Skip to content

[Tracker] Pull request handoff SLA #4925

Open

Description

@github-actions

Pull request handoff SLA tracker

Review during the daily stand-up alongside the architecture/RFC decision tracker.

The SLA is one business day per handoff; weekends are excluded and public holidays are not modeled.
Drafts, bots, and PRs labeled stale, paused, or do-not-track are excluded.
Last refreshed: 2026-10-22T14:30+00:00

Newly breached (17)

Waiting on reviewers (8)

PR Waiting on Reason Handoff SLA deadline Overdue
#4774 — fix(enterpriseops): retain databases until cleanup succeeds @mcuevas-nvidia review requested 2026-10-20 16:04 UTC 2026-10-21 16:04 UTC 22h
#4798 — feat(checkpoint): resume ToolSimulationAgent legacy runs @ananthsub review requested 2026-10-20 19:56 UTC 2026-10-21 19:56 UTC 18h
#4799 — fix(opensandbox): delete the sandbox of a create the caller gave up on @bxyu-nvidia, @cmunley1, @hemildesai, @theothermike, @yaoyu-33 review requested 2026-10-20 20:04 UTC 2026-10-21 20:04 UTC 18h
#4805 — fix(agents): hide the sandbox supervisor and runners from pkill python @bxyu-nvidia, @cmunley1, @hemildesai, @theothermike review requested 2026-10-20 20:33 UTC 2026-10-21 20:33 UTC 17h
#4812 — fix(physics_judge): carry domain through verify so per-domain metrics compute @yaoyu-33 review requested 2026-10-20 22:27 UTC 2026-10-21 22:27 UTC 16h
#4837 — build(env): AUT-2574 lock managed environments and use uv sync @anwithk review requested 2026-10-21 13:14 UTC 2026-10-22 13:14 UTC 1h
#4792 — fix(genrm_compare): keep a completed cohort through a checkpoint @yuhezhang-ai review requested 2026-10-21 13:22 UTC 2026-10-22 13:22 UTC 1h
#4858 — refactor(config): remove pre-v0.6.0 renamed config aliases @cmunley1, @marta-sd review requested 2026-10-21 14:17 UTC 2026-10-22 14:17 UTC 0h

Needs review assignment (9)

PR Waiting on Reason Handoff SLA deadline Overdue
#4791 — ci(autobot): enable comment reviews on ready transitions Maintainers reviewer not assigned 2026-10-20 16:51 UTC 2026-10-21 16:51 UTC 21h
#4765 — feat(equivalence-judge): support per-problem question extraction Maintainers reviewer not assigned 2026-10-20 18:43 UTC 2026-10-21 18:43 UTC 19h
#4637 — feat(sandbox): add cluster attribution label, configurable with a Slurm default Maintainers reviewer not assigned 2026-10-20 20:03 UTC 2026-10-21 20:03 UTC 18h
#4811 — feat(eval): add opt-in binary Bayes@N reward profiles Maintainers reviewer not assigned 2026-10-20 21:47 UTC 2026-10-21 21:47 UTC 16h
#4814 — fix: resolve issue #4810 - feat(eval): Add opt-in Bayes@N summaries to reward profiling Maintainers reviewer not assigned 2026-10-20 23:45 UTC 2026-10-21 23:45 UTC 14h
#4822 — fix(sandbox): reconcile Docker creates and resolve E2B service endpoints Maintainers reviewer not assigned 2026-10-21 03:53 UTC 2026-10-22 03:53 UTC 10h
#4823 — fix(core): close shared aiohttp clients on their owner loop Maintainers reviewer not assigned 2026-10-21 03:53 UTC 2026-10-22 03:53 UTC 10h
#4826 — docs(harbor): deprecate harbor_agent and harbor_agent_general in favour of the Harbor task format Maintainers reviewer not assigned 2026-10-21 04:55 UTC 2026-10-22 04:55 UTC 9h
#4827 — feat(harbor): pull datasets into environments/harbor/ (M38, first slice) Maintainers reviewer not assigned 2026-10-21 05:12 UTC 2026-10-22 05:12 UTC 9h
Older breach backlog (236)

Waiting on reviewers (99)

