fix: bound worker restarts and prepare 0.28.2

.claude-plugin/marketplace.json

@@ -12,7 +12,7 @@
       "displayName": "Agent Stack",
       "source": "./plugins/agent-stack",
       "description": "Four skills: agent-orchestrator — tool-calling loops, pipelines with checkpoints, provider routing with fallback, memory architecture, plus the wallet side of reselling LLM access; agent-evals — run/trace/thread evals, LLM judges, and fixtures grown from production; agent-interop — MCP servers and clients, A2A agent cards, the MCP Registry, and gateways; agent-harness — system prompts, tool shaping, workflow-vs-agent, and auditing an agent system.",
-      "version": "0.28.1",
+      "version": "0.28.2",
       "author": {
         "name": "ssheleg",
         "url": "https://x.com/sshlg93"

CHANGELOG.md

@@ -1,3 +1,18 @@
+## 0.28.2 — bounded restart admission
+
+- Resume only an owned stopped worker through the host's supported operation.
+  A quiet stream does not prove a stall; one failed lane does not authorize
+  stopping siblings or duplicating their in-flight work.
+- Reconcile uncertain outward effects before replay. Run IDs and unchanged
+  arguments do not guarantee cached completion or preserved context.
+- Record finite attempt and deadline bounds, including host retries, and retain
+  provider Retry-After/reset timing. Recheck task authority and ownership before
+  each attempt; preserve the best partial when the series ends.
+- Keep independent work moving while a watcher handles the eligible retry.
+  Billing, sign-in, consent and secret-entry gates remain human; no account/key
+  switching or cap bypass. Focused text controls cover known unsafe instructions,
+  without claiming observed model or scheduler behavior.
+
 ## 0.28.1 — a limit stop is not a wait
 
 There is no 0.28.0 package. The tag `v0.28.0` was pushed by mistake onto `cb8f699` (the 0.27.1

README.md

@@ -135,6 +135,10 @@ mechanisms for one idea, the four double-texting policies, streaming a dropped
 connection can rejoin, forking a past checkpoint to debug through the real loop,
 stateful versus stateless schedules, and the seven cross-cutting concerns pulled
 out of the loop into ordered interceptors — where order is semantics.
+Usage-limit recovery preserves lane ownership and uncertain effects, uses finite
+attempt/deadline budgets and provider timing, and keeps independent work moving.
+The [restart correction packet](docs/evidence/runs/2026-10-13-restart-boundaries.md)
+records the counterexamples and the limits of the instruction-only checks.
 
 **`references/governance.md`** — permission rather than cost. The four boundaries
 an agent crosses (model, tool, external server, agent-to-agent), each with its own

SKILL-CARD.md

@@ -5,7 +5,7 @@
 | Field | Value |
 |---|---|
 | Pack | `agent-stack` |
-| Version | `0.28.1` |
+| Version | `0.28.2` |
 | Skills | `agent-orchestrator`, `agent-evals`, `agent-interop`, `agent-harness` |
 | License | MIT |
 | Source | https://github.com/ssheleg/agent-stack |

docs/HANDOFF.md

@@ -1,11 +1,21 @@
-# Current work — published agent-stack 0.27.2
+# Current work — agent-stack 0.28.2 release candidate (not published)
 
-R30 owner-board and release-ancestry gates are merged and published as0.27.2,
-source `792e3f6cb2e46517f0187066db55edc63ae7d12c`. All four release jobs passed,
-and all44 npm payload files matched that source. The
-[release packet](evidence/runs/2026-10-13-release-0.27.2.md#published-readback)
-is the durable entry with exact sources, gates and limitations.
+Agent-stack0.28.1 source `cd6421b8ca7d9440bc87e366ebd8a96bb38b69b8` is published,
+but independent review found the restart instruction overbroad. The current
+[restart correction packet](evidence/runs/2026-10-13-restart-boundaries.md) owns
+scope, counterexamples, checks, remaining gates and the exact next action.
 
-Exact next task: coordinator pins0.27.2 in the next family release and verifies
-installed files. B-161 source/publication scope is closed; the family-wide R30
-and native runtime remain distinct. Preserve foreign worktrees and old tags.
+Independent source review of the doctrine correction passed at
+`ff797c6341fe6cdb4d9b0dd2e113f7ebb8f24f6f`, recorded in the hub's
+[peer receipt](https://github.com/ssheleg/sshlg-skills/blob/82310a5/docs/evidence/completion-1013/agent-stack-restart-peer.json).
+Version metadata now targets 0.28.2; this is not publication evidence.
+The [release packet](evidence/runs/2026-10-13-release-0.28.2.md) owns pending gates.
+
+Next: independent metadata review, one coordinated local owner aggregate, normal
+exact-head PR checks and merge, then exact accepted-main local full chain before
+the annotated tag and release/readback; do not change either existing0.28.0 or0.28.1 tag. Hub keeps its
+accepted0.27.2 pin until the correction is accepted and published. No native retry
+mechanism or model behavior is claimed by these documentation tests.
+
+Earlier0.27.2 source/publication evidence remains in the
+[release packet](evidence/runs/2026-10-13-release-0.27.2.md#published-readback).

