v0.24.0 — Portable research workbench and source-bound writing

v0.24.0 — Portable research workbench and source-bound report writing

This release connects selected research state, author baselines and local figure scripts across
folders and tool-using hosts. CLI/Python and the optional local MCP server use the same business
functions. The package supplies scientific tools and current context; the host develops the argument.

Complete task paths

• Export, inspect and restore a selected workbench, including its current sources, author companion,
  baseline and local scripts. Internal declarations become relative; explicitly mapped external
  result roots relocate without rewriting author prose or script strings.
• Use nine task-sized functions for material inspection, analysis, saved reports, surface research,
  writing, manuscript export, author return and workbench movement. No model API is mandatory.
• Select exact JSON numbers from native reports, named method/setup records and supplied reference
  files. Prepare bound Methods/Results/Discussion and a native Results table without manufacturing
  a general topic proposal or copying numbers into a second evidence list.
• Compare defined scalars using differences, ratios, relative changes or reductions. The reference
  denominator, both parent supports and located method definition travel with each comparison.
• Supply source-linked structured formulas once and use their math tokens in the draft. Word keeps
  native editable fractions, subscripts and superscripts; author-local expressions take priority.
• Continue from retained current reports and author drafts. Recalculate affected values and table
  cells while keeping wording; source-definition or interpretation changes still need host judgment.

Installed examples

Install the matching wheel with [docs] for Word output or [docs,mcp] for a local MCP host.
Run the public scripts from any writable directory using absolute script/output paths:

python /path/to/report-writing/run_example.py /path/to/new-report-output
python /path/to/author-collaboration/run_example.py /path/to/new-author-output
python /path/to/portable-workbench/run_example.py /path/to/new-author-output /path/to/new-portable-output

The report example uses constructed records and tests a denominator/source refresh plus native
math. The portable example moves an analytical author manuscript and preserves its current values,
wording and comments. These runnable examples demonstrate installation and collaboration behavior,
not independent CFD validation. See the host guide for real-task use.

Evaluation and release evidence

Known-project pressure-report and conjugate-heat-transfer paths use existing mature cooling results.
Actual second-host execution uses Claude Code configured with DeepSeek, not an official Claude model; connection,
tool execution, first writing and feedback-assisted revision are evaluated separately. Earlier
report trials exposed denominator and interpretation failures rather than being recorded as passes.
The single follow-up completed preparation, bound writing and native Word export through the new
interface, with a CLI-reported cost of USD 2.446 under its USD 3 setting. All 64 selected/derived
values matched the unchanged source records, relative percentages used their declared reference
denominators, and native fractions/superscripts rendered correctly in a two-page candidate.
This numerical agreement does not mean the first draft passed scientific review.
It still called one static-pressure difference a loss and inadequately named one mixed-weight
ratio. These are residual interpretation errors, not accepted scientific findings. The first
candidate is retained unchanged; the result supports the tool path, not autonomous scientific writing.

The run also exposed an oversized full-report read and an unnecessary question-ID coverage
requirement. A compact writing view and evidence duties that distinguish a question from a finding
address those concrete product problems; they are post-trial fixes, not another host retry.

The six Windows/Linux Python 3.10–3.12 test jobs and installed-package smoke check passed.
The wheel and source distribution below are the artifacts of the verified PR build; the merge
has the same source tree. An isolated wheel replay outside the checkout also passed.
Verified PR CI and
push CI; source: 1c17c383f0b60b2d176a218bf01f4291d7bebaf3.

Retained scope

The flow and heat-transfer tasks are different paths in a known project, not two unfamiliar domains.
Suitable mature real transient or multiphase CFD materials remain unavailable. This third-class
evaluation stays not yet evaluated; saved-time operators and battery control-model regressions
do not replace it. This version does not establish broad scientific generality or submission readiness.

Selective workbench transfer is not whole-project SQLite migration. Native solver archives remain
external; their verified extraction scope is unchanged. A matching support-domain declaration is not
proof of physical equivalence or causality. Formula trees provide native notation, not a new formula
solver. Structured comparisons and automatic token refresh cannot revise scientific interpretations.

Scientific interpretations still require author and host review.

v0.23.0 — Formal manuscripts and Word author collaboration

v0.23.0 — Formal manuscripts and Word author collaboration

This release connects the research workspace to an editable manuscript, an author's Word return,
and a fresh candidate with current scientific values. Analysis definitions, evidence tokens,
figures and native equations continue to use the existing research pipeline.

