, and structural risks.
30
30
6. If a template exists, adapt the finished content to the template order and field names.
31
31
7. If the user explicitly wants a filled local docx output and the template matches common report patterns, generate a field map with `scripts/generate-docx-field-map.ps1` and then run `scripts/apply-docx-field-map.ps1`.
@@ -51,8 +51,9 @@ description: Write Chinese university lab reports and course-design reports, or
51
51
52
52
- A user-supplied DOCX template has the highest formatting priority. Default to `-TemplateStyleMode preserve`; do not replace its page setup, fonts, paragraph spacing, title hierarchy, tables, borders, headers, footers, sections, or caption style with repository defaults.
53
53
- A user upload always wins over a selected internal template. Do not blend the uploaded template with an internal design.
54
-
- When no user template exists, choose exactly one of: `neutral-classic-lab`, `neutral-bordered-lab`, `neutral-engineering-lab`, `neutral-course-design`, or `neutral-modern-minimal`. Use `-BuiltInTemplateId` for local wrappers when the choice is explicit.
55
-
- Prefer `neutral-engineering-lab` for computer-network, operating-system, database, Java, Web, Android, programming, and software-engineering experiments; prefer `neutral-course-design` for course-design or comprehensive-project reports; otherwise default to `neutral-classic-lab`. Use the bordered or modern template only when the requested visual direction calls for it.
54
+
- When no user template exists, choose exactly one of the ten catalog entries: `neutral-classic-lab`, `neutral-bordered-lab`, `neutral-engineering-lab`, `neutral-course-design`, `neutral-modern-minimal`, `neutral-compact-header-lab`, `neutral-review-panel-lab`, `neutral-code-notebook-lab`, `neutral-data-analysis-lab`, or `neutral-project-dossier`. Use `-BuiltInTemplateId` for local wrappers when the choice is explicit.
55
+
- Prefer `neutral-code-notebook-lab` for programming, algorithm, code, test, and debug experiments; `neutral-data-analysis-lab` for measurement, statistics, raw-data, and error-analysis work; `neutral-compact-header-lab` for short weekly records; and `neutral-review-panel-lab` when teacher comments, score, or signature fields are required.
56
+
- Prefer `neutral-engineering-lab` for computer-network, operating-system, database, Web, Android, and software-engineering experiments. Prefer `neutral-project-dossier` for long-form system projects and `neutral-course-design` for ordinary course-design reports. Otherwise default to `neutral-classic-lab`; use the bordered or modern template when the requested visual direction calls for it.
56
57
- Internal templates must pass `python -m universal_report audit-template-catalog --repo-root <repo>`. They may not contain real school/college names, logos, watermarks, example student identities, embedded third-party media, or copied sample report content.
57
58
- Treat the internal DOCX files as original neutral reconstructions under the repository license. Public templates and GitHub projects are design references only; do not copy their branding, sample text, media, or restricted assets into the internal files.
58
59
- Before rendering, create or reuse a `TemplateStyleContract`. After rendering, run `scripts/validate-docx-format.ps1` and report the exact failed contract items instead of saying only that formatting is wrong.
short_description: "Generate Chinese lab and course-design reports from templates, screenshots, code, tutorial pages, and results"
4
-
default_prompt: "Use $experiment-report to create or complete a Chinese lab or course-design report. If I need a DOCX and have not said whether I have a preferred template, ask once; preserve my uploaded template, otherwise choose from the five neutral built-in templates."
4
+
default_prompt: "Use $experiment-report to create or complete a Chinese lab or course-design report. If I need a DOCX and have not said whether I have a preferred template, ask once; preserve my uploaded template, otherwise choose from the ten neutral built-in templates."
0 commit comments