[Repo Assist] Monthly Activity 2026-06

[github-actions]

🤖 Repo Assist here — I'm an automated AI assistant for this repository.

Activity for June 2026

Suggested Actions for Maintainer

Comprehensive list of all pending actions requiring maintainer attention.

• Review PR [Repo Assist] fix: implement asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers #1432: RA fix — asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers (closes Add asymptotic confidence intervals for average treatment effect for linear regression with effect modifiers #336) — Review
• Review PR [Repo Assist] test(utils): add Encoders class tests and one_hot_encode edge cases #1580: RA test — adds 11 Encoders class tests + 3 one_hot_encode edge-case tests — Review
• Review PR [Repo Assist] perf(gcm): avoid pre-allocated DataFrame in compute_data_from_noise and compute_noise_from_data #1561: perf(gcm) — avoid pre-allocated DataFrame in _noise.py — Review
• Review PR [Repo Assist] improve(gcm): add random_seed parameter to interventional_samples, counterfactual_samples, and average_causal_effect #1549: improve(gcm) — add random_seed to interventional_samples etc. — Review
• Review PR [Repo Assist] fix: preserve mediation confounders and no_directed_path in IdentifiedEstimand.__deepcopy__ #1548: fix — preserve mediation_first_stage_confounders in IdentifiedEstimand.__deepcopy__ — Review
• Review PR [Repo Assist] fix(cit): correct conditional_MI entropy computation for discrete variables #1547: fix(cit) — correct conditional_MI entropy computation (consolidate with [Repo Assist] fix(cit): correct conditional_MI column indexing to use Series instead of DataFrame #1522) — Review
• Review PR [Repo Assist] perf(gcm): avoid redundant get_ordered_predecessors calls and pre-allocated DataFrame in fitting_sampling #1544: perf — avoid redundant get_ordered_predecessors calls — Review
• Review PR [Repo Assist] improve: replace generic Exception with specific exception types in remaining modules #1543: improve — replace generic Exception with specific types — Review
• Review PR [Repo Assist] feat(gcm): add optional nodes filter to anomaly_scores() #1540: feat — add optional nodes filter to gcm.anomaly_scores() — Review
• Review PR [Repo Assist] docs: improve refutation documentation with p-value interpretation guide #1535: docs — improve refutation docs with p-value interpretation table (closes Guide on refutations and how to interpret p-values #532) — Review
• Review PR [Repo Assist] feat(datasets): add return_all_coefficients parameter to linear_dataset #1530: feat — add return_all_coefficients to linear_dataset() (closes Causal Effect of Common Causes to Outcome #466) — Review
• Review PR [Repo Assist] fix: make refuter significance tests robust to array-valued estimates #1529: fix — perform_bootstrap_test robust to array-valued estimates — Review
• Review PR [Repo Assist] feat(gcm): expose plotting functions at dowhy.gcm level #1528: feat — expose plotting functions at dowhy.gcm top level (closes Move gcm plotting utility one level up #586) — Review
• Review PR feat: Add z-identifiability via surrogate experiments as identification method in identify_ate_effect #1531: contributor (Omar-Camara) — add z-identifiability via surrogate experiments — Review
• Review PR [Repo Assist] fix(cit): correct conditional_MI column indexing to use Series instead of DataFrame #1522: fix(cit) — correct conditional_MI column indexing (consolidate with [Repo Assist] fix(cit): correct conditional_MI entropy computation for discrete variables #1547) — Review
• Review PR chore: relax scipy upper bound constraint #1590: contributor — relax scipy upper bound constraint — Review
• Review PR fix: add descriptive error for out-of-bounds interventions in weighting_sampler (Closes #1344) #1589: contributor (TasfinMahmud) — add descriptive error for out-of-bounds interventions in WeightingSampler (closes Error in do() operator for interventions outside original treatment distribution #1344) — Review
• Review PR Fix ConfounderDistributionInterpreter crash on categorical confounders #1586: contributor (immu4989) — fix ConfounderDistributionInterpreter crash on categorical confounders (closes ConfounderDistributionInterpreter crashes on categorical confounders (Cannot cast str dtype to float64) #1588) — Review
• Review PR [Repo Assist] fix: raise ValueError when estimate_effect() called with method_name=None #1594: RA fix — raise ValueError in estimate_effect() when method_name=None (closes CausalModel.estimate_effect - UnboundLocalError: local variable 'identifier_name' referenced before assignment #365) — Review
• Review PR [Repo Assist] fix(graph_refuter): detect all integer/bool dtypes as discrete, not just int32/int64 #1595: RA fix — GraphRefuter dtype detection: handle int8, uint8, bool and all integer types — Review
• Review PR [Repo Assist] fix: route distance metric params through metric_params= in NearestNeighbors (fixes #1390) #1577: RA fix — route distance metric params through metric_params= in NearestNeighbors (conflicts resolved by Copilot) — Review
• Review PR (new RA PR, branch repo-assist/perf-select-model-shortcircuit-20260621): perf — skip encoder fitting in select_model() when X.shape[1] > 5 (one-hot encoding can only expand columns) — PR number pending push
• Merge PR build(deps): bump undici from 7.24.4 to 7.28.0 in /docs/version_patcher #1592: Dependabot security update — undici 7.28.0 (7 CVEs including 3 High) in docs/version_patcher — Review
• Close PR [Repo Assist] fix: preserve DataFrame index in one_hot_encode to fix DataSubsetRefuter with categorical columns #1533: superseded by merged contributor PR Fix data subset refuter index misalignment #1546; has merge conflict; contributor already noted this — Close
• Close PR [Repo Assist] fix: raise ValueError in estimate_effect() when identified estimand is None #1554: superseded by merged fix: raise ValueError when no valid estimand is identified instead of silent None return #1552 — Close
• Close PR [Repo Assist] fix: guard against KeyError in get_backdoor_variables when default_backdoor_id is None #1538: superseded by merged [Repo Assist] fix: update _target_estimand for pre-instantiated stage models in TwoStageRegressionEstimator #1495 — Close
• Close PR [Repo Assist] fix: suppress pandas ≥2.2 GroupBy.apply DeprecationWarning in _estimate_conditional_effects #1523: superseded by merged [Repo Assist] fix: suppress pandas ≥2.2 GroupBy.apply DeprecationWarning and remove dead code #1575 — Close
• Close PR [Repo Assist] fix: route distance metric-specific params through metric_params= in NearestNeighbors #1508: superseded by merged fix(estimator): pass distance metric parameters as metric_params to NearestNeighbors #1563 — Close
• Close issue numpy.dual is dropped but it still occurs in dowhy #1181: resolved — upgrading DoWhy fixes numpy.dual — View
• Close issue The direction of the graph's edges seems not right #859: not a bug — contributor confirmed — View
• Consolidate PRs for conditional_MI: [Repo Assist] fix(cit): correct conditional_MI column indexing to use Series instead of DataFrame #1522 and [Repo Assist] fix(cit): correct conditional_MI entropy computation for discrete variables #1547 fix same function — pick one

