Skip to content

Redesign CausalPy method selection skill#931

Draft
drbenvincent wants to merge 4 commits into
mainfrom
cursor/redesign-method-selection-skill
Draft

Redesign CausalPy method selection skill#931
drbenvincent wants to merge 4 commits into
mainfrom
cursor/redesign-method-selection-skill

Conversation

@drbenvincent
Copy link
Copy Markdown
Collaborator

Summary

  • Replace the flat method-selection skill routing with ordered intake, output contracts, and a text-based routing algorithm optimized for agent use.
  • Add explicit not-in-CausalPy handling so agents can avoid force-fitting unsupported causal designs.
  • Add capability gates and focused disambiguation references for common routing confusions like ITS vs Piecewise ITS, DiD vs staggered DiD, SC vs SDiD, and IPW vs IV.

Test plan

  • prek run --files causalpy/skills/choosing-causalpy-methods/SKILL.md causalpy/skills/choosing-causalpy-methods/reference/experiment_decision_guide.md causalpy/skills/choosing-causalpy-methods/reference/decision_tree.md causalpy/skills/choosing-causalpy-methods/reference/not_in_causalpy.md causalpy/skills/choosing-causalpy-methods/reference/method_capability_matrix.md causalpy/skills/choosing-causalpy-methods/reference/disambiguation/its_vs_piecewise_its.md causalpy/skills/choosing-causalpy-methods/reference/disambiguation/its_vs_sc_vs_did.md causalpy/skills/choosing-causalpy-methods/reference/disambiguation/did_vs_staggered_vs_panel.md causalpy/skills/choosing-causalpy-methods/reference/disambiguation/sc_vs_sdid.md causalpy/skills/choosing-causalpy-methods/reference/disambiguation/ipw_vs_iv_vs_panel.md causalpy/skills/choosing-causalpy-methods/reference/disambiguation/prepostnegd_vs_did.md
  • prek run --all-files

Made with Cursor

drbenvincent and others added 2 commits May 25, 2026 17:43
Make method routing more reliable for agents by replacing flat fast routing with ordered intake, text-based routing, explicit no-fit outcomes, and focused disambiguation references.

Co-authored-by: Cursor <cursoragent@cursor.com>
Make the method-selection skill easier to discover for plain-language impact questions that imply quasi-experimental causal inference.

Co-authored-by: Cursor <cursoragent@cursor.com>
@read-the-docs-community
Copy link
Copy Markdown

read-the-docs-community Bot commented May 25, 2026

Documentation build overview

📚 causalpy | 🛠️ Build #32840606 | 📁 Comparing eafc26a against latest (c107d4e)

  🔍 Preview build  

1 file changed
± 404.html

@codecov
Copy link
Copy Markdown

codecov Bot commented May 25, 2026

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.08%. Comparing base (c107d4e) to head (eafc26a).

Additional details and impacted files
@@            Coverage Diff             @@
##             main     #931      +/-   ##
==========================================
+ Coverage   95.07%   95.08%   +0.01%     
==========================================
  Files          87       87              
  Lines       13701    13701              
  Branches      812      812              
==========================================
+ Hits        13026    13028       +2     
+ Misses        479      478       -1     
+ Partials      196      195       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

drbenvincent and others added 2 commits May 25, 2026 18:00
Route comparative interrupted time-series questions to InterruptedTimeSeries with comparison series as predictors, and clarify when to prefer synthetic control.

Co-authored-by: Cursor <cursoragent@cursor.com>
Keep CITS covered in routing without elevating it in the main skill introduction.

Co-authored-by: Cursor <cursoragent@cursor.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant