Agentic CI: Issue & PR Triage Tracker #562

Activity

1. andreatnvidia

   pinned this issue on Apr 20, 2026
2. github-actions commented on May 11, 2026

   

   github-actionsBot

   on May 11, 2026 – with GitHub Actions

   Last edited by github-actions

   Contributor

   More actions

   Repository Triage Report — 2026-11-09

   Open issues: 55 | Open PRs: 7 | Healthy (no action needed): 26

   ---

   Close as resolved (1)

   #	Title	Action	Evidence	Rationale
   #1103	Complete Slurm proxy health integration, draining, telemetry, and overload validation	Close	Closes #1103 in #1104 (merged 2026-10-30)	Merged PR "fix(slurm): complete proxy lifecycle and overload validation" matches the issue scope; it probably didn't auto-close because #1104 targeted a non-default branch.

   Close as duplicate (0)

   No exact duplicates found. Related but separate scopes: #92 / #1033 (Vertex AI provider support) and #695 / #727 / #741 (async run diagnostics and tuning); noted below where relevant.

   Needs maintainer decision (7)

   #	Title	Action	Evidence	Rationale
   #1129	Default OpenAI reasoning/vision model gpt-5 is retired on 2026-12-11	Pick the replacement default, then review #1141	needs-discussion; PR #1141 open, 0 failing checks	Deadline is 2026-12-11; #1141 proposes gpt-5.6-luna as the replacement.
   #1033	Responses API (/v1/responses) and Vertex AI: model access for a multi-turn benchmark	Decide scope and approach	needs-discussion, on-roadmap	Overlaps the older #92 on Vertex AI; decide whether to merge the two or split the Responses API work out.
   #711	Allow model providers to restrict workflows to approved models	Decide direction	needs-discussion, no activity since 2026-05-29	Policy/design question with no recorded decision.
   #677	Add first-class reranking model support	Decide direction	needs-discussion, no activity since 2026-05-19	New model type affects config and engine; needs a go/no-go.
   #354	Add other common params to ChatCompletionInferenceParams	Decide which params to add	needs-discussion, updated 2026-09-23	Open question about which params to expose versus leaving them to extra_body.
   #352	Ability to control order of multi-modal context along with prompts	Decide API shape	needs-discussion, updated 2026-10-30	Recent activity but no settled design.
   #182	Improve validation in structured generation columns	Decide direction or close	needs-discussion, no activity since 2026-01-30	May be partly covered by #1125 (native schema enforcement); decide whether to keep it.

   Ready for assignment (11)

   #	Title	Action	Evidence	Rationale
   #1102	Local seed path containing a single quote breaks the DuckDB query	Assign	bug, triaged, no assignee	Clear repro; small quoting/escaping fix.
   #1082	Preserve existing destination files when dataset exports fail	Assign	bug, no assignee, updated 2026-10-28	Well-defined behavior (write to a temp file, then move into place).
   #1056	refactor: finalize dataset directory before media initialization	Assign (a contributor proposed an approach on 2026-11-05)	task, detailed plan, no assignee	Implementation plan is in the issue; a contributor is waiting for a go-ahead.
   #1140	Validate US phone area-code candidates using sources with documented reuse terms	Assign	bug, follow-up to #1139, concrete repro ZIPs	Accepted regression from #1139; expected outcome is listed in detail.
   #1138	Normalize localized sex labels across managed person datasets	Label and assign	No labels, no assignee	Recent and scoped; needs triage labels.
   #1125	Add an opt-in native schema-enforcement mode to LLMStructuredColumnConfig	Assign	enhancement, no assignee	Opt-in flag with clear scope; related to #182.
   #1111	[Feature] Python 3.15 support: pyarrow cap and duckdb/jsonpath wheels	Label and assign	No labels, no assignee	Dependency-cap work, possibly blocked on upstream wheels.
   #1094	ci: standardize uv version selection across workflows	Assign to a maintainer	task, no assignee	Mechanical CI change; edits under .github/workflows/ need a maintainer.
   #1054	chore: remove nbconvert workaround after upstream nh3 release	Check for the nh3 release, then assign	task, dependencies	Small cleanup once upstream has released.
   #727	Public async runs do not expose capacity diagnostics needed to explain endpoint idle time	Assign	enhancement, updated 2026-10-28	Active; related to #695 and #741.
   #211	Batch-level error rates for early shutdown	Assign	bug, enhancement, updated 2026-10-30	Old issue with recent activity, still unowned.