Future Work for Repo Assist

• Many RA PRs pending CI approval — maintainer needs to approve CI runs for bot PRs.
• Once stale PRs are closed, continue rebasing/refreshing older fixes.
• Engineering investments (dependency updates, CI improvements) blocked by protected .github/ files.

Run History

2026-06-21 01:40 UTC — Run

• 🔍 Task 6: Checked all open RA PRs — CI failures are infrastructure (disk space), not code failures; no code fixes needed
• 💬 Task 6: Commented on PR [Repo Assist] fix: preserve DataFrame index in one_hot_encode to fix DataSubsetRefuter with categorical columns #1533 — recommended closure (superseded by Fix data subset refuter index misalignment #1546, has merge conflict)
• 🔧 Task 8: Created perf PR (branch repo-assist/perf-select-model-shortcircuit-20260621) — skip expensive encoder fitting in select_model() when X.shape[1] > 5 (one-hot can only expand columns); also remove redundant is_categorical(Y) call. All 25 tests pass.

2026-06-20 13:15 UTC — Run

• 🔧 Task 3: Created PR [Repo Assist] fix(graph_refuter): detect all integer/bool dtypes as discrete, not just int32/int64 #1595 (branch repo-assist/fix-graph-refuter-dtype-detection-20260620) — fix GraphRefuter dtype detection: use pd.api.types.is_integer_dtype() to handle int8, uint8, bool, and all integer types; adds tests/causal_refuters/test_graph_refuter.py
• 🔍 Task 2: Investigated all open issues — all have existing RA comments; no new uncommented issues since last run
• 📝 Task 11: Updated monthly summary — added [Repo Assist] fix: raise ValueError when estimate_effect() called with method_name=None #1594, new graph_refuter PR, Dependabot security PR build(deps): bump undici from 7.24.4 to 7.28.0 in /docs/version_patcher #1592