What changes for the author

• Paragraphs, figures, captions, tables and equations have a shared order and stable identity.
  DOCX, Markdown and LaTeX use this common scientific content.
• cfdpaper author prepares a Word manuscript and recovers ordinary prose, captions, supported
  local formatting and anchored comments. The original return is kept alongside the new source
  candidate. Non-overlapping edits merge; conflicting edits retain both versions for resolution.
• Local Figure/Fig. presentation follows subsequent numbering. A clear whole-sentence removal
  can remove its binding; changing a number is not treated as a scientific source update.
• research --author-candidate carries recovered local-ID drafts and presentation into the
  existing research workspace. Current source definitions and author-local plotting scripts drive
  recalculation and redrawing, rather than being replaced by the Word file.
• Research, report and conference profiles share scientific facts. Word retains native tables
  and equations; LaTeX keeps figures with their captions and separates table notes from bottom rules.

Try the complete public example

Install the wheel with the docs extra and run
examples/author-collaboration/run_example.py
from any writable directory. It starts with analytical pipe-pressure CSV rows, prepares Word,
demonstrates a clearly attributed automated edit/comment, changes the source viscosity and recovers
the wording with the new pressure values. No solver or external model service is required.

python -m pip install "cfd_paper_agent-0.23.0-py3-none-any.whl[docs]"
python /path/to/author-collaboration/run_example.py /path/to/new-output

The source-distribution and public-examples archive include the runnable example.

Real-task use

The retained cooling study was exercised with the author's actual Word edits and comment, then
with isolated source-value and observation-definition changes. The original results, manuscript
and plotting scripts were preserved. Research and template-based Word outputs were opened in
Word and inspected as rendered pages. A conference view was actually compiled using the official
IEEEtran class through a local bridge; this is a template exercise, not an IEEE submission claim.

The author edits exposed and helped fix copied figure locators, Word-localized style identities,
reference abbreviations and complete bound-clause deletions. The research workspace now provides
an explicit continuation route rather than requiring the host to paste all recovered prose back.

Software verification

The release is checked with the full test suite on Windows and Linux using Python 3.10–3.12,
plus lint/format checks and a clean-wheel installation replay. The public example is run outside
the checkout, without editable imports. Current run details are available in
GitHub Actions.
The release assets contain the wheel, source distribution and portable public examples with
rendered research/report previews.

Scope

Word return is a managed, bounded collaboration route, not arbitrary lossless DOCX conversion.
Numerical cells, formulas, citation fields, modified image content, ambiguous anchors and
unaccepted tracked changes require object-specific resolution. Ordinary comments are suggestions,
not approval. Numerical/definition changes identify affected interpretations; the host and author
still decide how to revise mechanisms, conclusions and literature comparisons.

Word template import uses named styles and final-section geometry. Legacy template compatibility
and automatic caption numbering are handled in the generated copy; arbitrary multi-section
publisher layouts are not reproduced. A profile is not evidence of journal compliance or
scientific readiness. Existing v0.22 scientific analyses and native-backend scope remain unchanged.

Verified CI: https://github.com/Xiuyiw/CFD-SCIPaper-Agent/actions/runs/39370692394

Source: 02d994f8a6affcf3981425a6a5996ada35b4d8a0 (same source tree as the verified PR build).

v0.22.0 — Physical diagnostics and mechanism evidence

v0.22.0 physical diagnostics and mechanism evidence

This increment connects original-field regions, signed transfers, matched spatial changes
and defined control-volume budgets to the existing research workspace. The host can use
those observations to test a physical explanation instead of assembling separate scalar
spreadsheets, plots and prose by hand. Scientific interpretation remains with the host and author.

Implemented

• field_integral distinguishes density times physical measure, already integrated element
  totals, vector flux dotted with oriented area, weighted means and separately supplied
  positive/negative integrated powers. Temperature-area sums are not heat. Missing values
  remain unavailable rather than becoming zero.
• Regions use explicit overlap fractions, declared centroid assignment or original planar
  polygon clipping. The geometric partition check detects actual overlaps and uncovered
  portions, not merely fractions adding to one. Curved walls require native reports or
  suitable externally defined fractions. Shapely is an optional geometry dependency.
• field_difference matches physical coordinates, measures and original polygons when
  supplied, retaining both original means and the difference. Equal row counts are not a
  spatial mapping. Absolute-temperature differences use K; percentages of absolute-temperature
  integrals are refused because they depend on temperature origin.
• energy_budget retains input, output, source, storage and the resulting residual with
  a stated control volume. A steady-state storage omission must be explicit. A remainder
  does not become an independently observed path or a model-validation result.
• An optional PyFluent reference backend reads saved CASE/DATA, original boundary polygons,
  temperature, heat flux and native whole-surface or coordinate-clipped reports without
  solving or writing back. The actual backend trial used Windows, Fluent 2023 R1 and
  PyFluent 0.20. Neutral NPZ/CSV consumption does not require Fluent or a license.
• research --diagnostics attaches those calculations, complementary figures and source-bound
  evidence to a new copy of an existing workspace. Compact current scalar readings in
  research-context.json point to complete result definitions and source locations.
  Generated figure sources, editable SVG/PDF, PNG/TIFF and author-local scripts travel together.
• Computed comparisons consume explicit physical supports: shared steady/fixed/window states,
  event-relative observations or consistently defined event timings. Native events and saved
  threshold markers are not interchangeable; only named conditions claimed fixed are matched.
• Literature excerpts can be checked at exact source lines or pages of page-delimited extracted
  text. PDF page index and a printed-page label remain distinct. This checks location, not
  whether the source's scientific claim transfers to a new problem.
• Source-value and definition changes refresh related diagnostics, figures and bound values
  while retaining author scripts and prose. In-place editing of a returned workspace and
  changed diagnostic source paths are recognized. Affected interpretation still needs the host.
• Five existing bundled skills now explain these actual tools and their scientific use.
  Adjacent native Word table captions have explicit spacing without changing body typography.

Install and try

Use CPython 3.10–3.12 and the matching source checkout/archive:

python -m pip install ".[docs,geometry]"
python examples/physical-diagnostics/run_example.py output/physical-demo

Open output/physical-demo/candidate/manuscript.docx. The
tutorial provides six constructed unequal-area
faces, opposing transfers, a matched difference and a separate budget. It replays disclosed
recorded host prose through Methods, Results, Discussion and Conclusions, five figures and
two native tables. The original arrays are not CFD results; the separate budget does not
validate them. --prepare-only leaves the drafts for a host to write.

Examples are in the source archive, not the wheel. For a release wheel:

python -m pip install "./cfd_paper_agent-0.22.0-py3-none-any.whl[docs,geometry]"

Use the optional fluent extra only in an environment that can access the licensed solver.
The research guide covers saved-results extraction, diagnostic
declarations, manuscript attachment and continued author editing. The core works without a
model API; the host supplies the research judgment and prose.

Actual research use and what improved

A known four-case conjugate-cooling task used existing original surface fields, regional
native reports and signed reconstructed conduction. It distinguished a moving hot footprint
from fixed-coordinate changes, mean-temperature composition from heat accounting, and local
wetted area from per-area heat transfer. The figures and manuscript retained the differing
structure–location responses rather than forcing a universal enhancement explanation.

Two isolated first writers received the same new-evidence materials. Both produced supported
scientific interpretations. The preceding tool route required task-specific regional mathematics
and a scalar bridge; the candidate used reusable physical calculations and bound figure/text
outputs. This supports less bespoke processing, not superior autonomous discovery or a measured
percentage saving in time/tokens. Initial candidates and their differing strengths were retained.
Two local wording corrections were kept separate from product improvements.

The first-use trial exposed stale attached-result context; the entry now publishes current
diagnostic readings before writing. Independent numerical review also found an origin-dependent
temperature-integral comparison and a false polygon-partition acceptance. Both have reproductions
and targeted fixes; valid real-region numerical values were unchanged.

Existing thermal-model histories separately demonstrated valid common-time comparisons and
rejection of own-event observations consumed as equal-time evidence. This was a time-support
regression, not a multiphase CFD validation. Separate source/region-definition perturbations
exercised targeted update and retention of author prose and local plotting scripts. Those copies
are software exercises, not additional scientific cases.

Actual Word/PDF pages and figures were inspected locally. Generic platform CI and installed-wheel
examples exercise the software; they do not establish cross-platform native solver behavior or
independent physical validation. Release checks are recorded in the matching GitHub workflow runs.

Scope and next work

The native reference path is limited to the tested surface quantities and backend environment.
STAR-CCM+ recognition/export guidance and the existing limited neutral VTK path are unchanged.
Planar clipping does not become a generic mesh engine. Gradient reconstruction and a budget
remainder from the same saved solution remain complementary diagnostics, not independent validation.

The Workbench plan retains limited Word-edit return
and document profiles for v0.23, and second-host/heterogeneous evaluation for v0.24. No automatic
SCI readiness, universal solver coverage or new simulation capability is claimed. The
v0.21.0 notes remain the previous release record.

Validated by Windows/Linux Python 3.10–3.12 and installed-package CI: all seven jobs passed. The merged source tree is identical to the verified candidate. Attached wheel and source archive are the CI-built distributions; the example ZIP contains only redistributable constructed data.

v0.21.0 — Connected Research Workbench

Connected surface research workbench

v0.21.0 connects saved surface results to analysis, complementary figures, source-bound writing and an editable research manuscript. The host AI and author remain responsible for the scientific argument.

What is new

• A portable cfdpaper research workspace: original face values, physical areas and polygon geometry feed weighted statistics, exact threshold-exceedance curves, tables and four connected manuscript sections.
• Targeted source/definition refresh with author-local scripts and drafts retained. Changed interpretations are identified for host review rather than silently rewritten.
• paper-framework-figure-studio-pro guidance and research --attach-figure connect reviewed mechanism artwork to captions, body references and manuscript numbering. Raster concepts remain explicit and schematic-only.
• A runnable public surface-research tutorial plus a reviewed synthetic Word/PDF example, figure formats and source CSVs.
• Cross-platform fixes for narrow-map fallback-font labels and Windows short/long temporary-path aliases.

Validation

Release-candidate CI passed all six Windows/Linux × Python 3.10–3.12 jobs and the independent Windows wheel installation/CLI/example replay job. The merged main tree is identical to that validated candidate.

An existing four-field cooling task also produced a connected research draft and a reviewed five-page candidate with mechanism artwork. This is a known-project workflow demonstration, not independent scientific validation or an automatic SCI-readiness claim.

Download and use

Download the wheel below and install Word support:

python -m pip install "./cfd_paper_agent-0.21.0-py3-none-any.whl[docs]"
cfdpaper research --help

The examples ZIP contains runnable tutorials and the reviewed synthetic manuscript/figures. Alternatively, use the matching source archive.

Full release notes · Surface-research tutorial

Scope

The new route consumes declared exported fields; it is not arbitrary native-solver ingestion. Automatic Word-edit merging and deeper physical diagnostics remain future work. Concept illustrations are not heat-flux evidence, and existing private CFD materials are not redistributed.

Post-merge main CI also passed all seven jobs before publication.

v0.20.0 Effective methods and saved contrasts

What's new

• Static Python method indexes and exact qualified-name reads through materials/inspect, without executing the source project.
• Same-record temporal differences with original readings, source locations, exact saved windows and shared numeric/plot declarations; temperature differences in K.
• Optional inline CSL metadata separates publication references from evidence descriptions.
• Explicit section-end figures preserve default layout, dimensions and fonts.
• Public effective-methods example and Chinese engineering trial guide.

Verification and scope

Windows/Linux Python 3.10–3.12 tests and exact installed-wheel smoke pass. The public tutorial includes a real editable Word document, native table/math, plots/source data and portable continuation. Actual Word/PDF pages were inspected. Original scientific materials remain unchanged; no new solver run or fit.

A same-material installed v0.19/v0.20 first-writing exercise was independently reviewed as mixed: a useful boundary-direction diagnostic, but remaining host method/support/locator issues. Original first drafts are preserved; one feedback revision is not claimed as autonomous or first-draft improvement. Scientific interpretation and final publication decisions remain author/host responsibilities.

Release notes · 中文工程试用指南

v0.19.0 — Source-bound methods and paired observations

Highlights

• Bind finite method parameters directly to exact JSON fields, with explicit units and precision.
• Refresh the same source values in prose, native tables/equations and linked manuscript sections.
• Compute paired model/reference changes while retaining both original observations.
• Show signed deviations and conservative endpoint envelopes in reproducible panels; temperature differences use K.
• Develop scientific Methods and multi-observation arguments with the existing specialist skills.

