lobes-cli 0.78.1__tar.gz → 0.78.3__tar.gz

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/.devague/deliveries/orin-associate-at-1m.json

@@ -2,7 +2,7 @@
   "plan_slug": "orin-associate-at-1m",
   "schema_version": 2,
   "created": "2026-09-13T05:41:01Z",
-  "updated": "2026-09-13T08:52:22Z",
+  "updated": "2026-09-13T12:08:46Z",
   "deviations": [
     {
       "id": "d1",
@@ -78,7 +78,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 4
     },
@@ -97,7 +97,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 5
     },
@@ -116,7 +116,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 6
     },
@@ -135,7 +135,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 7
     },
@@ -154,7 +154,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 8
     },
@@ -173,7 +173,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 9
     },
@@ -192,7 +192,7 @@
         "commit": "6eeacca"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 10
     },
@@ -211,7 +211,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 11
     },
@@ -230,7 +230,7 @@
         "commit": "6eeacca"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 12
     },
@@ -249,7 +249,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 13
     },
@@ -268,7 +268,7 @@
         "commit": "6eeacca"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 14
     },
@@ -287,7 +287,7 @@
         "commit": "6eeacca"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 15
     },
@@ -306,7 +306,7 @@
         "commit": "6eeacca"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 16
     },
@@ -322,7 +322,7 @@
       "outcome": "pass",
       "run": null,
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 17
     },
@@ -341,7 +341,7 @@
         "commit": "6eeacca"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 18
     },
@@ -357,7 +357,7 @@
       "outcome": "pass",
       "run": null,
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 19
     },
@@ -373,7 +373,7 @@
       "outcome": "pass",
       "run": null,
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 20
     },
@@ -392,7 +392,7 @@
         "commit": "b6a287b"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 21
     },
@@ -408,7 +408,7 @@
       "outcome": "fail",
       "run": null,
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 22
     },
@@ -424,7 +424,7 @@
       "outcome": "pass",
       "run": null,
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 23
     },
@@ -440,7 +440,7 @@
       "outcome": "pass",
       "run": null,
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 24
     },
@@ -459,7 +459,7 @@
         "commit": "4b9be8c"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 26
     },
@@ -478,7 +478,7 @@
         "commit": "4b9be8c"
       },
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false,
       "seq": 27
     }
@@ -495,7 +495,7 @@
         "e2"
       ],
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false
     },
     {
@@ -509,7 +509,7 @@
         "e17"
       ],
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false
     },
     {
@@ -523,7 +523,7 @@
         "e18"
       ],
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false
     },
     {
@@ -537,7 +537,7 @@
         "e13"
       ],
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
       "superseded": false
     },
     {
@@ -549,9 +549,32 @@
       ],
       "evidence_refs": [],
       "origin": "llm",
-      "status": "proposed",
+      "status": "approved",
+      "superseded": true
+    },
+    {
+      "id": "b6",
+      "kind": "added",
+      "behavior_text": "shapes hosting associate boot it before the pooling gears on a plain compose up (shape override reverses the base template's gears-first depends_on)",
+      "caused_by": [
+        "d4"
+      ],
+      "evidence_refs": [
+        "e22",
+        "e23"
+      ],
+      "origin": "llm",
+      "status": "approved",
       "superseded": false
     }
   ],
-  "supersessions": []
+  "supersessions": [
+    {
+      "id": "s1",
+      "action": "supersede",
+      "target_ref": "b5",
+      "replacement_ref": "b6",
+      "origin": "user"
+    }
+  ]
 }

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/.devague/frames/orin-associate-at-1m.json

@@ -4,7 +4,7 @@
   "schema_version": 6,
   "status": "exported",
   "created": "2026-09-13T01:51:16Z",
-  "updated": "2026-09-13T08:52:22Z",
+  "updated": "2026-09-13T12:24:51Z",
   "claims": [
     {
       "id": "c1",
@@ -575,8 +575,8 @@
       "text": "whether vLLM v0.27.1 streaming emits any bytes (role chunk, headers) before the first generated token; if it does, a streamed cold 1M request might survive a 600 s gateway read timeout while a non-streamed one does not; unprobed",
       "kind": "unknown_nonblocking",
       "claim_id": null,
-      "resolved": false,
-      "resolution": "",
+      "resolved": true,
+      "resolution": "Answered by docs/evidence/2026-09-13-accept-orin-associate-1m.txt step 10: vLLM v0.27.1 emits no bytes before the first generated token, so a streamed cold 1M request fails a 600 s gateway read timeout exactly like a non-streamed one. The Orin card's GATEWAY_READ_TIMEOUT=7200 covers both; both cold >=1M requests passed through the gateway (non-streamed 2,495.1 s wall; streamed first byte at 1,971.42 s). Tracked in #265.",
       "resolution_claim_id": null
     }
   ],
@@ -796,7 +796,7 @@
       "behavior": "ASSOCIATE_MAX_NUM_SEQS set renders --max-num-seqs=N; unset renders no argv token",
       "source_text": "the vllm-associate lane gains an ASSOCIATE_MAX_NUM_SEQS knob rendered with the worker lane's conditional idiom (unset renders no argv token), so a 1M-window lane can cap concurrent sequences; documented in the lane's knob comment and pinned in tests/test_associate_compose.py",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o2",
@@ -805,7 +805,7 @@
       "behavior": "VLLM_ALLOW_LONG_MAX_MODEL_LEN renders 0 by default and 1 when ASSOCIATE_ALLOW_LONG_MAX_MODEL_LEN=1",
       "source_text": "the vllm-associate lane gains a VLLM_ALLOW_LONG_MAX_MODEL_LEN environment passthrough (ASSOCIATE_ALLOW_LONG_MAX_MODEL_LEN, default 0) mirroring vllm-primary, required only when the served checkpoint declares max_position_embeddings below the requested window",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o3",
@@ -814,7 +814,7 @@
       "behavior": "--host=0.0.0.0 with no published host port and no network_mode; exposure tests unmodified",
       "source_text": "the associate lane's exposure contract is unchanged: --host=0.0.0.0 paired with no published host port (expose-only behind the gateway bearer gate), as pinned by tests/test_associate_exposure.py",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o4",
@@ -823,7 +823,7 @@
       "behavior": "exactly one associate entry (the NVIDIA NVFP4 checkpoint); an injected duplicate is detected",
       "source_text": "the catalog keeps exactly one role_hint=\"associate\" entry (the NVIDIA NVFP4 checkpoint), pinned by a new test_exactly_one_associate_gear; the useful-quants W4A16 is NOT added to the catalog (A/B 2026-09-13 missed the c24 bar, v2 resolved)",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o5",
@@ -832,7 +832,7 @@
       "behavior": "orin card renders GATEWAY_READ_TIMEOUT=7200 (spark/thor/base do not) and a cold >= 1M associate request through the Orin gateway completes, streamed and non-streamed",
       "source_text": "a cold 1M request through the Orin gateway as model=associate completes: the Orin gateway's upstream read timeout (GATEWAY_READ_TIMEOUT, template default 600 s, live Orin value 600 s, applied as the socket read timeout on every upstream POST at lobes/gateway/server.py:616) is raised above the measured cold 1.04M TTFT (2,390 s NVFP4 on 2026-09-13) with margin, set through a shipped knob (env.example + orin card or shape), not a hand-typed key",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o6",
@@ -841,7 +841,7 @@
       "behavior": "renders max_model_len 1048576, gpu_mem_util 0.70, max_num_batched_tokens 8192, max_num_seqs 2, citing the measure transcript",
       "source_text": "orin-associate.toml's associate override declares the measured 1M budget (max_model_len 1048576 plus the util, batched-token cap and max_num_seqs the A/B measured) with each value citing a docs/evidence transcript, replacing the 128000 window",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o7",
@@ -850,7 +850,7 @@
       "behavior": "hosts associate, embedder and reranker; drops cortex, senses, hand, muse, worker",
       "source_text": "orin-associate.toml's hosts list is associate, embedder and reranker (the pooling gears return to the shape), its header comment describes exactly that, and tests/test_orin_associate_shape.py plus the orin-associate golden assert the three hosted roles and the dropped ones",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o8",
@@ -859,7 +859,7 @@
       "behavior": "after rollout associate advertises context 1048576 and the lane reports max_model_len 1048576",
