[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M71RVCY0SAGWAVETZ3A3HES0":3,"skill-same:01M71RVCY0SAGWAVETZ3A3HES0":74},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":28,"publishedAt":46,"linkOnly":47,"markdownContent":48,"markdownBody":49,"frontMatter":50,"refName":51,"commitSha":52,"skillDirectory":53,"forksCount":54,"repositoryHtmlUrl":55,"repositoryHomepageUrl":6,"ownerHtmlUrl":56,"ownerBlogUrl":57,"ownerDescription":58,"updatedAt":59,"install":60,"resources":73},"01M71RVCY0SAGWAVETZ3A3HES0","debug-agent-session",null,"Use when an Eve agent is behaving wrong (stuck, gaming the score, bad reasoning) and you need to read its own rollout to find out why","scaling-group\u002Feve","scaling-group","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F181708022?v=4","01M71RTM8RQW6CX3M1934895JA","Scientific Computing and Intelligence Group","01M71RTM99S28ZTFK9BK17R4GC",66,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Fdebug-agent-session\u002FSKILL.md",[18,23],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",{"categoryNo":24,"categoryName":25,"taxonomyType":26,"taxonomyTypeDesc":27},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45],"session","debug","debug agent","agent session","debug agent session","Wrong","gaming","reasoning","read","why","find","you","rollout","when","use","eve","bad","2026-11-10 13:02:02",false,"---\nname: debug-agent-session\ndescription: \"Use when an Eve agent is behaving wrong (stuck, gaming the score, bad reasoning) and you need to read its own rollout to find out why\"\n---\n\n# Debug an agent's session\n\nWhen a run looks wrong (stalled, gaming, no progress) and the score\u002Fpopulation don't explain it, read the agent's own rollout, its full conversation and tool calls, to see *why* it did what it did. Local files only. run_root layout is in `docs\u002Fusing-eve.md`.\n\n## Where the rollout session is\n\nEach active solver worker runs in `\u003Crun_root>\u002Fsolver_workspaces\u002F\u003Cworkspace_id>\u002F`. The agent's own session record (JSONL, one event per line) lives there:\n\n- Codex: `\u003Cworkspace_id>\u002F.codex-driver-transcripts\u002F\u003Csession>.jsonl` (full rollout also under `\u003Cworkspace_id>\u002F.codex-driver-home\u002F.codex\u002Fsessions\u002FYYYY\u002FMM\u002FDD\u002Frollout-*.jsonl`)\n\nA `*-live.jsonl` is an in-progress session; the non-live file is the completed snapshot. (Note: `\u003Crun_root>\u002Fartifacts\u002F\u003Crun-id>_solver\u002Ftranscripts\u002F` holds the optimization *log trees*, not the agent conversation, use the workspace JSONL for behavior.) To find the workspace for a suspect candidate, locate its `..._step_\u003Cn>_...` directory under `\u003Crun_root>\u002Fsolver_workspaces\u002F` (see `..\u002Finspect-population\u002FSKILL.md`); there is no DB column mapping an entry to its workspace path.\n\nIf the run has been resumed, failed post-checkpoint workspaces may have been moved to `\u003Crun_root>\u002F.resume_archive\u002Fresume_*\u002Fsolver_workspaces\u002F\u003Cworkspace_id>\u002F`. Check active `solver_workspaces\u002F` first for the current logical run, then `.resume_archive\u002F` only when debugging the interrupted attempt itself.\n\n## What to look for\n\nRead the JSONL for the agent's reasoning + tool calls. Common failure signatures, with the `runner.log` markers that flag them:\n\n- **Optimizer no-op** (`the guidance tree was not modified; no optimizer candidate will be produced`): the optimizer agent isn't editing its guidance. Read its session to see whether it understood the task.\n- **Synced 0 optimizers** (`Phase 3: updated Elo for N optimizers and synced 0 Phase 2 optimizers`): the optimizer side produces nothing; two in a row is a stop-and-flag signal (see `..\u002Fsupervise-run\u002FSKILL.md`).\n- **Family lock-in:** `optimizer_lineage.db` shows only one optimizer `entry_id` reused across iterations (selection keeps picking the same one), exploration collapsed.\n- **Gaming the eval:** the candidate scores high but its `solver\u002F` (via `..\u002Finspect-population\u002FSKILL.md`) games the metric instead of solving the task; the session often shows the agent reverse-engineering the evaluator.\n\n## Note\n\nA mechanical halt (crash, killed process) is not an agent-behavior problem, recover it with `..\u002Fresume-run\u002FSKILL.md`.\n","# Debug an agent's session\n\nWhen a run looks wrong (stalled, gaming, no progress) and the score\u002Fpopulation don't explain it, read the agent's own rollout, its full conversation and tool calls, to see *why* it did what it did. Local files only. run_root layout is in `docs\u002Fusing-eve.md`.\n\n## Where the rollout session is\n\nEach active solver worker runs in `\u003Crun_root>\u002Fsolver_workspaces\u002F\u003Cworkspace_id>\u002F`. The agent's own session record (JSONL, one event per line) lives there:\n\n- Codex: `\u003Cworkspace_id>\u002F.codex-driver-transcripts\u002F\u003Csession>.jsonl` (full rollout also under `\u003Cworkspace_id>\u002F.codex-driver-home\u002F.codex\u002Fsessions\u002FYYYY\u002FMM\u002FDD\u002Frollout-*.jsonl`)\n\nA `*-live.jsonl` is an in-progress session; the non-live file is the completed snapshot. (Note: `\u003Crun_root>\u002Fartifacts\u002F\u003Crun-id>_solver\u002Ftranscripts\u002F` holds the optimization *log trees*, not the agent conversation, use the workspace JSONL for behavior.) To find the workspace for a suspect candidate, locate its `..._step_\u003Cn>_...` directory under `\u003Crun_root>\u002Fsolver_workspaces\u002F` (see `..\u002Finspect-population\u002FSKILL.md`); there is no DB column mapping an entry to its workspace path.\n\nIf the run has been resumed, failed post-checkpoint workspaces may have been moved to `\u003Crun_root>\u002F.resume_archive\u002Fresume_*\u002Fsolver_workspaces\u002F\u003Cworkspace_id>\u002F`. Check active `solver_workspaces\u002F` first for the current logical run, then `.resume_archive\u002F` only when debugging the interrupted attempt itself.\n\n## What to look for\n\nRead the JSONL for the agent's reasoning + tool calls. Common failure signatures, with the `runner.log` markers that flag them:\n\n- **Optimizer no-op** (`the guidance tree was not modified; no optimizer candidate will be produced`): the optimizer agent isn't editing its guidance. Read its session to see whether it understood the task.\n- **Synced 0 optimizers** (`Phase 3: updated Elo for N optimizers and synced 0 Phase 2 optimizers`): the optimizer side produces nothing; two in a row is a stop-and-flag signal (see `..\u002Fsupervise-run\u002FSKILL.md`).\n- **Family lock-in:** `optimizer_lineage.db` shows only one optimizer `entry_id` reused across iterations (selection keeps picking the same one), exploration collapsed.\n- **Gaming the eval:** the candidate scores high but its `solver\u002F` (via `..\u002Finspect-population\u002FSKILL.md`) games the metric instead of solving the task; the session often shows the agent reverse-engineering the evaluator.\n\n## Note\n\nA mechanical halt (crash, killed process) is not an agent-behavior problem, recover it with `..\u002Fresume-run\u002FSKILL.md`.\n",{"name":5,"description":7},"main","979065fd2cf1b3be9caf70596478fe27df7815ad","docs\u002Fskills\u002Fdebug-agent-session",9,"https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve","https:\u002F\u002Fgithub.com\u002Fscaling-group","https:\u002F\u002Fscaling-group.github.io\u002F","Department of Mathematics, National University of Singapore","2026-08-20 15:35:50",{"source":8,"skillName":5,"commands":61,"prompt":71,"zipAvailable":72},[62,65,68],{"tool":63,"command":64},"npx","npx skills add scaling-group\u002Feve --skill debug-agent-session",{"tool":66,"command":67},"bunx","bunx skills add scaling-group\u002Feve --skill debug-agent-session",{"tool":69,"command":70},"pnpm","pnpm dlx skills add scaling-group\u002Feve --skill debug-agent-session","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Fdebug-agent-session\u002FSKILL.md",true,[],[75,95,113,127,146,162],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":87,"publishedAt":46},"01M71RW4VG8A6FFRGT7H41J16J","writing-skills","Use when creating new skills, editing existing skills, or verifying skills work before deployment","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Fwriting-skills\u002FSKILL.md",[81,84],{"categoryNo":82,"categoryName":83,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":85,"categoryName":86,"taxonomyType":26,"taxonomyTypeDesc":27},"01JBSEEDCAT000000000000036","内容创作者",[88,89,90,91,92,93,43,94,42],"writing","writing skills","work","editing","deployment","new","Existing",{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":104,"publishedAt":46},"01M71RW2PFRVWH24AP25T6AV4E","theorem-search","Use TheoremSearch to find, verify, and cite mathematical theorem statements, named results, proof routes, and literature dependencies.","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Ftheorem-search\u002FSKILL.md",[101],{"categoryNo":102,"categoryName":103,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000021","资料检索与研究",[105,106,43,107,108,109,39,110,111,112],"theorem search","search","literature","results","verify","dependencies","proof","Mathematical",{"skillNo":114,"name":115,"nameZh":6,"description":116,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":46},"01M71RW0N0GQR9GW1B3HKBB800","supervise-run","Use when monitoring a running Eve experiment and recovering from problems while it runs","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Fsupervise-run\u002FSKILL.md",[],[120,121,122,123,124,125,126,42,43,44],"supervise run","supervise","run","monitoring","experiment","it","running",{"skillNo":128,"name":129,"nameZh":6,"description":130,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":131,"categories":132,"keywords":133,"publishedAt":46},"01M71RVYA101VVX3CKR2HZHBT3","run-math-proof-quickstart","Use when running a Math Proof quickstart from a user-provided problem: create a local task copy, launch a real EvE attempt, supervise it, and report proof\u002Fevaluation results.","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Frun-math-proof-quickstart\u002FSKILL.md",[],[122,134,135,136,137,111,138,139,126,140,141,44,43,42,142,143,108,121,144,145],"math","run math proof quickstart","quickstart","math proof","copy","launch","problem","local","task","create","report","Evaluation",{"skillNo":147,"name":148,"nameZh":6,"description":149,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":150,"categories":151,"keywords":152,"publishedAt":46},"01M71RVWYKH6TF0WB6X57GMWJ2","run-circle-packing-smoke","Use when a fast end-to-end smoke run is needed for the Eve loop development, will run on circle packing task.","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Frun-circle-packing-smoke\u002FSKILL.md",[],[153,154,122,155,156,42,157,158,159,44,43,160,142,161],"circle","packing","smoke","run circle packing smoke","development","for","end","loop","fast",{"skillNo":163,"name":164,"nameZh":6,"description":165,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":166,"categories":167,"keywords":171,"publishedAt":46},"01M71RVVCE1NEF4ZM405WX815Z","resume-run","Use when continuing an interrupted or paused Eve run from where it stopped (same run, same id), optionally after editing the experiment","https:\u002F\u002Fgithub.com\u002Fscaling-group\u002Feve\u002Fblob\u002Fmain\u002Fdocs\u002Fskills\u002Fresume-run\u002FSKILL.md",[168],{"categoryNo":169,"categoryName":170,"taxonomyType":26,"taxonomyTypeDesc":27},"01JBSEEDCAT000000000000038","人力资源",[172,122,173,91,44,43,42,174,175,124,125],"resume","resume run","id","same"]