PR Waiting on Reason Handoff SLA deadline Overdue
#799 — fix: preserve optional fields in structured_outputs strictify_schema_json @jkyi-nvidia review requested 2026-03-02 20:59 UTC 2026-03-03 20:59 UTC 166d 17h
#879 — Yev/scicode env v1 @dhruvnathawani, @jubick1337 review requested 2026-03-20 18:09 UTC 2026-03-23 18:09 UTC 152d 20h
#946 — Turing/ Playwright-based CUA environment with multi-provider adapter support @bxyu-nvidia review requested 2026-04-29 19:09 UTC 2026-04-30 19:09 UTC 124d 19h
#1193 — docs: address repo-health audit nits in README and CONTRIBUTING @cmunley1 review requested 2026-04-30 13:30 UTC 2026-05-01 13:30 UTC 124d 0h
#1264 — feat: artifact-root resolution for external plugins @ananthsub, @bxyu-nvidia review requested 2026-05-07 20:20 UTC 2026-05-08 20:20 UTC 118d 18h
#1334 — structured_outputs: add LLM-as-a-judge semantic verification @jkyi-nvidia review requested 2026-05-18 20:37 UTC 2026-05-19 20:37 UTC 111d 17h
#1491 — Add compute_eval benchmark and resource server @ananthsub, @bxyu-nvidia review requested 2026-06-02 00:07 UTC 2026-06-03 00:07 UTC 101d 14h
#1560 — docs: clarify resources server example-data contract and license validation @cwing-nvidia, @lbliii review requested 2026-06-10 12:04 UTC 2026-06-11 12:04 UTC 95d 2h
#1541 — Add SC_Bench benchmark @adil-a, @ananthsub, @bxyu-nvidia, @cwing-nvidia review requested 2026-06-13 00:57 UTC 2026-06-15 00:57 UTC 93d 13h
#1598 — Eshachar/openclaw swe @sdevare-nv review requested 2026-06-16 18:19 UTC 2026-06-17 18:19 UTC 90d 20h
#1653 — docs: remove model recipe column from RL framework compatibility table @cwing-nvidia review requested 2026-06-23 03:23 UTC 2026-06-24 03:23 UTC 86d 11h
#1752 — fix: handle vllm context length errors in nano v3 recipe @yfw review requested 2026-06-26 05:53 UTC 2026-06-29 05:53 UTC 83d 8h
#1964 — feat(model-server): add OpenAI-compatible /v1/embeddings endpoint @ffrujeri review requested 2026-07-15 16:18 UTC 2026-07-16 16:18 UTC 69d 22h
#2048 — feat: sandbox_agent swe resources server @adil-a, @ananthsub, @bxyu-nvidia, @hemildesai review requested 2026-07-15 21:13 UTC 2026-07-16 21:13 UTC 69d 17h
#2052 — hermes_agent: repin to upstream hermes and fix the wrapper for its API @Glorf review requested 2026-07-17 22:38 UTC 2026-07-20 22:38 UTC 67d 15h
#2155 — [GDPVal] Fix Stirrup reference paths and make head-server wait configurable @agronskiy review requested 2026-07-29 16:40 UTC 2026-07-30 16:40 UTC 59d 21h
#2160 — [GDPVal] Fix what reaches the rubric judge, and stop silent zeros @agronskiy review requested 2026-07-29 16:40 UTC 2026-07-30 16:40 UTC 59d 21h
#2162 — [GDPVal] Add model-agnostic Slurm orchestration for multi-allocation rollouts @agronskiy review requested 2026-07-29 16:40 UTC 2026-07-30 16:40 UTC 59d 21h
#2304 — default port range to the 5000-5999 band @adil-a review requested 2026-08-04 13:48 UTC 2026-08-05 13:48 UTC 56d 0h
#2311 — docs: rewrite Configure Agents index and document external agent harnesses @cwing-nvidia, @lbliii review requested 2026-08-04 14:30 UTC 2026-08-05 14:30 UTC 56d 0h
#2386 — feat: harden SWE against filesystem overflow @sdevare-nv review requested 2026-08-06 17:31 UTC 2026-08-07 17:31 UTC 53d 20h
#2142 — Add EnterpriseOps-Gym benchmark: resources server, benchmark registration, and per-turn telemetry agent @bxyu-nvidia, @cwing-nvidia review requested 2026-08-08 08:52 UTC 2026-08-10 08:52 UTC 53d 5h
#2481 — feat: swe_trace_converter change for the opencode backend @bxyu-nvidia, @sdevare-nv review requested 2026-08-11 18:44 UTC 2026-08-12 18:44 UTC 50d 19h
#2346 — feat(sandbox): add Tenki provider @hemildesai review requested 2026-08-12 16:45 UTC 2026-08-13 16:45 UTC 49d 21h
#2592 — Fix compute_pass_majority_metrics: score unanswered rollouts as failures in pass@k @gwarmstrong, @jkyi-nvidia review requested 2026-08-18 16:55 UTC 2026-08-19 16:55 UTC 45d 21h
#2703 — docs(verifier): require verify to return the scored response @mpatel31415 review requested 2026-08-24 05:09 UTC 2026-08-25 05:09 UTC 42d 9h
#2673 — fix: preserve structured MCP tool identity for verification @adil-a, @bxyu-nvidia review requested 2026-08-25 00:10 UTC 2026-08-26 00:10 UTC 41d 14h
#2742 — chore(hermes_agent): use pinned upstream Hermes release @Glorf, @ananthsub, @laszkiewiczp review requested 2026-08-25 03:18 UTC 2026-08-26 03:18 UTC 41d 11h
#2733 — feat(openclaw_agent): consume rollout-scoped Gym MCP metadata @cmunley1, @ffrujeri review requested 2026-08-25 04:28 UTC 2026-08-26 04:28 UTC 41d 10h
#2771 — feat: add raw_rewards_and_lengths to group-level reward profile metrics @bxyu-nvidia, @jkyi-nvidia review requested 2026-08-25 21:53 UTC 2026-08-26 21:53 UTC 40d 16h
#2783 — refactor: decouple rollout identity from capture @Glorf, @ffrujeri, @pthombre review requested 2026-08-26 08:14 UTC 2026-08-27 08:14 UTC 40d 6h
#2758 — feat: sandbox provider image prepare hook + update prepare step in AnyTerminal @hemildesai review requested 2026-08-26 16:17 UTC 2026-08-27 16:17 UTC 39d 22h
#2732 — chore(data): apply the dataset decoupling migration (agent_ref out of data files) @ananthsub review requested 2026-08-26 23:21 UTC 2026-08-27 23:21 UTC 39d 15h
#2879 — feat(sandbox): IdeGYM sandbox provider @hemildesai review requested 2026-08-31 23:09 UTC 2026-09-01 23:09 UTC 36d 15h
#3050 — fix(token-id-capture): fingerprint the served response dialect @pthombre review requested 2026-09-03 08:30 UTC 2026-09-04 08:30 UTC 34d 6h
#3058 — fix(code-gen): replace Manager proxies with bounded IPC @fgalko-oss review requested 2026-09-03 19:21 UTC 2026-09-04 19:21 UTC 33d 19h
#3117 — feat(server_utils): opt-in CRLF heartbeat so long HTTP/1.1 model calls survive Global Accelerator's 340s idle timeout @ananthsub, @kajalj22 review requested 2026-09-08 04:58 UTC 2026-09-09 04:58 UTC 31d 9h
#3133 — docs: standardize verification task data convention @adil-a review requested 2026-09-09 12:16 UTC 2026-09-10 12:16 UTC 30d 2h
#3189 — fix(responses): preserve refusal conversion semantics @bxyu-nvidia, @cmunley1 review requested 2026-09-09 14:04 UTC 2026-09-10 14:04 UTC 30d 0h
#2954 — docs: document gym env prefetch CLI #2937 @ritaneves review requested 2026-09-09 16:54 UTC 2026-09-10 16:54 UTC 29d 21h
#2580 — docs: fix Fern polish issues from NVBugs 6570461 @cwing-nvidia, @sephmard review requested 2026-09-09 16:55 UTC 2026-09-10 16:55 UTC 29d 21h
#3317 — fix(capture): keep unknown usage counts for streamed /v1/responses calls @cmunley1, @yaoyu-33 review requested 2026-09-15 03:38 UTC 2026-09-16 03:38 UTC 26d 10h
#3212 — feat(capture): expose complete rollout training traces @ananthsub, @bxyu-nvidia review requested 2026-09-15 13:07 UTC 2026-09-16 13:07 UTC 26d 1h
#3611 — feat(combibench): add CombiBench Lean 4 combinatorics benchmark @lilithgrigoryan review requested 2026-09-22 12:01 UTC 2026-09-23 12:01 UTC 21d 2h
#3348 — feat(translation_with_judge): add Indic-focused LLM-judge translation… @yaoyu-33 review requested 2026-09-22 12:03 UTC 2026-09-23 12:03 UTC 21d 2h
#3612 — feat(nmrarena): add NMRArena NMR structure elucidation benchmark @lilithgrigoryan review requested 2026-09-22 12:36 UTC 2026-09-23 12:36 UTC 21d 1h
#3599 — feat(metrics): collect optional vLLM and router inference metrics @oyilmaz-nvidia review requested 2026-09-22 12:51 UTC 2026-09-23 12:51 UTC 21d 1h
#3664 — fix(switchyard_model): match vllm_model for reasoning models; pin nemo-switchyard 0.3.0 @ananthsub, @bxyu-nvidia review requested 2026-09-25 00:04 UTC 2026-09-28 00:04 UTC 18d 14h
#3765 — fix(osworld): complete native AgentEnv evaluation and health reporting @JeffPengCoder review requested 2026-09-29 12:33 UTC 2026-09-30 12:33 UTC 16d 1h
#3780 — Enable multinode K8 @prokotg review requested 2026-09-29 19:07 UTC 2026-09-30 19:07 UTC 15d 19h
#3893 — feat(checkpoint): checkpoint evaluation runs from rollout collection @pthombre, @yuhezhang-ai, @zyzhou5 review requested 2026-10-01 18:57 UTC 2026-10-02 18:57 UTC 13d 19h
#3909 — feat(checkpoint): resources, agent, and environment servers with several workers @pthombre, @zyzhou5 review requested 2026-10-01 23:03 UTC 2026-10-02 23:03 UTC 13d 15h
#4109 — fix(opensandbox): retry a timed-out sandbox reconnect @hemildesai, @theothermike review requested 2026-10-06 16:49 UTC 2026-10-07 16:49 UTC 10d 21h
#4150 — feat(gdpval): add media preparation for judge inputs @agronskiy, @marta-sd review requested 2026-10-07 00:03 UTC 2026-10-08 00:03 UTC 10d 14h
#4154 — feat(agent): honor skip_verification in six agents that ignored the global flag @bxyu-nvidia, @yfw review requested 2026-10-07 05:49 UTC 2026-10-08 05:49 UTC 10d 8h
#4153 — fix(token-id-capture): split instructions out of the request envelope @ananthsub review requested 2026-10-07 07:00 UTC 2026-10-08 07:00 UTC 10d 7h
#4201 — feat(capture): commit the vLLM turn-boundary route in call extras @pthombre review requested 2026-10-08 00:03 UTC 2026-10-09 00:03 UTC 9d 14h
#4005 — feat(sandbox): add apptainer create.private_tmp option @ananthsub review requested 2026-10-08 00:30 UTC 2026-10-09 00:30 UTC 9d 13h
#4006 — feat(sandbox): fall back to a writable APPTAINER_CONFIGDIR @ananthsub review requested 2026-10-08 00:30 UTC 2026-10-09 00:30 UTC 9d 13h
#4034 — fix(swe_agents): mask rollouts whose OpenHands runtime server died @ananthsub review requested 2026-10-08 00:31 UTC 2026-10-09 00:31 UTC 9d 13h
#4218 — feat(token-capture): low-overhead metrics for capture operations, outcomes, digests, and caches @macandro96 review requested 2026-10-08 16:26 UTC 2026-10-09 16:26 UTC 8d 22h
#4237 — feat(sandbox): add Modal provider @hemildesai review requested 2026-10-08 17:16 UTC 2026-10-09 17:16 UTC 8d 21h
#3937 — fix(token-capture): use striped lock files that are never deleted @macandro96, @pthombre, @zyzhou5 review requested 2026-10-08 21:35 UTC 2026-10-09 21:35 UTC 8d 16h
#4268 — fix(token-capture): bound the capture-ledger cache @macandro96, @pthombre, @zyzhou5 review requested 2026-10-08 21:35 UTC 2026-10-09 21:35 UTC 8d 16h
#4272 — build(deps): bump nemo-lens to v0.3.0 to match NeMo-RL @oyilmaz-nvidia review requested 2026-10-09 00:24 UTC 2026-10-12 00:24 UTC 8d 14h
#4277 — feat(swebench): seed typed resources sessions for environment server episodes @hemildesai, @yaoyu-33 review requested 2026-10-09 05:57 UTC 2026-10-12 05:57 UTC 8d 8h
#4278 — feat(agent): return token capture produced outside the model server from an agent session @hemildesai, @yaoyu-33 review requested 2026-10-09 05:59 UTC 2026-10-12 05:59 UTC 8d 8h
#4302 — feat(swe): serve typed resources sessions from SWE-Gym and R2E-Gym @hemildesai, @sdevare-nv review requested 2026-10-09 16:49 UTC 2026-10-12 16:49 UTC 7d 21h
#4303 — feat(swe): serve typed resources sessions from SWEmer Ultra, SWEmer OML, and SWE internal v1 @hemildesai, @sdevare-nv, @yaoyu-33 review requested 2026-10-09 16:50 UTC 2026-10-12 16:50 UTC 7d 21h
#4304 — feat(swe): serve typed resources sessions from the SWE training servers @hemildesai, @sdevare-nv, @yaoyu-33 review requested 2026-10-09 16:50 UTC 2026-10-12 16:50 UTC 7d 21h
#4321 — Add variable resolution during gym eval @bxyu-nvidia review requested 2026-10-09 19:31 UTC 2026-10-12 19:31 UTC 7d 18h
#4259 — claude code token capture @pthombre review requested 2026-10-09 19:40 UTC 2026-10-12 19:40 UTC 7d 18h
#3850 — feat(nooa): add the sandbox process entrypoint @hemildesai review requested 2026-10-12 04:37 UTC 2026-10-13 04:37 UTC 7d 9h
#3851 — feat(nooa): supervise the borrowed sandbox process @hemildesai review requested 2026-10-12 04:37 UTC 2026-10-13 04:37 UTC 7d 9h
#3849 — feat(nooa): wire sandbox execution into agent sessions @hemildesai review requested 2026-10-12 04:37 UTC 2026-10-13 04:37 UTC 7d 9h
#2890 — perf(server): replace HTTP middleware wrappers with ASGI @bxyu-nvidia review requested 2026-10-12 14:37 UTC 2026-10-13 14:37 UTC 6d 23h
#4480 — [ApexAgents] Resume interrupted rollouts in prebuilt worlds @Kh4L review requested 2026-10-12 16:06 UTC 2026-10-13 16:06 UTC 6d 22h
#4358 — feat(big_finance): add BigFinanceBench environment @ushnish-de review requested 2026-10-12 16:45 UTC 2026-10-13 16:45 UTC 6d 21h
#3302 — feat(eval): persist failures and resume unfinished evaluations @giuliolovisotto review requested 2026-10-12 17:55 UTC 2026-10-13 17:55 UTC 6d 20h
#4220 — feat(sandbox): sandbox and snapshot lifecycle for partial-rollout checkpoints @ananthsub, @hemildesai review requested 2026-10-13 16:40 UTC 2026-10-14 16:40 UTC 5d 21h
#4612 — fix(telemetry): record metrics whenever metrics export, independent of span groups @iguehring-nvidia, @oyilmaz-nvidia review requested 2026-10-14 19:41 UTC 2026-10-15 19:41 UTC 4d 18h
#3980 — feat(token-capture): resolve a refused call's capture intent and report the refusal @ananthsub, @linj-glitch, @pthombre, @yaoyu-33 review requested 2026-10-14 23:37 UTC 2026-10-15 23:37 UTC 4d 14h
#4222 — feat(checkpoint): ask a long-running agent harness to park when a checkpoint opens @ananthsub review requested 2026-10-15 02:39 UTC 2026-10-16 02:39 UTC 4d 11h
#4221 — feat(checkpoint): process-level evidence for resources-owned sandbox checkpoints, their release, and lease expiry @ananthsub review requested 2026-10-15 02:40 UTC 2026-10-16 02:40 UTC 4d 11h
#3644 — feat(indic mbpp): add Indic MBPP @anwithk review requested 2026-10-15 14:22 UTC 2026-10-16 14:22 UTC 4d 0h
#4579 — docs: make r0.7.0 training commands copy-pasteable and bump version examples to v0.7.0 @yaoyu-33 review requested 2026-10-15 17:14 UTC 2026-10-16 17:14 UTC 3d 21h
#4638 — perf(rollout): validate the token-capture config once per run @marta-sd, @pthombre review requested 2026-10-15 17:34 UTC 2026-10-16 17:34 UTC 3d 20h
#4654 — fix(agents): disable Cline auto-update so the pinned CLI holds @yaoyu-33 review requested 2026-10-15 21:49 UTC 2026-10-16 21:49 UTC 3d 16h
#4693 — fix(google_search): send API key in a header and keep request details out of tool output @anwithk, @bxyu-nvidia review requested 2026-10-16 18:07 UTC 2026-10-19 18:07 UTC 2d 20h
#4727 — feat(eval): add the HLE-Diamond benchmark with text and sandboxed search presets @Kipok review requested 2026-10-19 05:04 UTC 2026-10-20 05:04 UTC 2d 9h
#4753 — fix(agent): sum token usage across every model call in four loop agents @yaoyu-33 review requested 2026-10-19 18:25 UTC 2026-10-20 18:25 UTC 1d 20h
#3910 — docs(nemo-user-sim): add multi-turn evaluation guides @ananthsub, @cwing-nvidia, @maygolk review requested 2026-10-19 18:46 UTC 2026-10-20 18:46 UTC 1d 19h
#4567 — fix(mcp): support MCP 1 and 2 tool exposure @ananthsub review requested 2026-10-19 19:13 UTC 2026-10-20 19:13 UTC 1d 19h
#4576 — feat(sandbox): sandbox owner base for resources servers, with SWE-bench on it @ananthsub review requested 2026-10-19 19:52 UTC 2026-10-20 19:52 UTC 1d 18h
#4274 — feat(checkpoint): harness driver for sandboxed agents that pauses them at checkpoints, with OpenCode on it @ananthsub review requested 2026-10-19 19:53 UTC 2026-10-20 19:53 UTC 1d 18h
#4761 — fix(agent): report summed policy usage instead of null or zeroed details @yaoyu-33 review requested 2026-10-19 20:47 UTC 2026-10-20 20:47 UTC 1d 17h
#4770 — fix(ccc): explain sandbox failures, fix the IOI run path, and mask unmeasured samples @cmunley1, @yaoyu-33 review requested 2026-10-20 00:53 UTC 2026-10-21 00:53 UTC 1d 13h
#4773 — test(checkpoint): add a checkpoint test agent for deterministic turn recovery @ananthsub review requested 2026-10-20 04:12 UTC 2026-10-21 04:12 UTC 1d 10h
#4284 — fix(opencode): optionally pass the prompt to opencode run verbatim @cmunley1 review requested 2026-10-20 06:03 UTC 2026-10-21 06:03 UTC 1d 8h

