Expose designed-mode comparison as the sensory_compare_products agent tool#473
Open
kgdunn wants to merge 3 commits into
Open
Expose designed-mode comparison as the sensory_compare_products agent tool#473kgdunn wants to merge 3 commits into
kgdunn wants to merge 3 commits into
Conversation
… tool Add a @tool_spec wrapper (sensory_compare_products) over process_improve.sensory.compare_products, so the designed-treatment factorial ANOVA + Tukey/Dunnett post-hoc is callable as a JSON-in / JSON-out agentic tool alongside the existing sensory_* tools and through the MCP server. Inputs are panel row-records plus the factor/block/within/control arguments; the result carries the ANOVA table, Tukey and Dunnett contrasts, the compact-letter display and per-level means (NaN cleaned to null for JSON). Tests in tests/test_sensory.py drive it through execute_tool_call and assert it recovers a planted formulation-by-condition interaction and the Dunnett flags. Version bump 1.60.0 -> 1.61.0 with matching CITATION.cff and CHANGELOG entry. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018Ra2KAdJeGtnUzi68yDFaR
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
…anches
Include the optional `within` column in the missing-columns guard so a bad
`within` returns a clean {ok: false} error instead of raising outside the
try/except. Add tests for the pooled single-factor / no-block path (empty
Dunnett, 'stratum' column) and the bad-within error path, lifting patch
coverage over the previously untested branches.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018Ra2KAdJeGtnUzi68yDFaR
Add a test where a bad 'primary' passes the column guard but makes the
underlying compare_products raise, exercising the tool's (KeyError, ValueError)
handler that returns {ok: false}. Closes the last uncovered lines in the wrapper.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018Ra2KAdJeGtnUzi68yDFaR
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
sensory_compare_products, a@tool_specagent-callable wrapper overprocess_improve.sensory.compare_products(merged in Add designed-mode panel comparison (factorial ANOVA + Tukey/Dunnett post-hoc) #471), so the designed-treatment factorial ANOVA + Tukey HSD / Dunnett-vs-control post-hoc is available as a JSON-in / JSON-out tool alongside the existingsensory_reshape_to_long/sensory_validate_descriptive/sensory_analyze_descriptive/sensory_panel_checktools, and through the MCP server (get_sensory_tool_specs()).factors/block/primary/within/control/interactions/alpha/conf_level; the result carries the ANOVA table, Tukey and Dunnett contrasts, the compact-letter display, per-level means, and the resolved config. NaN cells (e.g. theResidualrow's F/p) are cleaned tonullso the payload is JSON-serialisable.{ok: false, errors: [...]}(mirroringsensory_panel_check); factor/level errors from the underlying call are surfaced the same way.Test plan
uv run pytest tests/test_sensory.py -k compare_products --no-cov(2 tests): drives the tool throughexecute_tool_call, asserts the payload is JSON-serialisable, that it recovers a planted formulation-by-condition interaction (T2 collapses only under HB in the letter display) and the Dunnett-vs-control flags, and that missing columns are rejected.sensory_compare_productsappears inget_sensory_tool_specs()(MCP discovery).uv run ruff check .-> All checks passed;uv run mypy src/process_improveclean.tests/test_tool_spec.py::...::test_registered_in_registryis a pre-existing xdist +--new-firstordering flake (fails identically on cleanmainin a local run, green in CI); this PR does not touch it.Checklist
pyproject.toml(MINOR: new agent-tool API surface),1.60.0->1.61.0ruff check .passesCHANGELOG.mdupdated (andCITATION.cffversion synced)Generated by Claude Code