The public tutorial connects analysis, an absolute-state table, a paired figure, source-bound numerical settings and four manuscript sections. The examples ZIP includes actual installed-package Word/PDF output and editable plotting sources. Install the wheel with the docs extra for Word support.

A new controlled thermal-probe exercise used the same raw materials with installed v0.18/v0.19 hosts. A context-isolated AI review found correct principal numbers and mixed strengths: feedback/observation coupling improved, while a distinct boundary-exchange diagnostic and some publication details remained weaker. This is host-assisted writing, not a demonstrated uniform automatic scientific gain. Original first drafts are retained; private research assets are not included.

Release notes · Paired-observation tutorial

v0.18.0 — Complementary observations for scientific writing

Highlights

• Compute model/reference differences and signed distances outside explicit source ranges.
• Keep reported ranges, figure-reading envelopes and declared bounds distinct.
• Share exact calculation definitions between numerical prose and automatic plots, including temporal windows.
• Combine saved profiles and reference comparisons in editable, source-bound numerical panels.
• Carry complementary observations through connected Methods, Results, Discussion and Conclusions.

Includes a recorded invented four-section tutorial, native Word equation, source CSV, runnable plotting script and SVG/PDF/PNG/TIFF. The same real raw materials were also used for isolated v0.17/v0.18 first drafts: scientific strengths were mixed, not uniformly improved. One bounded feedback revision and a separate chip-cooling regression were completed without solving or refitting original projects.

The attached examples ZIP includes an actual installed-package Word/PDF example. Install the wheel with the docs extra for editable Word output. Private research materials are excluded.

Release notes · Multi-observation tutorial

v0.17.0 — Transient evidence for scientific writing

Highlights

• Calculate nonuniform time integrals and time-weighted means on exact saved windows.
• Keep cumulative increments separate from instantaneous-rate integration.
• Bind sampled peaks, threshold times and temperature contrasts to prose and native Word tables.
• Compare compatible calculated diagnostics with their source units and actual windows.
• Carry transient writing skills, computed evidence and editable figures between model hosts.

Includes an invented three-panel thermal tutorial with source data, runnable plotting code, SVG/PDF/PNG/TIFF and an actual installed-package Word example. A separate known battery-project first-writing trial identified a capacity-conditioned reversal of conductivity effects; original project results remain unchanged. It is not independent experimental validation or a proof of cross-provider scientific equivalence.

Windows/Linux Python 3.10–3.12 integration and installed-package replay passed. Install the wheel with its docs extra for editable Word output. Private research files are excluded.

Release notes · Transient tutorial

v0.16.0 — Scientific context for research writing

Highlights

• Carry located case, solution-stage and reporting-operator facts with manuscript tasks.
• Read literal JSON method parameters through exact Pointers, with explicit scope and truncation.
• Propagate method-source changes into existing section and passage update reports.
• Display each DOI once in the default bibliography.
• Run the CLI with python -m cfdpaper.

The release includes a portable recorded example with native Word table and editable math output. The focused private first-draft trial found mixed strengths; its bounded feedback revision remains separate from first-pass performance.

Windows/Linux Python 3.10–3.12 tests and installed-package example replay passed. Install the wheel with its docs extra for editable Word output. Private research materials are excluded.

Release notes · Scientific context guide

v0.15.0 — Scientific intake and manuscript continuation

Highlights

• Find method definitions and evidence in full source text with exact file/line context.
• Select existing material files directly for focused analysis preparation.
• Continue older manuscript reviews with an explicit display-only refresh, preserving author drafts and scientific inputs.
• Carry the five workflow skills in manuscript and section task packages.
• Strengthen method-stage reporting, mechanism evidence selection and nearest-study discussion.
• Correct tiny publication font-rounding rejections.

Scientific evaluation

A paired known-project first-draft comparison found local gains and remaining weaknesses, not uniform superiority. Both complete drafts and their actual Word pages were examined; one bounded feedback-assisted revision remains distinct from first-pass performance. Software release is not certification that the research example is ready to submit.

Install the wheel with its docs extra for editable Word output. The example ZIP contains public synthetic data, recorded prose, a portable manuscript, DOCX and PDF. Private research data are not included.

Full release notes