       "source_text": "after the window change the Orin gateway advertises context 1048576 for associate on GET /capabilities and lobes capabilities: the gateway container is recreated (not only vllm-associate), and acceptance checks both the advertised context and the live fingerprint max_model_len read from the lane's /v1/models",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o9",
@@ -868,7 +868,7 @@
       "behavior": "health 200, probes PASS, >= 1M needle PASS, capabilities 1048576, zero OOM across a 2-session agentic run, associate restarts 0",
       "source_text": "on the physical Orin with the rendered shape: /health 200; known-answer, multi-step and tool-call probes PASS; a needle at >= 1,000,000 prompt tokens PASSES; GET /capabilities reports associate context == 1048576; zero OOMKilled/engine-dead events across a 2-session 12-turn agentic run; associate restarts == 0 after the run",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o10",
@@ -877,7 +877,7 @@
       "behavior": "one row per arm for every listed metric and a verdict applying the c24 rule",
       "source_text": "the A/B transcript records, for NVFP4 and W4A16 at 1,048,576: boot result, KV pool tokens, decode tok/s at >= 3 depths, cold TTFT at ~1M, concurrency 1/2/4 aggregate tok/s, agentic tool-call validity and recall counts, peak tj and minimum available memory; the winner is chosen by a stated rule, not by feel",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o11",
@@ -886,7 +886,7 @@
       "behavior": "embed/rerank RestartCount quoted after rollout; the Orin deployment doc names the gear first-start race",
       "source_text": "the accept transcript records embed and reranker restart counts after rollout, and the Orin deployment doc names the observed gear first-start race: on 2026-09-13 model-gear-vllm-embed failed its first start every time the gears started together after associate (02:29, 03:34, 04:31 UTC; 'No available memory for the cache blocks') and recovered via restart=unless-stopped",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o12",
@@ -895,7 +895,7 @@
       "behavior": "shape comment quotes the 128000 / 0.70 / 8192 rollback; the accept transcript quotes the dated .env backup",
       "source_text": "orin-associate.toml keeps the pre-change 128000-window associate values as a documented rollback in its comment block (the spark-lobe d4 idiom), and the rollout records a dated backup of the Orin .env before rendering",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o13",
@@ -904,7 +904,7 @@
       "behavior": "docs quote the shipped-shape minimum (1,925 MiB, deviation d2) and the A/B's 2,588 MiB with transcript paths, plus the memory-capped side-process rule",
       "source_text": "the Orin associate docs record the measured zero-swap headroom at 1M (minimum available host memory 2,588 MiB during the NVFP4 1.04M prefill, 2026-09-13) together with the operating rule that any side process on the Orin runs memory-capped",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o14",
@@ -913,7 +913,7 @@
       "behavior": "shape header/summary and card doc block match what ships; retired budget literals only historical; shape validation claimed only by citing the accept transcript",
       "source_text": "stale Orin associate statements are reconciled with what ships: the orin-associate header (still describes hand + pooling gears), orin.toml's 0.56 LOCKSTEP doc block, docs/deployment-shapes.md:91 (pre-d2 util 0.63 KV figures) and docs/orin-associate-deployment.md (16384 batched tokens vs the live 8192)",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o15",
@@ -922,7 +922,7 @@
       "behavior": "CLAUDE.md, deployment-shapes, the Lightning doc, machine-profiles and the Orin deployment doc cite both 1M transcripts",
       "source_text": "every doc that states associate's Orin budget or window as fact is updated to the measured 1M values: CLAUDE.md associate paragraph, docs/deployment-shapes.md orin-associate rows, docs/nemotron-3.5-lightning-30b-a3b-nvfp4.md Orin section, docs/machine-profiles.md totals, docs/orin-associate-deployment.md",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     },
     {
       "id": "o16",
@@ -931,7 +931,7 @@
       "behavior": "the generated shape override makes vllm-embed and vllm-rerank wait for vllm-associate healthy and clears vllm-associate depends_on, so a plain compose up boots associate first",
       "source_text": "rendering orin-associate on the Orin produces a lane that boots at max_model_len 1048576 with the A/B-measured checkpoint, util, batched-token cap and max_num_seqs; the gateway advertises context 1048576; and every value cites a docs/evidence transcript",
       "origin": "llm",
-      "status": "proposed"
+      "status": "approved"
     }
   ]
 }

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/.devague/plans/orin-associate-at-1m.json

@@ -5,7 +5,7 @@
   "schema_version": 5,
   "status": "exported",
   "created": "2026-09-13T05:20:28Z",
-  "updated": "2026-09-13T05:50:27Z",
+  "updated": "2026-09-13T12:25:15Z",
   "targets": [
     {
       "id": "c1",
@@ -507,8 +507,8 @@
       "text": "re-scaffolding the Orin with lobes init --shape orin-associate --apply --force OVERWRITES docker-compose.yml, Dockerfile.gateway, qwen3_reranker.jinja, mg-logwrap.sh and the tool-parser plugin from the packaged templates with NO backup (lobes/runtime/_compose.py:471-525, :242-259); docker-compose.override.yml (mesh block) is never written by init (_compose.py:130-137); .env rendered keys are rewritten in place via set_env while unrendered operator keys (ASSOCIATE_IMAGE, ASSOCIATE_SPECULATIVE_CONFIG, GATEWAY_API_KEY, LOBES_MESH_*) are preserved (init.py:588-624). --force is required for the t1 compose knob to exist on the box, so t5 must tar-backup ~/.lobes first (including the local-wheel Dockerfile.gateway) and diff afterwards",
       "kind": "unknown_nonblocking",
       "task_id": "t5",
-      "resolved": false,
-      "resolution": ""
+      "resolved": true,
+      "resolution": "Mitigated, nothing lost: before rendering, ~/.lobes was tarred (including the local-wheel Dockerfile.gateway) and .env backed up, and the result was diffed afterwards (docs/evidence/2026-09-13-accept-orin-associate-1m.txt step 1). docker-compose.override.yml was untouched and operator-typed .env keys were preserved. The only side effect was the stale compose coming up to date, which gave the reranker the #227 chat template (delta b4). The product gap (init --force overwrites without a backup) is noted in #265."
     },
     {
       "id": "r2",
@@ -523,8 +523,8 @@
       "text": "acceptance wall time and associate downtime: two cold >= 1M requests through the gateway (~40 min each, measured 2,390 s TTFT) plus the 2-session agentic run and boot; with max_num_seqs=2 they must run sequentially (head-of-line, frame park v3)",
       "kind": "unknown_nonblocking",
       "task_id": "t5",
-      "resolved": false,
-      "resolution": ""
+      "resolved": true,
+      "resolution": "Realized as planned: associate offline 06:00:31-06:06:36Z; embed and rerank lanes back at 06:09:40Z; the cold >=1M needles and the 2-session agentic run then ran sequentially against the live lane (accept transcript). The separate ~7 min 08:41-08:48Z outage came from the d4 live check and is tracked in #262. Tracked in #265."
     }
   ],
   "obligations": []

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/CHANGELOG.md

@@ -4,6 +4,18 @@ All notable changes to this project are documented here. The format is based on
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project
 adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.78.3] - 2026-09-13
+
+### Changed
+
+- Resolve frame park v5 and plan risks r1/r3 for plan orin-associate-at-1m from the accept transcript; parks v3/v4 stay open, tracked in #265.
+
+## [0.78.2] - 2026-09-13
+
+### Changed
+
+- Operator approved the full delivery ledger for plan orin-associate-at-1m: obligations o1-o16, evidence e1-e23, behavioral deltas b1-b5.
+
 ## [0.78.1] - 2026-09-13
 
 ### Changed

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: lobes-cli
-Version: 0.78.1
+Version: 0.78.3
 Summary: lobes — run, assess, and switch the local vLLM model.
 Project-URL: Homepage, https://github.com/agentculture/lobes-cli
 Project-URL: Issues, https://github.com/agentculture/lobes-cli/issues

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "lobes-cli"
-version = "0.78.1"
+version = "0.78.3"
 description = "lobes — run, assess, and switch the local vLLM model."
 readme = "README.md"
 license = "Apache-2.0"

{lobes_cli-0.78.1 → lobes_cli-0.78.3}/uv.lock

@@ -440,7 +440,7 @@ wheels = [
 
 [[package]]
 name = "lobes-cli"
-version = "0.78.1"
+version = "0.78.3"
 source = { editable = "." }
 
 [package.optional-dependencies]