Waiting on authors (5)

PR Waiting on Reason Handoff SLA deadline Overdue
#1600 — docs(fern): add SEO meta descriptions + headlines to nightly docs (#1589) @lbliii changes requested 2026-06-29 14:05 UTC 2026-06-30 14:05 UTC 82d 0h
#2339 — Katherine/gdpval devel staging @katherineh123 changes requested 2026-08-09 22:03 UTC 2026-08-10 22:03 UTC 52d 16h
#2570 — feat(vllm_model): optional session-affinity header for upstream routers @aoshen02 changes requested 2026-08-18 16:53 UTC 2026-08-19 16:53 UTC 45d 21h
#3589 — feat(milu): add Indic benchmark integration @anushaknvidia changes requested 2026-10-05 17:47 UTC 2026-10-06 17:47 UTC 11d 20h
#3870 — feat(orchestration): give services on a shared node their own GPUs @wprazuch changes requested 2026-10-07 09:13 UTC 2026-10-08 09:13 UTC 10d 5h

Needs review assignment (132)

PR Waiting on Reason Handoff SLA deadline Overdue
#834 — Add Tau^2 Bench training environment Maintainers reviewer not assigned 2026-03-06 22:23 UTC 2026-03-09 22:23 UTC 162d 16h
#1103 — structured_outputs: support oneOf/anyOf branches and $defs/$ref schemas Maintainers reviewer not assigned 2026-04-21 05:08 UTC 2026-04-22 05:08 UTC 131d 9h
#1233 — structured_outputs: extend XML type coercion to handle oneOf/anyOf/allOf composition Maintainers reviewer not assigned 2026-05-05 21:17 UTC 2026-05-06 21:17 UTC 120d 17h
#1315 — Add GitHub issue template for environment integrations Maintainers reviewer not assigned 2026-05-13 19:42 UTC 2026-05-14 19:42 UTC 114d 18h
#1429 — docs: clarify SWE-bench dataset downloads Maintainers reviewer not assigned 2026-05-27 08:53 UTC 2026-05-28 08:53 UTC 105d 5h
#1590 — feat(equivalence_llm_judge): cap extracted-answer length sent to judge Maintainers reviewer not assigned 2026-06-12 22:06 UTC 2026-06-15 22:06 UTC 92d 16h
#1720 — feat(instruction_following): add loose-mode evaluation and compute_metrics Maintainers reviewer not assigned 2026-06-25 01:12 UTC 2026-06-26 01:12 UTC 84d 13h
#1745 — fix(config): seed no-model refs for prepare-data Maintainers reviewer not assigned 2026-06-26 00:31 UTC 2026-06-29 00:31 UTC 83d 13h
#1797 — fix(inference_provider): normalize provider endpoint errors Maintainers reviewer not assigned 2026-06-27 04:43 UTC 2026-06-29 04:43 UTC 83d 9h
#1800 — Harden SWE agent test file parsing Maintainers reviewer not assigned 2026-06-27 10:32 UTC 2026-06-29 10:32 UTC 83d 3h
#1847 — Rgala/browsecomp search providers newstable Maintainers reviewer not assigned 2026-07-01 00:35 UTC 2026-07-02 00:35 UTC 80d 13h
#1908 — Actions: sync project status to release label Maintainers reviewer not assigned 2026-07-02 13:16 UTC 2026-07-03 13:16 UTC 79d 1h
#1939 — Tb e2e Maintainers reviewer not assigned 2026-07-06 18:22 UTC 2026-07-07 18:22 UTC 76d 20h
#2080 — feat: Switchyard OpenCode integration Maintainers reviewer not assigned 2026-07-20 12:50 UTC 2026-07-21 12:50 UTC 67d 1h
#2109 — fix(server_utils): use unpredictable per-process secret for session signing Maintainers reviewer not assigned 2026-07-22 21:53 UTC 2026-07-23 21:53 UTC 64d 16h
#2027 — Simplify ModelCallRecord; Refactor call capture logic; Extend to agents Maintainers reviewer not assigned 2026-07-28 06:26 UTC 2026-07-29 06:26 UTC 61d 8h
#2254 — fix(hermes_agent): never present fabricated tokens as policy output Maintainers reviewer not assigned 2026-07-30 21:27 UTC 2026-07-31 21:27 UTC 58d 17h
#2437 — feat(agents): add aggregate metrics proxy helper Maintainers reviewer not assigned 2026-08-08 12:41 UTC 2026-08-10 12:41 UTC 53d 1h
#2412 — fix(pinchbench): address the model server per rollout so capture attributes calls Maintainers reviewer not assigned 2026-08-07 13:59 UTC 2026-08-10 13:59 UTC 53d 0h
#2513 — fix: serialize Jest in nv-internal SWE Maintainers reviewer not assigned 2026-08-12 13:56 UTC 2026-08-13 13:56 UTC 50d 0h
#2527 — fix: bound model request retries Maintainers reviewer not assigned 2026-08-13 09:39 UTC 2026-08-14 09:39 UTC 49d 4h
#2553 — feat: add OrcaRouter as a named inference provider Maintainers reviewer not assigned 2026-08-14 03:26 UTC 2026-08-17 03:26 UTC 48d 11h
#2578 — [Super 3.5 evals] Apply measured full-rack vLLM configuration Maintainers reviewer not assigned 2026-08-17 18:10 UTC 2026-08-18 18:10 UTC 46d 20h
#2594 — feat(swebench): configurable image namespace and arch for eval containers Maintainers reviewer not assigned 2026-08-18 14:50 UTC 2026-08-19 14:50 UTC 45d 23h
#2602 — docs: mention Cline harness in SWE agents README Maintainers reviewer not assigned 2026-08-18 20:29 UTC 2026-08-19 20:29 UTC 45d 18h
#2605 — Add live web client for Kuhn poker Maintainers reviewer not assigned 2026-08-19 01:15 UTC 2026-08-20 01:15 UTC 45d 13h
#2700 — feat(gymv): add multimodal environment integration Maintainers reviewer not assigned 2026-08-24 04:01 UTC 2026-08-25 04:01 UTC 42d 10h
#2772 — fix(sandbox): implement canonical task environment resolution Maintainers reviewer not assigned 2026-08-25 22:45 UTC 2026-08-26 22:45 UTC 40d 15h
#2841 — feat(primevul): add PrimeVul paired vulnerability-detection benchmark Maintainers reviewer not assigned 2026-08-28 08:45 UTC 2026-08-31 08:45 UTC 38d 5h
#2836 — perf(vllm): derive required prefix token IDs in Gym Maintainers reviewer not assigned 2026-08-31 17:20 UTC 2026-09-01 17:20 UTC 36d 21h
