fix(release): synchronize canonical verification ledger for 0.25.2

.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": "0.25.1",
+      "version": "0.25.2",
       "author": {
         "name": "ssheleg",
         "url": "https://x.com/sshlg93"

CHANGELOG.md

@@ -1,3 +1,12 @@
+## 0.25.2 — the verification ledger follows the harness release
+
+- Add the current harness requirements to `docs/evidence/verification.md` in
+  the same commit as this release. The umbrella correctly refused 0.25.1 because
+  the member ledger still named 0.24.3; the dedicated research handoff did not
+  satisfy the canonical ledger contract.
+- Keep the published 0.25.1 evidence and the unreleased 0.25.0 tag incident
+  explicit. No runtime, reference behavior or dependency changes.
+
 ## 0.25.1 — workbench harness contracts, with a bounded ECC transfer
 
 - Distinguish a kernel harness from the operating layer around an existing agent.

SKILL-CARD.md

@@ -5,7 +5,7 @@
 | Field | Value |
 |---|---|
 | Pack | `agent-stack` |
-| Version | `0.25.1` |
+| Version | `0.25.2` |
 | Skills | `agent-orchestrator`, `agent-evals`, `agent-interop`, `agent-harness` |
 | License | MIT |
 | Source | https://github.com/ssheleg/agent-stack |

docs/evidence/research/2026-09-21-ecc-handoff.md

@@ -104,3 +104,11 @@ and submodule pin to 0.25.1 / 0b76d1a, then refresh installed skills and separat
 check the host's loaded receipt. Those operations belong to the parent and are
 not implied by this member publication. The tested research/reference contract
 has no new runtime or behavioral-model evaluation result.
+
+## Canonical ledger follow-up — 0.25.2
+
+The umbrella gate rejected the 0.25.1 member because its canonical verification
+ledger still named 0.24.3. The research handoff was insufficient. Version 0.25.2
+adds the matching ledger heading in the release commit without weakening the
+guard or changing the reviewed reference. The parent must use 0.25.2 after its
+exact registry verification; 0.25.1 evidence above remains historical.

docs/evidence/verification.md

@@ -10,6 +10,34 @@ This file exists because its absence read as zero exposure. `sshlg-skills` board
 
 ---
 
+## Shipped state — v0.25.2 (2026-09-21)
+
+Ledger correction for the ECC workbench-contract transfer. This release changes
+version metadata and records the requirements in their canonical ledger; the
+reference bytes are the same as the independently reviewed and published 0.25.1.
+The rows distinguish published evidence for those bytes from candidate validation.
+
+| REQ | What ships | How it was confirmed | Confirmed |
+|---|---|---|---|
+| ECC-1 | A pinned, selective ECC adoption record | [Research ledger](research/2026-09-21-ecc-harness.md) records commit 2b6e839771e53096d8451a213d40dc64ec8acac0, read scope, adaptation decisions and 15 SHA-256 digests; all 15 compared successfully to the frozen source | **observed** |
+| ECC-2 | Explicit kernel versus workbench harness boundary | `agent-harness/references/layers.md` and `workbench-contracts.md` distinguish host runtime and permissions from family work contracts; coordinating agent independently reviewed the change before PR #33 | **observed** — semantic review, not a model benchmark |
+| ECC-3 | Install ownership, bounded hooks, durable handoff and artifact-bound evaluation contracts | `npm test` and both strict Claude manifest checks passed for the published 0.25.1 reference; make-skill house audit returned 0 GAP / 19 PASS | **observed** — structural checks, not behavioral improvement |
+| ECC-4 | No new runtime, hooks or dependencies | Registry 0.25.1 tarball contains 40 files, all byte-identical to commit 0b76d1afc19c778db0228b1c0f383c2c1d6a9949; no research/private machine files; this patch does not change the skill tree | **observed** |
+| ECC-5 | Current canonical ledger in the release commit | Package, plugin, marketplace, skill card and this heading all name 0.25.2; `python3 test/validate.py` checks version and ledger validity | **observed** — `npm test` EXIT=0 on the 0.25.2 candidate, including `test/validate.py` |
+
+Published predecessor evidence and exact integrity are in the
+[release handoff](research/2026-09-21-ecc-handoff.md). v0.25.0 remains unreleased;
+its cancelled workflow is not a successful publication. Host installation and
+loaded receipts belong to the parent rollout, not these package checks.
+
+## Shipped state — v0.25.1 (2026-09-21)
+
+Published from 0b76d1afc19c778db0228b1c0f383c2c1d6a9949. Release workflow
+35584641319 passed all four jobs; npm returned version 0.25.1 and the same
+`gitHead`. A fresh checkout passed `npm test`; all 40 registry tarball files
+matched it byte-for-byte. The canonical ledger was updated only in 0.25.2,
+after the umbrella gate exposed that this section was missing.
+
 ## Shipped state — v0.24.3 (2026-09-14)
 
 Hygiene from the 2026-09-13 family audit (HK-11).

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ssheleg/agent-stack",
-  "version": "0.25.1",
+  "version": "0.25.2",
   "scripts": {
     "test": "python3 test/validate.py && python3 test/plant_guard_test.py && node test/installer_test.js && npm run test:audit",
     "test:audit": "for t in test/audit_regressions/*.py; do python3 \"$t\" || exit 1; done"

plugins/agent-stack/.claude-plugin/plugin.json

@@ -3,7 +3,7 @@
   "name": "agent-stack",
   "displayName": "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": "0.25.1",
+  "version": "0.25.2",
   "author": {
     "name": "ssheleg",
     "url": "https://x.com/sshlg93"