   Stuck PR (0)

   All 7 open PRs have 0 failing checks and were updated within the last 7 days.

   Duplicate PRs (0)

   No two open PRs reference the same issue.

   Stale, consider closing (17)

   #	Title	Action	Evidence	Rationale
   #885	Add keyed execution for group-scoped consistent generation	Close with note or schedule	80d no activity, no assignee	No traction since it was filed.
   #860	Expose terminal failed-column and seed-row provenance in public results	Close with note or schedule	89d no activity, no assignee	No traction since it was filed.
   #837	chore: consolidate docs/ and fern/ into one documentation tree	Re-check relevance, then close or schedule	102d no activity, no assignee	Docs work since then may have changed the scope.
   #809	refactor: interface-layer OpenTelemetry runtime follow-ups deferred from #802	Close with note or schedule	124d no activity, no assignee	Deferred follow-up that nobody has picked up.
   #790	feat: generate an exact number of rows matching a declared criterion	Close with note or schedule	131d no activity, no assignee	Feature idea with no traction.
   #760	enforce from __future__ import annotations via ruff FA102 rule	Assign or close	143d no activity; FA102 still not in ruff select	Still valid and a one-line config change plus fixes; cheap to pick up.
   #758	Add native provider request support for non-chat plugin model calls	Close with note or schedule	145d no activity, no assignee	No traction.
   #695	Report per-column idle time in end-of-run metrics	Fold into #727 or close	172d no activity	Overlaps the more active #727 (capacity diagnostics).
   #694	Consolidate Data Designer user configuration into a single TOML file	Close with note or schedule	124d no activity, no assignee	Design-heavy with no follow-up.
   #659	uv tool update removes DataDesigner plugin installs	Re-check against current uv, then close	179d no activity, bug	uv behavior may have changed since it was filed.
   #637	Split docs preview build and deploy workflows	Close with note or schedule	180d no activity, no assignee	Workflow task with no traction.
   #619	Push Data Designer datasets to Harbor Hub	Close with note	186d no activity, no assignee	Integration request with no follow-up.
   #601	tests: cover structural_impact.py pure helpers; thread repo_root and align path resolution	Re-check relevance, then close	189d no activity, no assignee	Tooling code may have moved since it was filed.
   #573	Concrete schemas for all nemotron-personas datasets	Re-check against the persona locales added since, then close or schedule	199d no activity, no assignee	Persona datasets have been added since (#1061, #1136).
   #563	Skill: auto-port custom columns to plugin packages	Close with note	203d no activity, no assignee	No traction.
   #386	Add structured metadata to log records for programmatic consumption	Close with note or schedule	213d no activity, no assignee	Related to the logging hook in #503.
   #362	Allow user to keep record fields that failed to parse as null.	Close with note or schedule	245d no activity, no assignee	No traction; may relate to #182 / #1125.

   ---

   Healthy items (19 issues, 7 PRs)