#2910 — keep resolved config out of process arguments Maintainers reviewer not assigned 2026-09-01 03:16 UTC 2026-09-02 03:16 UTC 36d 11h
#2915 — test(ci): AUT-2197 catch untestable component directories Maintainers reviewer not assigned 2026-09-01 06:51 UTC 2026-09-02 06:51 UTC 36d 7h
#2923 — Merge main into gdpval-devel Maintainers reviewer not assigned 2026-09-01 11:42 UTC 2026-09-02 11:42 UTC 36d 2h
#2978 — fix(osworld): stabilize long-horizon Molt rollouts Maintainers reviewer not assigned 2026-09-02 15:38 UTC 2026-09-03 15:38 UTC 34d 22h
#2980 — feat(anyterminal): OpenSandbox oracle config, resource knobs, cgroup metrics, TB2.1 prepare Maintainers reviewer not assigned 2026-09-02 17:43 UTC 2026-09-03 17:43 UTC 34d 20h
#2932 — [GDPVal] Merge main into gdpval-devel Maintainers reviewer not assigned 2026-09-03 17:12 UTC 2026-09-04 17:12 UTC 33d 21h
#3123 — docs(skills): align repository skills with current contracts Maintainers reviewer not assigned 2026-09-05 05:26 UTC 2026-09-07 05:26 UTC 33d 9h
#3085 — feat(nooa-agent): add native lifecycle integration and capability example Maintainers reviewer not assigned 2026-09-04 12:06 UTC 2026-09-07 12:06 UTC 33d 2h
#3096 — feat: add SAV tracking verifier resources server Maintainers reviewer not assigned 2026-09-04 18:27 UTC 2026-09-07 18:27 UTC 32d 20h
#3132 — [ApexAgents] Serve GLM-5.2 with CUDA graphs, pin replicas to one rack, tolerate partial readiness Maintainers reviewer not assigned 2026-09-05 21:55 UTC 2026-09-07 21:55 UTC 32d 16h
#3109 — feat(aviary): support post-terminal model response Maintainers reviewer not assigned 2026-09-04 22:04 UTC 2026-09-07 22:04 UTC 32d 16h
#3209 — feat(benchmarks): remove the dataset-level agent pin Maintainers reviewer not assigned 2026-09-10 05:34 UTC 2026-09-11 05:34 UTC 29d 8h
#3318 — Jnolan/gdpval hermes upstream Maintainers reviewer not assigned 2026-09-11 08:11 UTC 2026-09-14 08:11 UTC 28d 6h
#3372 — feat(cli): add gym eval intake to submit a rundir to a benchmark-run data layer Maintainers reviewer not assigned 2026-09-14 17:42 UTC 2026-09-15 17:42 UTC 26d 20h
#3412 — feat(checkpoint): prefix token level recovery Maintainers reviewer not assigned 2026-09-15 18:17 UTC 2026-09-16 18:17 UTC 25d 20h
#3494 — feat: post-train environment Maintainers reviewer not assigned 2026-09-17 16:55 UTC 2026-09-18 16:55 UTC 23d 21h
#3502 — feat: Allow explicit input files in eval runs Maintainers reviewer not assigned 2026-09-17 20:45 UTC 2026-09-18 20:45 UTC 23d 17h
#3503 — feat(browsecomp): add Brave search provider Maintainers reviewer not assigned 2026-09-17 21:32 UTC 2026-09-18 21:32 UTC 23d 16h
#3504 — feat(video): add SA-V tracks verifier Maintainers reviewer not assigned 2026-09-17 21:52 UTC 2026-09-18 21:52 UTC 23d 16h
#3507 — feat: add SAV tracks verifier resources server Maintainers reviewer not assigned 2026-09-17 23:20 UTC 2026-09-18 23:20 UTC 23d 15h
#3560 — feat(core): add a warm process pool for CPU-heavy verification Maintainers reviewer not assigned 2026-09-20 00:05 UTC 2026-09-21 00:05 UTC 23d 14h
#3512 — docs(example_session_state_mgmt): document in-memory session state Maintainers reviewer not assigned 2026-09-18 01:59 UTC 2026-09-21 01:59 UTC 23d 12h
#3520 — fix(terminus): parallel tool calls formatting Maintainers reviewer not assigned 2026-09-18 07:42 UTC 2026-09-21 07:42 UTC 23d 6h
#3146 — Add the Blobfish enterprise-operations resources server and ERPBench-100 Maintainers reviewer not assigned 2026-09-20 15:30 UTC 2026-09-21 15:30 UTC 22d 23h
#3556 — feat(sandbox): support Docker pipe process sessions Maintainers reviewer not assigned 2026-09-19 20:24 UTC 2026-09-21 20:24 UTC 22d 18h
#3548 — feat(checkpoint): add indirect prompt injection state recovery Maintainers reviewer not assigned 2026-09-18 22:51 UTC 2026-09-21 22:51 UTC 22d 15h
#3563 — feat(checkpoint): multi-worker support for policy model server Maintainers reviewer not assigned 2026-09-21 03:43 UTC 2026-09-22 03:43 UTC 22d 10h
#3585 — feat(kidbench): add KIDBench child-facing safety benchmark Maintainers reviewer not assigned 2026-09-21 23:28 UTC 2026-09-22 23:28 UTC 21d 15h
#3587 — feat(injecagent): add InjecAgent indirect prompt injection benchmark Maintainers reviewer not assigned 2026-09-22 00:44 UTC 2026-09-23 00:44 UTC 21d 13h
#3588 — feat(facts-parametric): add official public benchmark Maintainers reviewer not assigned 2026-09-22 00:47 UTC 2026-09-23 00:47 UTC 21d 13h
#3567 — feat(asb): add Agent Security Bench environment Maintainers reviewer not assigned 2026-09-22 02:06 UTC 2026-09-23 02:06 UTC 21d 12h
#3590 — feat(indic_gsm8k): add Indic GSM8K benchmark Maintainers reviewer not assigned 2026-09-22 03:50 UTC 2026-09-23 03:50 UTC 21d 10h
#3595 — feat(toolalignbench): add reproducible benchmark environment Maintainers reviewer not assigned 2026-09-22 07:13 UTC 2026-09-23 07:13 UTC 21d 7h
#3598 — feat(indic_arc_c): add Indic ARC-Challenge benchmark Maintainers reviewer not assigned 2026-09-22 08:16 UTC 2026-09-23 08:16 UTC 21d 6h
#3600 — feat(launcher): add Slurm benchmark launcher and model profiles Maintainers reviewer not assigned 2026-09-22 08:55 UTC 2026-09-23 08:55 UTC 21d 5h