docs/evidence/runs/2026-10-13-release-0.28.2-candidate.json

@@ -0,0 +1,295 @@
+{
+  "schema": "release-candidate/1",
+  "version": "0.28.2",
+  "status": "CANDIDATE_NOT_PUBLISHED",
+  "reviewedDoctrine": "ff797c6341fe6cdb4d9b0dd2e113f7ebb8f24f6f",
+  "observedAt": "2026-10-13T08:45:01.162095+00:00",
+  "checks": [
+    {
+      "argv": [
+        "python3",
+        "-B",
+        "test/validate.py"
+      ],
+      "exit": 0,
+      "outputSha256": "6822712dfac56cb79ac20c371d64adc75758eb5c021c53984e5439546a6ac73a"
+    },
+    {
+      "argv": [
+        "python3",
+        "-B",
+        "test/audit_regressions/restart-boundaries.py"
+      ],
+      "exit": 0,
+      "outputSha256": "cd0858c45bc89bcad050bf7426a8af8e400eeb63851b89e81fe647cbeb161e2d"
+    },
+    {
+      "argv": [
+        "claude",
+        "plugin",
+        "validate",
+        ".",
+        "--strict"
+      ],
+      "exit": 0,
+      "outputSha256": "e681c9722b621a912c34443001df8b4c02c4276cc50f4004a70a37f4f957d358"
+    },
+    {
+      "argv": [
+        "claude",
+        "plugin",
+        "validate",
+        "plugins/agent-stack",
+        "--strict"
+      ],
+      "exit": 0,
+      "outputSha256": "bc22c2f268f8d30ec7fe5e5cbcc6cb95fa6766bbf57deb4802cc94c229134df5"
+    },
+    {
+      "argv": [
+        "git",
+        "diff",
+        "--check"
+      ],
+      "exit": 0,
+      "outputSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
+    }
+  ],
+  "npmDryRunExit": 0,
+  "payload": [
+    {
+      "path": "CHANGELOG.md",
+      "sha256": "0fce6df83bb875438bbcd2dd1609665ae19348284405c1b2c7b075514a375e7e",
+      "sameAsReviewedDoctrine": false
+    },
+    {
+      "path": "LICENSE",
+      "sha256": "826df230e15df9b208c1e08c42ef3b764b21c4abec5a91a38976748e92a82e9d",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "README.md",
+      "sha256": "49953599fcd54dfcba0f1254636ddd46fca0e0ac6c141091bf72552c53131f3e",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "SECURITY.md",
+      "sha256": "87715bb85d820beb6179cf5294fc7ed461cb234657df810031c0eb527fb00b62",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "bin/agent-stack.js",
+      "sha256": "5706481b1f1266174eeefebc1f2fc5ff6f30f77b395289490b182e95d5c97b00",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "package.json",
+      "sha256": "42cbab0ed2a00d1bc950ada5688221b8d291bf82f1a681df6e3994935066ceed",
+      "sameAsReviewedDoctrine": false
+    },
+    {
+      "path": "plugins/agent-stack/.claude-plugin/plugin.json",
+      "sha256": "678aa34cceef4ff51bedad7f718baa901af911fdf33afd9bf22b091ac1461a3f",
+      "sameAsReviewedDoctrine": false
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/references/multi-agent.md",
+      "sha256": "e0c3175aa8858de1ba8117c613d2e201e32ff581453078c927a3ebddcff6c0bb",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/references/otel-genai.md",
+      "sha256": "d2cadad0a6a0632d234a51621b136f3001f5a7422e256f84d261d7c0ff1660d5",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/references/statistics.md",
+      "sha256": "77cd03cc4f258a097888f81b817e8c9487b4d777d9ac092f76fe48de8de692ca",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/SKILL.md",
+      "sha256": "8f533ce3e320e0416163a39a98207cd347a0d22332c4072afa1d628f77204b8b",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/audit.md",
+      "sha256": "b91b966250438edd91b87c392113404d830b06301518d9120246321c88d9a1aa",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/context-lifecycle.md",
+      "sha256": "f682b0d4894f7f58baf4f39e622ed1bf8bc773d3c76c552a09aecab4c1e23fef",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/instruction-budget.md",
+      "sha256": "3962dbe1643ca1f867575786597cfeea93a28b1d0cc940811020e2d4b49fc6a6",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/layers.md",
+      "sha256": "e2f4fd73424b0896c493616dbb98e7bd973eaad3b8365b3c1b31964fbac46d89",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/pi-sdk.md",
+      "sha256": "b7eb9375934887ed5f85e597e80c5f3d83c02c37cd585aff13960104a31ddcd5",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/pi.md",
+      "sha256": "684616062c9d109ff13eed83a5b9a10544040f1ee22790ee70e8d8156b286c08",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/system-prompt.md",
+      "sha256": "f8577ebfeb8ae7448b36b5cd7fcca1ef4ab7065d1aef9baf6f7946eb5ea2905a",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/techniques.md",
+      "sha256": "4198c8d2d8a0d6da2f8ba7f2370310578167819abbb9bd351a722491ee8f31e8",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/tools.md",