2026-06-20 06:35 UTC — Run

• 🏷️ Labelled ConfounderDistributionInterpreter crashes on categorical confounders (Cannot cast str dtype to float64) #1588 with bug
• 💬 Task 2: Commented on ConfounderDistributionInterpreter crashes on categorical confounders (Cannot cast str dtype to float64) #1588 — acknowledged bug report and contributor PR Fix ConfounderDistributionInterpreter crash on categorical confounders #1586 fixing ConfounderDistributionInterpreter categorical crash
• 🔧 Task 3: Created PR [Repo Assist] fix: raise ValueError when estimate_effect() called with method_name=None #1594 (branch repo-assist/fix-estimate-effect-method-name-none-20260620) — raise ValueError in estimate_effect() when method_name=None; completes fix started by merged PR [Repo Assist] improve: raise ValueError with helpful message when method_name=None in do() and refute_estimate() #1537 (closes CausalModel.estimate_effect - UnboundLocalError: local variable 'identifier_name' referenced before assignment #365)
• 💬 Task 3: Commented on CausalModel.estimate_effect - UnboundLocalError: local variable 'identifier_name' referenced before assignment #365 — linked fix PR

2026-06-11 01:35 UTC — Run

• 💬 Task 2: Commented on Unhandled traceback in one_hot_encode when columns contain nested or unhashable data structures #1578 — acknowledged bug, noted PR fix: wrap one_hot_encode execution blocks to catch nested structure failures #1579 fix approach, suggested catching specific exceptions, flagged test overlap with RA [Repo Assist] test(utils): add Encoders class tests and one_hot_encode edge cases #1580
• 🔍 Task 6: Verified RA PRs [Repo Assist] fix: route distance metric params through metric_params= in NearestNeighbors (fixes #1390) #1577, [Repo Assist] test(utils): add Encoders class tests and one_hot_encode edge cases #1580, [Repo Assist] fix(utils): correct OrderedSet iteration for falsy elements + add tests #1581 created correctly; no merge conflicts on key branches; all awaiting maintainer CI approval

2026-06-10 13:57 UTC — Run

• 🔧 Task 5: Fixed OrderedSet.__next__ falsy-element bug (if not element → if element is None); added 19 tests in tests/utils/test_ordered_set.py; PR [Repo Assist] fix(utils): correct OrderedSet iteration for falsy elements + add tests #1581
• 🧪 Task 9: Added 11 Encoders class tests + 3 one_hot_encode edge-case tests to tests/utils/test_encoding.py; PR [Repo Assist] test(utils): add Encoders class tests and one_hot_encode edge cases #1580

2026-06-10 01:31 UTC — Run

• 🔧 Task 3/6: Created PR [Repo Assist] fix: route distance metric params through metric_params= in NearestNeighbors (fixes #1390) #1577 — rebase of PR [Repo Assist] fix: route distance metric-specific params through metric_params= in NearestNeighbors #1508 onto current main; adds _build_nearest_neighbors() to DistanceMatchingEstimator properly routing p and metric_params= (closes #1390, supersedes [Repo Assist] fix: route distance metric-specific params through metric_params= in NearestNeighbors #1508, compare with fix(estimator): pass distance metric parameters as metric_params to NearestNeighbors #1563)
• 💬 Task 6: Commented on PR [Repo Assist] fix: route distance metric-specific params through metric_params= in NearestNeighbors #1508 — noted merge conflict, pointed to fix(estimator): pass distance metric parameters as metric_params to NearestNeighbors #1563 and new rebase PR as alternatives
• 📝 Task 11: Updated monthly summary

2026-06-09 13:37 UTC — Run

• 🔧 Task 3: Fixed CausalModel.do() — 3 bugs fixed (method_params=None TypeError, wrong fit() args, missing self._data); PR [Repo Assist] fix: correct CausalModel.do() method_params and fit() call signature #1576; closes #309
• 💬 Task 3: Commented on effect_estimate different using do operator and _estimate_effect function #309 — linked fix PR
• 📝 Task 11: Updated monthly summary

2026-06-09 01:26 UTC — Run

• 🔧 Task 4: Created PR [Repo Assist] fix: suppress pandas ≥2.2 GroupBy.apply DeprecationWarning and remove dead code #1575 — suppress pandas ≥2.2 GroupBy.apply DeprecationWarning (supersedes [Repo Assist] fix: suppress pandas ≥2.2 GroupBy.apply DeprecationWarning in _estimate_conditional_effects #1523)
• 💬 Task 2: Commented on Include conditional effects, p-value and confidence intervals for DML #325 — explained DML confidence intervals and CATE inference

2026-06-08 14:27 UTC — Run

• 🔧 Task 3/6: Rebased PR [Repo Assist] fix: implement asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers #1432 with current main
• 💬 Task 2: Commented on contributor PR chore: improve dowhy maintenance path #1566

2026-06-08 01:35 UTC — Run

• 🔍 Task 6: Confirmed all CI checks on PR [Repo Assist] fix: update _target_estimand for pre-instantiated stage models in TwoStageRegressionEstimator #1495 passing ✅
• 💬 Task 6: Commented on PR [Repo Assist] fix: update _target_estimand for pre-instantiated stage models in TwoStageRegressionEstimator #1495 — ready to merge

2026-06-07 13:11 UTC — Run

• 🔍 Task 6: Investigated CI failure on PR [Repo Assist] fix: update _target_estimand for pre-instantiated stage models in TwoStageRegressionEstimator #1495 — pre-existing test bug, not caused by [Repo Assist] fix: update _target_estimand for pre-instantiated stage models in TwoStageRegressionEstimator #1495
• 💬 Task 2: Commented on Issue with distance_metric_params for dowhy.causal_estimators.distance_matching_estimator #1390 — informed reporter about contributor PR fix(estimator): pass distance metric parameters as metric_params to NearestNeighbors #1563

2026-06-07 01:34 UTC — Run

• 🔧 Task 3/6: Updated PR [Repo Assist] fix: clone pre-instantiated stage estimators with correct backdoor estimand in TwoStageRegressionEstimator #1527; commented on Issue with Logistic Regression in Mediation Analysis #1335

2026-06-06 13:07 UTC — Run

• 🔍 Task 6: Verified merge conflicts; confirmed merges [Repo Assist] fix: make CausalEstimator pickle/deepcopy-safe by excluding logger #1564, [Repo Assist] fix: PropensityBalanceInterpreter works with arbitrary covariate and treatment names #1550, [Repo Assist] fix(docs): add defaults to TAGS/CURRENT_VERSION env vars in conf.py #1558, [Repo Assist] fix(refuter): derive distinct per-simulation seeds in DataSubsetRefuter #1560, [Repo Assist] test: add tests for RandomCommonCause refuter #1413

2026-06-06 01:30 UTC — Run

• 🔧 Task 3/6: Updated PR [Repo Assist] fix: implement asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers #1432; commented on Add asymptotic confidence intervals for average treatment effect for linear regression with effect modifiers #336

2026-06-05 01:31 UTC — Run

• ⚙️ Task 6: Checked RA PR status; commented on fix(estimator): pass distance metric parameters as metric_params to NearestNeighbors #1563

2026-06-04 13:46 UTC — Run

• 🔍 Task 6: Reviewed all open RA PRs; CI status confirmed

2026-06-04 02:00 UTC — Run

• 🔍 Task 3: Investigated open bug issues; contributor PRs fix: raise ValueError in iv branch of TwoStageRegressionEstimator for missing instrumental variables #1520 and fix: raise ValueError when no valid estimand is identified instead of silent None return #1552 ready for merge

2026-06-03 01:41 UTC — Run

• 💬 Task 2: Commented on Support polars data frames #1151; Task 6: Updated PR [Repo Assist] fix: make CausalEstimator pickle/deepcopy-safe by excluding logger #1477

2026-06-02 14:06 UTC — Run

• 🔧 Task 8: Created perf PR [Repo Assist] perf(gcm): avoid pre-allocated DataFrame in compute_data_from_noise and compute_noise_from_data #1561 — avoid pre-allocated DataFrame in _noise.py

2026-06-02 01:36 UTC — Run

• 🔧 Task 3: Fixed DataSubsetRefuter random_state bug; 4 new tests

2026-06-01 01:51 UTC — Run

• 🔧 Task 3: Fixed docs/source/conf.py — TAGS/CURRENT_VERSION defaults (closes #692)
• 📝 Task 11: Created June 2026 issue; closed May 2026 issue [Repo Assist] Monthly Activity 2026-05 #1494

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@11c9a2c442e519ff2b427bf58679f5a525353f76

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@11c9a2c442e519ff2b427bf58679f5a525353f76

[Huaian666]

our whole infra runs on this now. migration was smooth. haven't had a single issue since switching. can't say that about the previous solution.