#3602 — feat(launcher): add fast benchmark profiles and named resumable runs Maintainers reviewer not assigned 2026-09-22 09:08 UTC 2026-09-23 09:08 UTC 21d 5h
#3594 — feat(even-handedness): add Anthropic Political Even-handedness benchmark Maintainers reviewer not assigned 2026-09-22 14:52 UTC 2026-09-23 14:52 UTC 20d 23h
#3617 — feat(safe_child_llm): add Safe-Child-LLM developmental safety benchmark Maintainers reviewer not assigned 2026-09-22 15:06 UTC 2026-09-23 15:06 UTC 20d 23h
#3620 — feat(vera_mh): add VERA-MH clinical-rubric safety benchmark Maintainers reviewer not assigned 2026-09-22 16:09 UTC 2026-09-23 16:09 UTC 20d 22h
#3628 — feat(facts-multimodal): add image factuality benchmark Maintainers reviewer not assigned 2026-09-22 20:16 UTC 2026-09-23 20:16 UTC 20d 18h
#3634 — feat(checkpoint): resume proof refinement agent turns Maintainers reviewer not assigned 2026-09-22 22:19 UTC 2026-09-23 22:19 UTC 20d 16h
#3642 — [ApexAgents] Relay policy traffic over a unix socket for network-isolated sandboxes Maintainers reviewer not assigned 2026-09-23 00:53 UTC 2026-09-24 00:53 UTC 20d 13h
#3650 — feat(exploitbench): add ExploitBench V8 capability-ladder benchmark Maintainers reviewer not assigned 2026-09-23 06:34 UTC 2026-09-24 06:34 UTC 20d 7h
#3651 — feat(agentdyn): add AgentDyn indirect prompt-injection benchmark Maintainers reviewer not assigned 2026-09-23 06:36 UTC 2026-09-24 06:36 UTC 20d 7h
#3659 — [GDPVal] Coerce a rendered paths list nested inside the finish argument Maintainers reviewer not assigned 2026-09-23 10:24 UTC 2026-09-24 10:24 UTC 20d 4h
#3661 — feat(facts-grounding-v2): add public benchmark adapter Maintainers reviewer not assigned 2026-09-23 14:58 UTC 2026-09-24 14:58 UTC 19d 23h
#3738 — pjin/main-dev-20260928 Maintainers reviewer not assigned 2026-09-28 20:54 UTC 2026-09-29 20:54 UTC 16d 17h
#3742 — [ApexAgents] Add a context window setting for Stirrup summarization Maintainers reviewer not assigned 2026-09-28 22:16 UTC 2026-09-29 22:16 UTC 16d 16h
#3767 — feat(formal-conjectures): add the Formal Conjectures benchmark on lean_proof Maintainers reviewer not assigned 2026-09-29 13:57 UTC 2026-09-30 13:57 UTC 16d 0h
#3368 — refactor(agent): decouple simple-agent compaction from model serving Maintainers reviewer not assigned 2026-09-30 21:35 UTC 2026-10-01 21:35 UTC 14d 16h
#3830 — feat(gpqa): add Indic pretraining evaluation recipe Maintainers reviewer not assigned 2026-10-01 03:40 UTC 2026-10-02 03:40 UTC 14d 10h
#3982 — feat(agentic-mme): add the Agentic-MME environment, agent and benchmark Maintainers reviewer not assigned 2026-10-03 05:08 UTC 2026-10-05 05:08 UTC 13d 9h
#3924 — feat(longbenchv2-48k): add LongBench v2 resources server and benchmark for max 48k context length data Maintainers reviewer not assigned 2026-10-02 10:58 UTC 2026-10-05 10:58 UTC 13d 3h
#3928 — fix(simple-agent): preserve session cookies across response deltas Maintainers reviewer not assigned 2026-10-02 12:56 UTC 2026-10-05 12:56 UTC 13d 1h
#3998 — fix(core): retry transient gateway authentication failures Maintainers reviewer not assigned 2026-10-04 16:55 UTC 2026-10-05 16:55 UTC 12d 21h
#3947 — fix(gdpval): replace transport-blocked calibration tasks Maintainers reviewer not assigned 2026-10-02 18:49 UTC 2026-10-05 18:49 UTC 12d 19h
#3958 — feat(gdpval): add optional judge failure telemetry Maintainers reviewer not assigned 2026-10-02 21:14 UTC 2026-10-05 21:14 UTC 12d 17h
#3960 — feat(gdpval): add optional judge failure telemetry Maintainers reviewer not assigned 2026-10-02 21:30 UTC 2026-10-05 21:30 UTC 12d 16h
#4049 — feat(agentic-vbench): add the Agentic-VBench environment, agent and benchmark Maintainers reviewer not assigned 2026-10-05 10:42 UTC 2026-10-06 10:42 UTC 12d 3h
#4062 — claude code sandboxed agent Maintainers reviewer not assigned 2026-10-05 20:35 UTC 2026-10-06 20:35 UTC 11d 17h
#4082 — feat(rollout): add multi-harness fan-out training Maintainers reviewer not assigned 2026-10-06 01:07 UTC 2026-10-07 01:07 UTC 11d 13h
#4073 — fix(tau2): emit policy trajectory for rollout health Maintainers reviewer not assigned 2026-10-06 05:30 UTC 2026-10-07 05:30 UTC 11d 9h
#4094 — feat(orchestration): auto-size prefill/decode tier pools with nodes: auto Maintainers reviewer not assigned 2026-10-06 11:33 UTC 2026-10-07 11:33 UTC 11d 2h
#4104 — fix(local-vllm): reserve server sockets on the actor host Maintainers reviewer not assigned 2026-10-06 15:30 UTC 2026-10-07 15:30 UTC 10d 23h
#3981 — fix(agent): report native and captured token usage Maintainers reviewer not assigned 2026-10-06 16:07 UTC 2026-10-07 16:07 UTC 10d 22h
#4129 — feat(aa-briefcase-lite): add AA-Briefcase-Lite benchmark and artifact-aware judging Maintainers reviewer not assigned 2026-10-06 19:44 UTC 2026-10-07 19:44 UTC 10d 18h
#4209 — feat(osworld): OSWorld environment stack Maintainers reviewer not assigned 2026-10-08 01:26 UTC 2026-10-09 01:26 UTC 9d 13h
#4249 — feat(image_tools): reward terminal answers in the pivot environment (#3161) Maintainers reviewer not assigned 2026-10-08 17:31 UTC 2026-10-09 17:31 UTC 8d 20h
#4197 — test(checkpoint): check checkpoint invariants over random sequences of operations Maintainers reviewer not assigned 2026-10-09 07:09 UTC 2026-10-12 07:09 UTC 8d 7h