+      "sha256": "572822bc05a710019fc5454f2ddb3750537dbf3667fc73fad0c8bcdcd0f1346a",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/workbench-contracts.md",
+      "sha256": "a17af9cc84dcc8661085ebb606b09ac08ebe39f6c1fc436ab547ac21747bca80",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/scripts/audit_agent.py",
+      "sha256": "f0193565dd3851d3ea779a05f018099b9f2bca3eccacc370b9a43048baba656a",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/scripts/instruction_budget.py",
+      "sha256": "0f88ff1764cc5a87a2b523126c871e71af750bc29128e05b97f9e6955db999ab",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/SKILL.md",
+      "sha256": "003866190df99a02ea862860050c03c2ce8971479ced5ba75bfe42057a0ba4e0",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/a2a.md",
+      "sha256": "e39c4c6d1f19b39e09720c40a09a50e8f2a59824aeb83c5336443f8b709a6c34",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/gateway.md",
+      "sha256": "e74e8585eb6a1fba08af5b419a67c41f38689a29886a89dc149ef3465b9319a0",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/mcp-scale.md",
+      "sha256": "c69b0bfd73e189776fcc6ecdd7d3e88964073ab19a66ffcf79aab579b341a892",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/mcp-ship.md",
+      "sha256": "2918ca0beed94ca7f7e34bb30e2609495c421121d3a94a0e43e743fc76b644a2",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/mcp.md",
+      "sha256": "9b427194659408ae7f521ea689b8b23cbfd216f918f313cc961dce360b7e80b4",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/registry.md",
+      "sha256": "648e7649369b9aa5dac5f424c9a4373292e5d73c90d9466b58171934e5c7b51f",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/SKILL.md",
+      "sha256": "3e70a38f204edc6822814335dc6dfc4e8c3b825117137239da46b96fdd60bfec",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/context-engineering.md",
+      "sha256": "57811d800b75f25a5a242505e8e6be2b33775fb156893ebec51af5ea18fcfc1e",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/governance.md",
+      "sha256": "bb2e3389779a56d4ace610bb14b5df05ace7fa6efc1e5810b0ae929f7050e435",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/graph-engineering.md",
+      "sha256": "c594d1b0a4813f0d47e60a26e2b7905a964c3017556a400fcbd534f390811e70",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/kv-cache.md",
+      "sha256": "b8ae7c5ba9531ebb02ded2e403a8fcb4ae2b944940076a4f15f966d05f6caeae",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/llm-proxy-billing.md",
+      "sha256": "84c2144d6a5a917c042c5e31ecfacfe6238c60870da567abc00bd770e3bccccd",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/memory-architecture.md",
+      "sha256": "860466e4e54d0116063146401c3df54a25bbf07cdfba18b63d0fb0e06a285f1a",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/memory-landscape.md",
+      "sha256": "b297b121ebb07c50579f929f0e5361598625b024dbf49653b82e12b1dc0b4f4d",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/memory-lifecycle.md",
+      "sha256": "6c7252553bc6308e6f5ec88bb802a62e4bacd98f7946ec5b2e82fef8c1aa0073",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/patterns.md",
+      "sha256": "a239bc4549339b049b54b154f88536bed68256ffffdcdc4a0f41cf7185eeb360",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/pipeline.md",
+      "sha256": "89f160068715136c6f7793b305f78371831ebef9527cdb481e15b5af9a1af40b",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/provider-lifecycle.md",
+      "sha256": "d9b4462baf1c805daf05612d7ee8556ac5e666f72de62609666f1ee469294de3",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/runtime.md",
+      "sha256": "4c49fe64e16256fd78529e4a557782c5ccd9be8af8a5703bb92a773bdea12d70",
+      "sameAsReviewedDoctrine": true
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/SKILL.md",
+      "sha256": "8a80583133707ea69c9e7c0963a68f74b138b74e5148e31ceeea12097af32aba",
+      "sameAsReviewedDoctrine": true
+    }
+  ],
+  "payloadCount": 44,
+  "changedPayloadSinceReviewedDoctrine": [
+    "CHANGELOG.md",
+    "package.json",
+    "plugins/agent-stack/.claude-plugin/plugin.json"
+  ],
+  "fullOwnerGate": "NOT_RUN_QUEUED_AFTER_TP_AND_SEO",
+  "hostedPR": "NOT_RUN",
+  "release": "NOT_RUN",
+  "installedApply": "OUT_OF_SCOPE",
+  "notes": [
+    "Metadata homes package/marketplace/plugin/card aligned, historical changelog untouched below new section.",
+    "Exact accepted main will run local full npm test chain before annotated tag."
+  ]
+}