@catheadowl/dsh-eval 0.2.0 → 0.2.1

package/CHANGELOG.md

@@ -0,0 +1,57 @@
+---
+description: Release history of @catheadowl/dsh-eval — one entry per published version, following Keep a Changelog conventions
+---
+
+# Changelog
+
+All notable changes to `@catheadowl/dsh-eval` are documented here. Versions
+follow [Semantic Versioning](https://semver.org/); entries follow
+[Keep a Changelog](https://keepachangelog.com/) conventions.
+
+## [0.2.1] — 2026-09-09
+
+### Added
+
+- Subagent dispatch observability: `subagentChildren` projection plus
+  `subagentDispatched` / `subagentCompleted` matchers.
+- Multi-turn followups: cross-turn asynchronous driving for review
+  experiments, with dispatch/completion count assertions.
+
+### Changed
+
+- `dsh-review` boots a **blank environment by default**: staged out-of-tree
+  plugin rows are disabled via overlay, so host-profile gates/plugins can no
+  longer steer or crash a reviewer. Pass `--keep-plugin-rows` (or executor
+  option `keepPluginRows`) to opt back in deliberately.
+
+### Fixed
+
+- Review artifacts capture the reviewer's **answer**, not the last message:
+  conclusions use the last assistant text before any plugin-sourced injection
+  (trace-derived); the raw final message is kept as `run-N.stdout.txt` when it
+  diverges, and each report run cites its transcript.
+
+## [0.2.0] — 2026-09-06
+
+### Changed
+
+- **BREAKING (0.x minor)**: public API split into two tiers. The root entry
+  is now the SDK tier only (assertion DSL, step builders, `runEvalCase`,
+  `defineReviewExperiment`); low-level sandbox/overlay/trace primitives and
+  review execution moved to the `./experimental` escape-hatch subpath.
+- `exports` map added — deep path imports are now mechanically blocked.
+
+## [0.1.0] — 2026-09-05
+
+### Added
+
+- Initial public release: case runner over dsh headless runs,
+  session-trace assertions, and a scripted mock-LLM layer for plugin intent
+  tests.
+- CLI binaries `dsh-eval` (run cases) and `dsh-review` (LLM output review);
+  `@deepseek-ai/dsh-llm` is a peerDependency provided by the host
+  ecosystem.
+
+[0.2.1]: https://github.com/CatheadOwl/dsh-eval/releases/tag/v0.2.1
+[0.2.0]: https://github.com/CatheadOwl/dsh-eval/releases/tag/v0.2.0
+[0.1.0]: https://github.com/CatheadOwl/dsh-eval/releases/tag/v0.1.0

package/README.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority; after
 # editing either side, bring the other along and re-record with:
 #   node scripts/verify-readme-i18n.mjs --write
-README.md: 063d8659bc607bbefcc4a09aefa3c5b9280628ae249e150e41952a6b9f83239e
-README.zh.md: 4938e021edcf5fb49232c96b780ed121d51e00abe4a89b3583a17ecbbb2f14b1
+README.md: 94e4e6c1172544c94c65329abf8dc6db81c0ba8030c354e1c1e8e64f8bc8276b
+README.zh.md: a1f45726e695fe46931a2988c7293d53c61b3c275dc92e039e707b315f72fc60

package/README.md

@@ -1,7 +1,3 @@
----
-description: '@catheadowl/dsh-eval — a dsh-native agent evaluation layer for plugin authors: behavior cases run against real headless dsh traces, review experiments test whether fresh models understand plugin outputs'
----
-
 # @catheadowl/dsh-eval
 
 English | [中文](README.zh.md)
@@ -115,10 +111,11 @@ Unknown keys fail loudly (typos never degrade silently). The `disableRows` seman
 | Doc | Topic |
 |---|---|
 | [host-wiring](docs/host-wiring.md) | peer wiring (incl. the npm antique-peer trap), building the CLI, profiles, credentials, spawn requirements |
-| [review](docs/review.md) | comprehension review: experiment definition, sterile profile, artifacts, the six review rules |
+| [review](docs/review.md) | comprehension review: experiment definition, blank-environment reviewers, artifacts, the six review rules |
 | [matchers](docs/matchers.md) | the full trace-matcher and mock-helper set (tool face / text face / model-visible face) |
 | [disablerows](docs/disablerows.md) | `disableRows` and the turn-close gate boundary contract |
 | [rowconfig](docs/rowconfig.md) | the `rowConfig` per-row config override contract (whole-segment replacement, restate needed keys) |
+| [cross-turn](docs/cross-turn.md) | the `followups` cross-turn async driving contract (driver-row swap, subagent settle wait, single-cursor mock orchestration, bounded-redispatch assertions) |
 | [intent-cases](docs/intent-cases.md) | real intent-case spec: when to write one, assertion face, guards, CI semantics |
 | [report](docs/report.md) | machine-readable report structure (`--format json` / `--report`) |
 | [known-issues](docs/known-issues.md) | known issues and workarounds (e.g. REQUEST_EXTENSION in staged homes) |

package/README.zh.md

@@ -1,7 +1,3 @@
----
-description: '@catheadowl/dsh-eval 中文主页——dsh-native agent 评测层：behavior case 跑真实 headless dsh trace，review experiment 测 fresh model 能否理解插件输出'
----
-
 # @catheadowl/dsh-eval
 
 [English](README.md) | 中文
@@ -115,10 +111,11 @@ export default {
 | 文档 | 主题 |
 |---|---|
 | [host-wiring](docs/host-wiring.md) | peer 接线（含 npm 古董 peer 坑）、构建 CLI、profile、凭证、spawn 要求 |
-| [review](docs/review.md) | comprehension review：实验定义、sterile profile、产物、六条评审规则 |
+| [review](docs/review.md) | comprehension review：实验定义、空白环境 reviewer、产物、六条评审规则 |
 | [matchers](docs/matchers.md) | trace matcher 与 mock helper 全集（工具面 / 文本面 / 模型可见面） |
 | [disablerows](docs/disablerows.md) | `disableRows` 与 turn-close 门禁边界契约 |
 | [rowconfig](docs/rowconfig.md) | `rowConfig` 行 config 覆写契约（整段替换、重述所需键） |
+| [cross-turn](docs/cross-turn.md) | `followups` 跨轮异步驱动契约（driver 行换装、子 agent settle 等待、mock 单 cursor 编排、有界重派断言） |
 | [intent-cases](docs/intent-cases.md) | real 意图 case 规约：何时写、断言面、守卫、CI 语义 |
 | [report](docs/report.md) | 机器可读报告（`--format json` / `--report`）结构 |
 | [known-issues](docs/known-issues.md) | 已知问题与规避（如 staged home 的 REQUEST_EXTENSION） |