#4452 — feat(Indic Alpaca eval): add Indic AlpacaEval pointwise benchmark Maintainers reviewer not assigned 2026-10-11 12:12 UTC 2026-10-12 12:12 UTC 8d 2h
#4469 — Rshahbazyan/add verisoftbench benchmark Maintainers reviewer not assigned 2026-10-12 12:49 UTC 2026-10-13 12:49 UTC 7d 1h
#4564 — test(conformance): add P1 timeout scenarios Maintainers reviewer not assigned 2026-10-13 14:56 UTC 2026-10-14 14:56 UTC 5d 23h
#4009 — feat(model-server): add opt-in anthropic_max_tokens_as_end_turn Maintainers reviewer not assigned 2026-10-13 23:08 UTC 2026-10-14 23:08 UTC 5d 15h
#4010 — feat(vllm_model): add opt-in fold_nonleading_system_messages Maintainers reviewer not assigned 2026-10-13 23:08 UTC 2026-10-14 23:08 UTC 5d 15h
#4017 — feat(anyswe_agent): add opt-in sanitize_git_history Maintainers reviewer not assigned 2026-10-13 23:08 UTC 2026-10-14 23:08 UTC 5d 15h
#4025 — feat(sandbox): add apptainer exec.memory_limit_mib per-process memory cap Maintainers reviewer not assigned 2026-10-13 23:08 UTC 2026-10-14 23:08 UTC 5d 15h
#4036 — feat(swe_agents): add opt-in unfinished_episodes="fail" Maintainers reviewer not assigned 2026-10-13 23:08 UTC 2026-10-14 23:08 UTC 5d 15h
#4035 — feat(swe_agents): add opt-in sanitize_git_history Maintainers reviewer not assigned 2026-10-13 23:19 UTC 2026-10-14 23:19 UTC 5d 15h
#4019 — fix(anyswe_agent): make R2E-Gym grading resolve Maintainers reviewer not assigned 2026-10-13 23:20 UTC 2026-10-14 23:20 UTC 5d 15h
#4024 — fix(anyswe_agent): write grader files under the per-instance /sandbox dir Maintainers reviewer not assigned 2026-10-13 23:20 UTC 2026-10-14 23:20 UTC 5d 15h
#4026 — feat(openai_model): opt-in retry for reasoning-only empty completions Maintainers reviewer not assigned 2026-10-13 23:27 UTC 2026-10-14 23:27 UTC 5d 15h
#4027 — feat(model-server): opt-in HTTP 529 retry for thinking-only /v1/messages turns Maintainers reviewer not assigned 2026-10-13 23:27 UTC 2026-10-14 23:27 UTC 5d 15h
#4624 — fix(instruction_following): split counting_composition on * * * and stop counting punctuation as words Maintainers reviewer not assigned 2026-10-15 08:13 UTC 2026-10-16 08:13 UTC 4d 6h
#4627 — docs(skills): add-benchmark health checks and session rules Maintainers reviewer not assigned 2026-10-15 12:13 UTC 2026-10-16 12:13 UTC 4d 2h
#4630 — fix(observability): retain explicit verifier invocations Maintainers reviewer not assigned 2026-10-15 13:13 UTC 2026-10-16 13:13 UTC 4d 1h
#4631 — test(conformance): exercise model-backed verifier evidence Maintainers reviewer not assigned 2026-10-15 13:30 UTC 2026-10-16 13:30 UTC 4d 1h
#4636 — fix(harbor): reject directory symlinks before task hashing Maintainers reviewer not assigned 2026-10-15 14:54 UTC 2026-10-16 14:54 UTC 3d 23h
#4640 — perf(server): cache static global config lookups on request paths Maintainers reviewer not assigned 2026-10-15 17:46 UTC 2026-10-16 17:46 UTC 3d 20h
#4643 — perf(eval): tune garbage collection in the rollout driver Maintainers reviewer not assigned 2026-10-15 18:14 UTC 2026-10-16 18:14 UTC 3d 20h
#4652 — feat(environment): pass the rollout index to resources seeds Maintainers reviewer not assigned 2026-10-15 20:17 UTC 2026-10-16 20:17 UTC 3d 18h
#4679 — fix(resources): tighten the session contract for close, refusals and lost sessions Maintainers reviewer not assigned 2026-10-16 17:23 UTC 2026-10-19 17:23 UTC 2d 21h
#4678 — docs(evaluate): fix gym eval run examples that fail as written Maintainers reviewer not assigned 2026-10-16 21:34 UTC 2026-10-19 21:34 UTC 2d 16h
#4681 — docs(training): document task_source routing instead of agent_ref Maintainers reviewer not assigned 2026-10-16 21:34 UTC 2026-10-19 21:34 UTC 2d 16h
#4682 — docs(training): fix offline SFT/DPO rollout conversion snippets Maintainers reviewer not assigned 2026-10-16 21:34 UTC 2026-10-19 21:34 UTC 2d 16h
#4649 — feat(geobenchx): add the GeoBenchX multistep GIS benchmark Maintainers reviewer not assigned 2026-10-19 13:27 UTC 2026-10-20 13:27 UTC 2d 1h
#4754 — docs: correct ServerClient retry behavior Maintainers reviewer not assigned 2026-10-19 19:08 UTC 2026-10-20 19:08 UTC 1d 19h
#4680 — fix(agent): refuse unsupported agent session seeds with a final 422 Maintainers reviewer not assigned 2026-10-19 21:00 UTC 2026-10-20 21:00 UTC 1d 17h
#4692 — fix(resources): move swebench_pro and interactive_browser onto the session contract Maintainers reviewer not assigned 2026-10-19 21:00 UTC 2026-10-20 21:00 UTC 1d 17h
#4771 — fix(cli): quote the eval aggregate --input-glob override Maintainers reviewer not assigned 2026-10-20 02:23 UTC 2026-10-21 02:23 UTC 1d 12h
#4772 — fix(eval): count unanswered votes in majority@k/no_answer Maintainers reviewer not assigned 2026-10-20 02:23 UTC 2026-10-21 02:23 UTC 1d 12h
#4775 — fix(swebench): grade staged and committed fixes, not just unstaged changes Maintainers reviewer not assigned 2026-10-20 06:31 UTC 2026-10-21 06:31 UTC 1d 7h
#4784 — sandbox: add pty capability to e2b sandbox provider Maintainers reviewer not assigned 2026-10-20 10:11 UTC 2026-10-21 10:11 UTC 1d 4h
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions