Reject incomplete temporary-maker source discovery

docs/HANDOFF.md

@@ -1,3 +1,13 @@
+# FW01 source discovery correction — 2026-10-13
+
+The [source packet and receipt](evidence/runs/2026-10-13-fw01-discovery/README.md)
+records actual healthy/readable/unreadable baseline evidence and the bounded
+permission-error fix. New focused controls pass; source payload and version stay
+unchanged. Next: independent review, serialized owner aggregate and genuine normal
+PR admission. No full-scan/runtime completeness, publication or installation claim.
+
+---
+
 # FW00 referral dispositions — 2026-10-13
 
 [Current packet](evidence/runs/2026-10-13-fw00-referrals.md) closes only B-160's

docs/evidence/runs/2026-10-13-fw01-discovery/README.md

@@ -0,0 +1,79 @@
+# FW01 source discovery completeness — agent-stack
+
+## Frozen packet and baseline
+
+Owner source: [63e155d](https://github.com/ssheleg/agent-stack/commit/63e155d9c1ca9df789535926acc3546f33954e3b).
+Version remains 0.28.2; this packet changes owner validation and evidence only.
+The [actual whole-validator baseline](baseline.json) executes `python3 -B test/validate.py`
+in an immutable Git-source fixture: healthy exits 0, readable unauthorized maker exits
+1, then its parent directory is made unreadable (OS PermissionError independently
+confirmed) and the same validator exits 0. This proves a discovery omission, beyond
+an extracted-helper observation. No complete owner suite or native runtime was run.
+
+The upstream family finding is
+[the bounded central receipt](https://github.com/ssheleg/sshlg-skills/blob/38b69b1d3c022f45e4ac3464c5b83f4b96047852/docs/evidence/completion-1013/fw01-discovery-followup.json).
+SEO's same discovery seam was corrected at
+[76d25cd](https://github.com/ssheleg/seo-aeo-audit/commit/76d25cd5c283bce7c4fdd25c7c42fee7038fe1b2).
+Its lexical/exemption policy is not copied into this owner.
+
+## Contract and acceptance
+
+- Keep this owner's scanner predicate, skip directories, lexical matching and exact
+  exemptions unchanged. Only discovery error handling is in scope.
+- `os.walk` directory discovery failure must add an ordinary owner guard failure
+  identifying relative source path and error class. It must not become an empty,
+  apparently complete source set. Keep running readable checks to collect evidence.
+- Existing unreadable source-file behavior already exits nonzero; no success claim
+  or broad parser rewrite is introduced. Excluded dependency directories remain
+  excluded, including when unreadable. No symlink traversal policy is changed.
+- Add focused helper admission controls and actual whole-validator healthy,
+  readable allocation, unreadable directory, and restoration controls. Confirm the
+  permission plant actually fails under this user; root privilege must not produce
+  a fake passing permission test. Restore permissions in finally; all fixtures go
+  through the owner's residue ledger.
+- Preserve versions and every published payload file. Verify npm pack path/digest
+  equality against the accepted baseline. Independent peer review and normal PR
+  admission follow; broader owner gate waits for the coordinator's serial slot.
+
+Implementation is the next step. No scanner, runtime, board closure, publication
+or installation is claimed by this source-first packet.
+
+The first baseline harness attempt failed its own diagnostic-stream assertion:
+this owner emits failure diagnostics on stderr, while make-skill uses stdout.
+The actual readable plant already returned 1. The harness was corrected to inspect
+both streams, then all three baseline conditions above were rerun successfully.
+This was a harness failure, not a successful owner check or an implementation fix.
+
+## Implemented and focused verification
+
+Only the scanner's `os.walk` call gained an error callback (relative path and error
+class); the owner's lexical rules, exemptions and skip set are unchanged. The new
+`test/audit_regressions/fw01_discovery.py` passes 9 controls: helper healthy,
+permission refusal, restoration, excluded dependency and absent-root admission;
+actual whole-validator healthy, readable unauthorized maker, unreadable directory,
+and restoration. Permission tests explicitly prove OS denial and restore modes in
+finally. Both owned fixture trees were removed after success.
+
+The [pack comparison](payload-parity.json) checks all 44 current npm payload
+files (609,718 bytes) against accepted baseline Git blobs. Only test/docs
+paths change; package selection and version are unchanged. Registry metadata confirms
+that version exists; this is not a fresh immutable-tarball readback. No new release
+or installed update is needed for this excluded validation/evidence source change.
+
+## Independent review and owner aggregate
+
+[Root's independent source acceptance](https://github.com/ssheleg/sshlg-skills/blob/85d1758e02cbf1759be861a7af2108d267c339a3/docs/evidence/completion-1013/fw01-sibling-peer.json) binds the implementation above
+and independently replays all 9 controls. It accepts directory discovery refusal
+within the documented scanner scope, without a filesystem race or symlink proof.
+
+The exact implementation commit `bc0f66387850f30674ae14636f7e40d98e029d2b` then passed the complete owner
+`npm test` command, exit 0, in 12.04 seconds. The
+[structured result](owner-full.json) binds source, command and log digest;
+[full output](owner-full.log.gz) preserves every group. Minimum observed free
+space was 1533267968 bytes. The watchdog would stop only this owned
+process group below 300 MiB; it did not fire. Owned scratch/cache trees were removed.
+No provider, native-host or model behavior was exercised.
+
+Next: commit this documentation freeze, promote the existing draft PR once, and
+require genuine eligible PR jobs before normal integration. Hosted checks and merge
+remain pending; no release or installed update is needed for this excluded source.

docs/evidence/runs/2026-10-13-fw01-discovery/baseline.json

@@ -0,0 +1,25 @@
+{
+  "source": "63e155d9c1ca9df789535926acc3546f33954e3b",
+  "scope": "actual whole validator, no full owner suite",
+  "permission_plant_confirmed": true,
+  "results": [
+    {
+      "case": "healthy",
+      "exit": 0,
+      "stdout": "  unlooked: front matter vs a strict reader \u2014 no sshlg-skills umbrella above this checkout\nOK: agent-stack structurally valid (17 checks = 13 named + 4 per-skill, 4 skill(s), v0.28.2)\nresidue: this run left nothing \u2014 0 temp tree(s) created, 0 removed\n",
+      "stderr": ""
+    },
+    {
+      "case": "readable unauthorized allocation",
+      "exit": 1,
+      "stdout": "  unlooked: front matter vs a strict reader \u2014 no sshlg-skills umbrella above this checkout\nresidue: this run left nothing \u2014 0 temp tree(s) created, 0 removed\n",
+      "stderr": "FAIL: 1 problem(s)\n  - scripts/fw01_hidden/leak.py: creates a temp tree outside the residue ledger \u2014 'tempfile.mkdtemp(' at line 2. A shipped script cannot reach the suite's ledger: make the tree with a context manager that removes it on every path and give the file an exact-count entry, with its reason, in `TEMP_TREE_ALLOWED` \u2014 or move the work into the suite\n"
+    },
+    {
+      "case": "unreadable source directory",
+      "exit": 0,
+      "stdout": "  unlooked: front matter vs a strict reader \u2014 no sshlg-skills umbrella above this checkout\nOK: agent-stack structurally valid (17 checks = 13 named + 4 per-skill, 4 skill(s), v0.28.2)\nresidue: this run left nothing \u2014 0 temp tree(s) created, 0 removed\n",
+      "stderr": ""
+    }
+  ]
+}

docs/evidence/runs/2026-10-13-fw01-discovery/initial-harness-failure.json

@@ -0,0 +1,8 @@
+{
+  "source": "63e155d9c1ca9df789535926acc3546f33954e3b",
+  "harness_exit": 1,
+  "actual_readable_plant_exit": 1,
+  "wrong_assumption": "diagnostic must be in stdout",
+  "correction": "inspect stdout plus stderr; retain genuine nonzero condition",
+  "rerun_receipt": "baseline.json"
+}

docs/evidence/runs/2026-10-13-fw01-discovery/owner-full.json

@@ -0,0 +1,17 @@
+{
+  "owner": "agent-stack-fw01-scan",
+  "source": "bc0f66387850f30674ae14636f7e40d98e029d2b",
+  "command": "npm test",
+  "exit_code": 0,
+  "seconds": 12.04,
+  "minimum_free_bytes": 1533267968,
+  "disk_stopped": false,
+  "log": "owner-full.log.gz",
+  "log_sha256": "c30bf7ff7c13b4fbb04e87ffb77b4dc3147d38ba3513ba6a9ab50615d6e84d09",
+  "scratch_removed": true,
+  "independent_review": "https://github.com/ssheleg/sshlg-skills/blob/85d1758e02cbf1759be861a7af2108d267c339a3/docs/evidence/completion-1013/fw01-sibling-peer.json",
+  "scratch_remaining_before_owned_cleanup": [
+    "data-gym-cache",
+    "node-compile-cache"
+  ]
+}