   Issues

   • Annotated Parquet reads can corrupt nested Arrow arrays with null children #1143 Annotated Parquet reads can corrupt nested Arrow arrays with null children — lipikaramaswamy — 2026-11-09 (PR fix: preserve nested Arrow arrays on storage reads #1142)
   • Structured-output prompt schemas escape non-ASCII text as \uXXXX #1130 Structured-output prompt schemas escape non-ASCII text as \uXXXX — spped2000 — 2026-11-04 (PR fix: keep non-ASCII in prompt schemas #1131)
   • Support image columns and media collection in multi-shard Slurm runs #1114 Support image columns and media collection in multi-shard Slurm runs — nabinchha — 2026-10-29 (assigned)
   • Support plugin and custom columns in multi-shard Slurm runs #1113 Support plugin and custom columns in multi-shard Slurm runs — nabinchha — 2026-10-29 (assigned)
   • Allow huggingface-hub 2.x (widen cap from <2 to <3) #1110 Allow huggingface-hub 2.x — davanstrien — 2026-10-29 (PR chore: allow huggingface-hub 2.x #1117)
   • Jinja templates see string cells that look like JSON scalars as numbers, booleans and None #1088 Jinja templates see JSON-scalar-like string cells as numbers/booleans/None — v0ropaev — 2026-10-30 (PR fix: keep scalar strings as text in template contexts #1119)
   • Expose structured model cost/usage in DatasetCreationResults #956 Expose structured model cost/usage in DatasetCreationResults — jeremyjordan — 2026-10-06
   • Persist allocation-local performance telemetry for Slurm runs and benchmarks #947 Persist allocation-local performance telemetry for Slurm runs and benchmarks — andreatnvidia — 2026-09-17
   • Decouple HTTP connection pool limits from max_parallel_requests at high concurrency #946 Decouple HTTP connection pool limits from max_parallel_requests — andreatnvidia — 2026-09-22
   • Harden the Slurm runtime and validate sealed release artifacts #870 Harden the Slurm runtime and validate sealed release artifacts — nabinchha — 2026-08-15 (assigned)
   • Epic: Slurm batch execution v1 #850 Epic: Slurm batch execution v1 — andreatnvidia — 2026-10-28 (epic, tracked via feat(slurm): integrate Slurm execution for 1.0.0 release #1040)
   • Public async runs cannot tune row-group horizon/adaptive admission #741 Public async runs cannot tune row-group horizon/adaptive admission — eric-tramel — 2026-10-06
   • Agentic CI: Issue & PR Triage Tracker #562 Agentic CI: Issue & PR Triage Tracker — andreatnvidia — 2026-11-02 (tracking issue)
   • Add request/response logging hook to native model client #503 Add request/response logging hook to native model client — andreatnvidia — 2026-10-06
   • refactor: introduce AsyncRunController and explicit sinks for async build/preview #447 refactor: introduce AsyncRunController and explicit sinks — andreatnvidia — 2026-03-25 (assigned)
   • Enable non-existent tool fallback with MCP server #400 Enable non-existent tool fallback with MCP server — danecor — 2026-10-06
   • Update health-checks to account for multi-modality defined in the data designer workflow #246 Update health-checks to account for multi-modality — nabinchha — 2026-10-06
   • Add more stages to processors, and allow for custom processors via callbacks #159 Add more stages to processors, and custom processors via callbacks — andreatnvidia — 2026-03-02 (assigned)
   • Support more types of Model Providers like VertexAI and Bedrock #92 Support more types of Model Providers like VertexAI and Bedrock — kirit93 — 2026-09-29

   PRs

   • fix: preserve nested Arrow arrays on storage reads #1142 fix: preserve nested Arrow arrays on storage reads — lipikaramaswamy — 2026-11-09
   • fix: replace retired gpt-5 and gpt-4.1 with gpt-5.6-luna in OpenAI default models #1141 fix: replace retired gpt-5 and gpt-4.1 with gpt-5.6-luna in OpenAI default models — pcbeingused333 — 2026-11-08
   • fix: keep non-ASCII in prompt schemas #1131 fix: keep non-ASCII in prompt schemas — spped2000 — 2026-11-04
   • feat: add dtype_backend to result loading #1128 feat: add dtype_backend to result loading — binaryaaron — 2026-11-03
   • fix: keep scalar strings as text in template contexts #1119 fix: keep scalar strings as text in template contexts — v0ropaev — 2026-11-06
   • chore: allow huggingface-hub 2.x #1117 chore: allow huggingface-hub 2.x — davanstrien — 2026-11-03
   • feat(slurm): integrate Slurm execution for 1.0.0 release #1040 feat(slurm): integrate Slurm execution for 1.0.0 release (draft) — nabinchha — 2026-11-06

   Summary

   • 36 items flagged for action across 7 buckets (1 resolved, 0 duplicate, 7 decision, 11 ready, 17 stale)
   • 0 PRs flagged (0 stuck, 0 duplicate)
   • 26 healthy items (19 issues, 7 PRs; collapsed above)
   • Repeatedly-failed fix attempts: no daily-suite state files (.agentic-ci-state/) were available in this run, so this section was skipped.
3. NVIDIA-NeMo

   deleted a comment from github-actions on May 12, 2026