docs/evidence/runs/2026-10-13-fw01-discovery/payload-parity.json

@@ -0,0 +1,240 @@
+{
+  "baseline": "63e155d9c1ca9df789535926acc3546f33954e3b",
+  "package_version": "0.28.2",
+  "registry_observation": {
+    "version": "0.28.2",
+    "gitHead": "2b3d769775e368d0a1e6d0f20ea095ae24fbbd54"
+  },
+  "current_pack_files": 44,
+  "current_pack_bytes": 609718,
+  "all_current_pack_blobs_equal_baseline": true,
+  "changed_tracked_scope": [
+    "docs/evidence/runs/2026-10-13-fw01-discovery/README.md",
+    "docs/evidence/runs/2026-10-13-fw01-discovery/baseline.json",
+    "docs/evidence/runs/2026-10-13-fw01-discovery/initial-harness-failure.json",
+    "test/validate.py"
+  ],
+  "files": [
+    {
+      "path": "CHANGELOG.md",
+      "sha256": "0fce6df83bb875438bbcd2dd1609665ae19348284405c1b2c7b075514a375e7e",
+      "size": 111515
+    },
+    {
+      "path": "LICENSE",
+      "sha256": "826df230e15df9b208c1e08c42ef3b764b21c4abec5a91a38976748e92a82e9d",
+      "size": 1064
+    },
+    {
+      "path": "README.md",
+      "sha256": "49953599fcd54dfcba0f1254636ddd46fca0e0ac6c141091bf72552c53131f3e",
+      "size": 14601
+    },
+    {
+      "path": "SECURITY.md",
+      "sha256": "87715bb85d820beb6179cf5294fc7ed461cb234657df810031c0eb527fb00b62",
+      "size": 3071
+    },
+    {
+      "path": "bin/agent-stack.js",
+      "sha256": "5706481b1f1266174eeefebc1f2fc5ff6f30f77b395289490b182e95d5c97b00",
+      "size": 7391
+    },
+    {
+      "path": "package.json",
+      "sha256": "42cbab0ed2a00d1bc950ada5688221b8d291bf82f1a681df6e3994935066ceed",
+      "size": 1323
+    },
+    {
+      "path": "plugins/agent-stack/.claude-plugin/plugin.json",
+      "sha256": "678aa34cceef4ff51bedad7f718baa901af911fdf33afd9bf22b091ac1461a3f",
+      "size": 1081
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/references/multi-agent.md",
+      "sha256": "e0c3175aa8858de1ba8117c613d2e201e32ff581453078c927a3ebddcff6c0bb",
+      "size": 5473
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/references/otel-genai.md",
+      "sha256": "d2cadad0a6a0632d234a51621b136f3001f5a7422e256f84d261d7c0ff1660d5",
+      "size": 13131
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/references/statistics.md",
+      "sha256": "77cd03cc4f258a097888f81b817e8c9487b4d777d9ac092f76fe48de8de692ca",
+      "size": 13853
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-evals/SKILL.md",
+      "sha256": "8f533ce3e320e0416163a39a98207cd347a0d22332c4072afa1d628f77204b8b",
+      "size": 20773
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/audit.md",
+      "sha256": "b91b966250438edd91b87c392113404d830b06301518d9120246321c88d9a1aa",
+      "size": 11484
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/context-lifecycle.md",
+      "sha256": "f682b0d4894f7f58baf4f39e622ed1bf8bc773d3c76c552a09aecab4c1e23fef",
+      "size": 9055
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/instruction-budget.md",
+      "sha256": "3962dbe1643ca1f867575786597cfeea93a28b1d0cc940811020e2d4b49fc6a6",
+      "size": 7026
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/layers.md",
+      "sha256": "e2f4fd73424b0896c493616dbb98e7bd973eaad3b8365b3c1b31964fbac46d89",
+      "size": 6161
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/pi-sdk.md",
+      "sha256": "b7eb9375934887ed5f85e597e80c5f3d83c02c37cd585aff13960104a31ddcd5",
+      "size": 14100
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/pi.md",
+      "sha256": "684616062c9d109ff13eed83a5b9a10544040f1ee22790ee70e8d8156b286c08",
+      "size": 12889
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/system-prompt.md",
+      "sha256": "f8577ebfeb8ae7448b36b5cd7fcca1ef4ab7065d1aef9baf6f7946eb5ea2905a",
+      "size": 7712
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/techniques.md",
+      "sha256": "4198c8d2d8a0d6da2f8ba7f2370310578167819abbb9bd351a722491ee8f31e8",
+      "size": 7968
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/tools.md",
+      "sha256": "572822bc05a710019fc5454f2ddb3750537dbf3667fc73fad0c8bcdcd0f1346a",
+      "size": 10351
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/references/workbench-contracts.md",
+      "sha256": "a17af9cc84dcc8661085ebb606b09ac08ebe39f6c1fc436ab547ac21747bca80",
+      "size": 10349
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/scripts/audit_agent.py",
+      "sha256": "f0193565dd3851d3ea779a05f018099b9f2bca3eccacc370b9a43048baba656a",
+      "size": 19276
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/scripts/instruction_budget.py",
+      "sha256": "0f88ff1764cc5a87a2b523126c871e71af750bc29128e05b97f9e6955db999ab",
+      "size": 5863
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-harness/SKILL.md",
+      "sha256": "003866190df99a02ea862860050c03c2ce8971479ced5ba75bfe42057a0ba4e0",
+      "size": 14571
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/a2a.md",
+      "sha256": "e39c4c6d1f19b39e09720c40a09a50e8f2a59824aeb83c5336443f8b709a6c34",
+      "size": 11113
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/gateway.md",
+      "sha256": "e74e8585eb6a1fba08af5b419a67c41f38689a29886a89dc149ef3465b9319a0",
+      "size": 22093
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/mcp-scale.md",
+      "sha256": "c69b0bfd73e189776fcc6ecdd7d3e88964073ab19a66ffcf79aab579b341a892",
+      "size": 8841
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/mcp-ship.md",
+      "sha256": "2918ca0beed94ca7f7e34bb30e2609495c421121d3a94a0e43e743fc76b644a2",
+      "size": 7099
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/mcp.md",
+      "sha256": "9b427194659408ae7f521ea689b8b23cbfd216f918f313cc961dce360b7e80b4",
+      "size": 14716
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/references/registry.md",
+      "sha256": "648e7649369b9aa5dac5f424c9a4373292e5d73c90d9466b58171934e5c7b51f",
+      "size": 8377
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-interop/SKILL.md",
+      "sha256": "3e70a38f204edc6822814335dc6dfc4e8c3b825117137239da46b96fdd60bfec",
+      "size": 15786
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/context-engineering.md",
+      "sha256": "57811d800b75f25a5a242505e8e6be2b33775fb156893ebec51af5ea18fcfc1e",
+      "size": 11541
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/governance.md",
+      "sha256": "bb2e3389779a56d4ace610bb14b5df05ace7fa6efc1e5810b0ae929f7050e435",
+      "size": 9487
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/graph-engineering.md",
+      "sha256": "c594d1b0a4813f0d47e60a26e2b7905a964c3017556a400fcbd534f390811e70",
+      "size": 28552
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/kv-cache.md",
+      "sha256": "b8ae7c5ba9531ebb02ded2e403a8fcb4ae2b944940076a4f15f966d05f6caeae",
+      "size": 9911
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/llm-proxy-billing.md",
+      "sha256": "84c2144d6a5a917c042c5e31ecfacfe6238c60870da567abc00bd770e3bccccd",
+      "size": 19674
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/memory-architecture.md",
+      "sha256": "860466e4e54d0116063146401c3df54a25bbf07cdfba18b63d0fb0e06a285f1a",
+      "size": 20225
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/memory-landscape.md",
+      "sha256": "b297b121ebb07c50579f929f0e5361598625b024dbf49653b82e12b1dc0b4f4d",
+      "size": 8296
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/memory-lifecycle.md",
+      "sha256": "6c7252553bc6308e6f5ec88bb802a62e4bacd98f7946ec5b2e82fef8c1aa0073",
+      "size": 11365
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/patterns.md",
+      "sha256": "a239bc4549339b049b54b154f88536bed68256ffffdcdc4a0f41cf7185eeb360",
+      "size": 25555
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/pipeline.md",
+      "sha256": "89f160068715136c6f7793b305f78371831ebef9527cdb481e15b5af9a1af40b",
+      "size": 4549
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/provider-lifecycle.md",
+      "sha256": "d9b4462baf1c805daf05612d7ee8556ac5e666f72de62609666f1ee469294de3",
+      "size": 8966
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/references/runtime.md",
+      "sha256": "4c49fe64e16256fd78529e4a557782c5ccd9be8af8a5703bb92a773bdea12d70",
+      "size": 13205
+    },
+    {
+      "path": "plugins/agent-stack/skills/agent-orchestrator/SKILL.md",
+      "sha256": "8a80583133707ea69c9e7c0963a68f74b138b74e5148e31ceeea12097af32aba",
+      "size": 20286
+    }
+  ],
+  "scope": "source payload parity; registry metadata read, immutable tarball not re-downloaded"
+}