coding-agent-harness 1.0.8 → 1.1.0

package/presets/release-closeout/templates/task_plan.append.md

@@ -17,8 +17,8 @@ Run these preset entrypoints from the target root. The task scaffold only record
 
 | Output | Owner |
 | --- | --- |
-| `coding-agent-harness/governance/releases/{{release}}/INDEX.md` | `presets/release-closeout` |
-| `coding-agent-harness/governance/releases/{{release}}/task-aggregate.json` | `presets/release-closeout` |
-| `coding-agent-harness/governance/releases/{{release}}/task-archive-plan.md` | `presets/release-closeout` |
-| `coding-agent-harness/governance/releases/{{release}}/public-summary.md` | `presets/release-closeout` |
-| `coding-agent-harness/governance/releases/{{release}}/public-redaction-report.json` | `presets/release-closeout` |
+| `{{paths.governanceRoot}}/releases/{{release}}/INDEX.md` | `presets/release-closeout` |
+| `{{paths.governanceRoot}}/releases/{{release}}/task-aggregate.json` | `presets/release-closeout` |
+| `{{paths.governanceRoot}}/releases/{{release}}/task-archive-plan.md` | `presets/release-closeout` |
+| `{{paths.governanceRoot}}/releases/{{release}}/public-summary.md` | `presets/release-closeout` |
+| `{{paths.governanceRoot}}/releases/{{release}}/public-redaction-report.json` | `presets/release-closeout` |

package/presets/standard-task/preset.yaml

@@ -9,7 +9,7 @@ task:
 entrypoints:
   newTask:
     type: template
-    writes: [coding-agent-harness/planning/tasks/**]
+    writes: [{{paths.tasksRoot}}/**]
     audit: true
     templates:
       taskPlanAppend: templates/task_plan.append.md
@@ -27,5 +27,5 @@ audit:
   evidenceFiles: [preset-audit.json]
 writeScopes:
   taskDocs:
-    path: coding-agent-harness/planning/tasks/**
+    path: {{paths.tasksRoot}}/**
     access: write

package/references/agents-md-pattern.md

@@ -70,23 +70,29 @@ AGENTS.md 只包含两类内容：
 项目根目录/
 ├── AGENTS.md        ← concise canonical charter + routing index
 ├── CLAUDE.md        ← 轻量 shim，指向 AGENTS.md
-└── docs/
-    ├── Harness-Ledger.md
-    └── coding-agent-harness/governance/standards/
-        ├── testing-standard.md
-        ├── execution-workflow-standard.md
-        ├── repo-governance-standard.md
-        ├── ci-cd-standard.md
-        ├── long-running-task-standard.md
-        ├── adversarial-review-standard.md
-        ├── review-routing-standard.md
-        ├── engineering-standard.md
-        ├── frontend-standard.md
-        ├── docs-library-standard.md
-        ├── harness-ledger-standard.md
-        ├── regression-ssot-governance.md
-        ├── walkthrough-standard.md
-        └── ...（按需扩展）
+└── coding-agent-harness/
+    ├── governance/
+    │   ├── generated/
+    │   │   ├── Harness-Ledger.md
+    │   │   └── Closeout-Index.md
+    │   └── standards/
+    │       ├── testing-standard.md
+    │       ├── execution-workflow-standard.md
+    │       ├── repo-governance-standard.md
+    │       ├── ci-cd-standard.md
+    │       ├── long-running-task-standard.md
+    │       ├── adversarial-review-standard.md
+    │       ├── review-routing-standard.md
+    │       ├── engineering-standard.md
+    │       ├── frontend-standard.md
+    │       ├── docs-library-standard.md
+    │       ├── harness-ledger-standard.md
+    │       ├── regression-ssot-governance.md
+    │       ├── walkthrough-standard.md
+    │       └── ...（按需扩展）
+    └── planning/
+        ├── tasks/
+        └── modules/
 ```
 
 ### 行数控制

package/references/lessons-governance.md

@@ -22,7 +22,7 @@ coding-agent-harness/governance/
 ├── lessons/                     ← promoted lesson 详情文档
 │   ├── L-2026-05-07-001-xxx.md
 │   └── ...
-└── _archive/                    ← 已合入或废弃的历史详情归档
+└── archive/                     ← 已合入或废弃的历史详情归档
 
 coding-agent-harness/planning/tasks/<task-id>/
 ├── lesson_candidates.md          ← 候选索引和人工路由状态
@@ -146,7 +146,7 @@ pending governance integration -> approved -> merged
 
 ## 归档机制
 
-当 `coding-agent-harness/governance/lessons/` 中已合入或废弃的详情文档过多时，可以把 `merged`、`rejected`、`superseded` 的文档移入 `coding-agent-harness/governance/_archive/`。不要归档仍被 Closeout Index 或 Harness Ledger 当前行引用的文档，除非同步更新引用。
+当 `coding-agent-harness/governance/lessons/` 中已合入或废弃的详情文档过多时，可以把 `merged`、`rejected`、`superseded` 的文档移入 `coding-agent-harness/governance/archive/`。不要归档仍被 Closeout Index 或 Harness Ledger 当前行引用的文档，除非同步更新引用。
 
 ## 人的审批工作流
 

package/references/module-parallel-standard.md

@@ -90,11 +90,8 @@ coding-agent-harness/planning/modules/<key>/
     │   └── review.md (如需要)
     └── ...
 
-task-local walkthrough.md/
-├── <module-key>/
-│   ├── <PREFIX>-NN-walkthrough.md
-│   └── ...
-└── _shared/               ← 跨模块基础设施 task 的 walkthrough
+每个任务的收口记录保存在该任务目录的 `walkthrough.md`。不再维护集中式
+walkthrough 目录树。
 
 coding-agent-harness/planning/modules/_shared/tasks/
 └── YYYY-MM-DD-<name>/     ← 跨模块/基础设施 task
@@ -264,7 +261,7 @@ HARNESS_REQUIRE_GLOBAL_MODULE_SYNC=1 node dist/check-harness.mjs <repo-path>
 
 - 模块所有步骤完成后，状态改为 `completed`
 - 将模块目录移入 `coding-agent-harness/planning/modules/_archive/<key>/`
-- 对应的 walkthrough 保留在 `walkthrough.md/`（不归档）
+- 对应的收口记录保留在各任务目录的 `walkthrough.md`（不归档）
 
 ### 过期检测
 

package/references/ssot-governance.md

@@ -98,8 +98,8 @@ SSoT（Single Source of Truth，单一事实源）保存当前事实。任务生
 | Regression SSoT | active gates | gate 废弃或长期不再运行 | `coding-agent-harness/governance/regression/_archive/` |
 | Cadence Ledger | active cadence checks | cadence 废弃或合并到其他 gate | `coding-agent-harness/governance/regression/_archive/` |
 | Closeout Index | 当前 closeout 索引 | closed/superseded 超过保留窗口 | `coding-agent-harness/governance/archive/legacy-walkthrough/` |
-| Lesson detail docs | pending / approved / superseded 详情文档 | merged/rejected 超过 20 条 | `coding-agent-harness/governance/_archive/` |
-| Harness Ledger | 当前生成索引 | 重新生成前 archive 旧快照 | `coding-agent-harness/governance/_archive/` 或迁移会话 archive |
+| Lesson detail docs | pending / approved / superseded 详情文档 | merged/rejected 超过 20 条 | `coding-agent-harness/governance/archive/` |
+| Harness Ledger | 当前生成索引 | 重新生成前 archive 旧快照 | `coding-agent-harness/governance/archive/` 或迁移会话 archive |
 
 归档不改变 ID，不删除证据文件；Active 文件必须留下 archive index 或指向归档文件。
 

package/references/taskr-gap-analysis.md

@@ -2,7 +2,7 @@
 
 > 对比基准：
 > - **Taskr**: [xerrors/taskr-skill](https://github.com/xerrors/taskr-skill)（SKILL.md + protocol.ts + CLI + Board）
-> - **Coding Agent Harness**: 本仓库 `/Users/lizeyu/Projects/coding-agent-harness`（SKILL.md + 12 references + 30+ templates）
+> - **Coding Agent Harness**: 本仓库根目录（SKILL.md + references/ + 30+ templates）
 >
 > 分析日期：2026-05-21
 > 分析人：小宁（Codex Main）
@@ -558,10 +558,10 @@ Footer 格式：`Taskr: <task-id>`（注意不是旧的 `[taskr:<id>]` subject-l
 
 ```
 Before（当前）:
-  SKILL.md → 12 references → 30+ templates → agent 手动读写 md
+  SKILL.md → references/ → 30+ templates → agent 手动读写 md
 
 After（吸收后）:
-  SKILL.md → 12 references（更新） → 30+ templates（更新，带 frontmatter）
+  SKILL.md → references/（更新） → 30+ templates（更新，带 frontmatter）
     → scripts/validate-tasks.sh ← 新增
     → scripts/harness-doctor.sh ← 新增
     → scripts/reconcile-commits.sh ← 新增

package/run-dist.mjs

@@ -0,0 +1,34 @@
+#!/usr/bin/env node
+import fs from "node:fs";
+import path from "node:path";
+import { spawnSync } from "node:child_process";
+import { fileURLToPath } from "node:url";
+
+const [entrypoint, ...args] = process.argv.slice(2);
+if (!entrypoint) {
+  console.error("Usage: node run-dist.mjs <dist-entrypoint> [...args]");
+  process.exit(1);
+}
+
+const root = path.dirname(fileURLToPath(import.meta.url));
+const buildScript = path.join(root, "scripts/build-dist.mts");
+const distEntrypoint = path.join(root, "dist", entrypoint);
+
+if (fs.existsSync(buildScript)) {
+  const build = spawnSync(process.execPath, [buildScript, "--quiet"], {
+    cwd: root,
+    stdio: "inherit",
+  });
+  if (build.status !== 0) process.exit(build.status ?? 1);
+}
+
+if (!fs.existsSync(distEntrypoint)) {
+  console.error(`Missing dist runtime entrypoint: dist/${entrypoint}`);
+  process.exit(1);
+}
+
+const result = spawnSync(process.execPath, [distEntrypoint, ...args], {
+  cwd: root,
+  stdio: "inherit",
+});
+process.exit(result.status ?? 1);

package/skills/preset-creator/SKILL.md

@@ -41,10 +41,19 @@ Before writing files, answer these in the task notes or your response:
 
 Use the bundled references before writing files:
 
+- `references/structure-aware-paths.md`
 - `references/complex-task-skeleton/`
 - `references/preset-package-skeleton.md`
 
-The complex task skeleton reference contains the base `brief.md`, `task_plan.md`, `execution_strategy.md`, `visual_map.md`, `findings.md`, `lesson_candidates.md`, `progress.md`, `review.md`, `references/INDEX.md`, and `artifacts/INDEX.md` contracts. The preset package skeleton contains a copyable package tree, a complete `preset.yaml`, starter Markdown resources, and the verification checklist. Keep this `SKILL.md` focused on method and judgment; use the references when the task has moved from design to file creation.
+The structure-aware paths reference is mandatory before writing `preset.yaml`;
+it defines `{{paths.*}}`, safe write scopes, drift audit, and template refresh.
+The complex task skeleton reference contains the base `brief.md`, `task_plan.md`,
+`execution_strategy.md`, `visual_map.md`, `findings.md`,
+`lesson_candidates.md`, `progress.md`, `review.md`, `references/INDEX.md`, and
+`artifacts/INDEX.md` contracts. The preset package skeleton contains a copyable
+package tree, a complete `preset.yaml`, starter Markdown resources, and the
+verification checklist. Keep this `SKILL.md` focused on method and judgment; use
+the references when the task has moved from design to file creation.
 
 Use a simple package:
 
@@ -61,6 +70,19 @@ my-preset/
       input-packet.md
 ```
 
+## Structure-Aware Paths
+
+Preset authors must not hard-code Harness runtime paths such as
+`coding-agent-harness/planning/tasks/**`. Repositories may move the Harness
+folder, so every runtime write/read scope must use the target's resolved
+`{{paths.*}}` context.
+
+Before creating or reviewing a preset, read
+`references/structure-aware-paths.md`. For normal task-creating presets, use
+`{{paths.tasksRoot}}/**` in both `entrypoints.newTask.writes` and
+`writeScopes.*.path`. These values must match exactly; do not rely on partial
+overlap.
+
 ## Required Manifest Sections
 
 - `id`: lowercase letters, numbers, and hyphens.
@@ -93,6 +115,11 @@ Supported input types are `text`, `flag`, and `json-file`. Resource `index.type`
 
 Every value in `entrypoints.newTask.writes` must exactly match one `writeScopes.*.path` entry. Do not rely on partial overlap.
 
+For runtime paths in manifests or rendered Markdown, use `{{paths.*}}`.
+For ordinary preset values, use `{{valueName}}` from `templateValues`.
+Do not mix the two: `{{paths.*}}` is supplied by Harness from the target
+`harness.yaml`, while `{{valueName}}` is supplied by the preset manifest.
+
 ## Reference Bundle Pattern
 
 Use `resources.references` when the preset should preload common context for a group of tasks.
@@ -146,6 +173,7 @@ Do not confuse artifacts with evidence. Artifacts can be input packets or fixtur
 ## Safety Rules
 
 - `writeScopes` must be as narrow as possible.
+- Runtime write/read scopes must use `{{paths.*}}`, usually `{{paths.tasksRoot}}/**`.
 - Generated files must stay under the created task directory.
 - A preset must not mutate source code, Git state, or global governance tables during `new-task`.
 - Do not add JavaScript for `new-task` behavior.
@@ -159,13 +187,15 @@ Do not confuse artifacts with evidence. Artifacts can be input packets or fixtur
 2. For complex presets, open `references/complex-task-skeleton/README.md` and inspect the base task files the preset will overlay.
 3. Open `references/preset-package-skeleton.md` and copy only the files the preset actually needs.
 4. Create the preset directory with `preset.yaml`, templates, and resources.
-5. Keep task creation declarative: manifest inputs, `templateValues`, `metadata`, Markdown templates, `resources`, evidence declarations, and `writeScopes`.
-6. Run `harness preset check <path>`.
-7. Install with `harness preset install <path> --force` in a disposable or user-approved environment.
-8. Smoke test with `harness new-task <id> --budget <budget> --preset <preset-id> ... <target>`.
-9. For reference-bundle presets, create two different tasks from the same preset and verify both contain the same shared references but independent audit/evidence bundles.
-10. Run `harness status --json <target>`, `harness task-index --json <target>`, and `harness check --profile target-project <target>`.
-11. Inspect the generated `task_plan.md`, `references/INDEX.md`, and `artifacts/INDEX.md` manually before declaring success.
+5. Use `{{paths.tasksRoot}}/**` or another supported `{{paths.*}}` field for every runtime write/read scope; never hard-code `coding-agent-harness/...`.
+6. Keep task creation declarative: manifest inputs, `templateValues`, `metadata`, Markdown templates, `resources`, evidence declarations, and `writeScopes`.
+7. Run `harness preset check <path>`.
+8. Install with `harness preset install <path> --force` in a disposable or user-approved environment.
+9. Smoke test with `harness new-task <id> --budget <budget> --preset <preset-id> ... <target>`.
+10. For reference-bundle presets, create two different tasks from the same preset and verify both contain the same shared references but independent audit/evidence bundles.
+11. Run `harness status --json <target>`, `harness task-index --json <target>`, and `harness check --profile target-project <target>`.
+12. Run `harness preset audit --json` or `harness preset audit --project --json <target>` before replacing installed presets.
+13. Inspect the generated `task_plan.md`, `references/INDEX.md`, and `artifacts/INDEX.md` manually before declaring success.
 
 ## Quality Checklist
 
@@ -175,5 +205,7 @@ Do not confuse artifacts with evidence. Artifacts can be input packets or fixtur
 - Every required read points to a real `REF-*` row.
 - Every `REF-*` row explains why that reference matters.
 - Every generated artifact has an `ART-*` row when artifacts are used.
+- No runtime path is hard-coded to `coding-agent-harness/...`; scopes use `{{paths.*}}`.
+- `entrypoints.newTask.writes` and `writeScopes.*.path` match exactly.
 - The preset passes `preset check`, task creation, `status --json`, `task-index --json`, and target check.
 - A downstream agent can create a valid preset from this skill without editing Harness source.

package/skills/preset-creator/references/complex-task-skeleton/brief.md

@@ -1,7 +1,5 @@
 # {{TASK_TITLE}}
 
-## Brief
-
 ## Task ID
 
 `{{TASK_ID}}`
@@ -10,26 +8,52 @@
 
 {{DATE}}
 
-## Outcome
+## Outcome Statement
+
+One sentence stating the concrete result this task must produce.
+
+## Outcome Value
+
+Write 100-300 words explaining what the user, project, or next agent gets when
+this task is complete. Describe the decision, delivery, verification, or next
+development step this result unlocks. Focus on the usable result rather than the
+implementation process unless the implementation method is itself the
+deliverable.
+
+## Deliverables
 
-State the user-visible or project-visible result this task must deliver.
+- Visible artifact:
+- Modified surface:
+- Verification evidence:
+
+## First Human Read
+
+Name the files, evidence, or generated outputs a human should inspect first when
+they open this task.
 
 ## Boundaries
 
 - In scope: files, behavior, documentation, or verification this task may change.
 - Out of scope: work that must not be folded into this task.
-- Stop conditions: uncertainty, risk, or missing access that requires coordinator or user review.
+- Stop conditions: uncertainty, risk, or missing access that requires coordinator
+  or user review.
+
+## Completion Judgment
+
+List 3-5 concrete conditions that prove the outcome has been reached. Keep the
+full execution plan in `task_plan.md`.
 
 ## Execution Contract
 
 - Owner: coordinator
 - Lifecycle state: not_started
-- Required files: `task_plan.md`, `execution_strategy.md`, `visual_map.md`, `progress.md`, `findings.md`, `review.md`
+- Required files: `task_plan.md`, `execution_strategy.md`, `visual_map.md`,
+  `progress.md`, `findings.md`, `review.md`
 - Required closeout: verification evidence recorded in `progress.md`
 
-## First Next Step
+## Current Next Step
 
-Replace this line with the first concrete action before implementation starts.
+State the first concrete action before implementation starts.
 
 ## Scaffold Provenance
 

package/skills/preset-creator/references/preset-package-skeleton.md

@@ -33,7 +33,7 @@ task:
 entrypoints:
   newTask:
     type: template
-    writes: [docs/coding-agent-harness/planning/tasks/**]
+    writes: [{{paths.tasksRoot}}/**]
     audit: true
     templates:
       taskPlanAppend: templates/task_plan.append.md
@@ -90,7 +90,7 @@ audit:
   evidenceFiles: [preset-audit.json, preset-manifest.json, write-scope.json]
 writeScopes:
   taskDocs:
-    path: docs/coding-agent-harness/planning/tasks/**
+    path: {{paths.tasksRoot}}/**
     access: write
 ```
 
@@ -173,7 +173,7 @@ task:
 entrypoints:
   newTask:
     type: template
-    writes: [docs/coding-agent-harness/planning/tasks/**]
+    writes: [{{paths.tasksRoot}}/**]
     audit: true
     templates:
       taskPlanAppend: templates/task_plan.append.md
@@ -196,7 +196,7 @@ audit:
   evidenceFiles: [preset-audit.json, preset-manifest.json, write-scope.json]
 writeScopes:
   taskDocs:
-    path: docs/coding-agent-harness/planning/tasks/**
+    path: {{paths.tasksRoot}}/**
     access: write
 ```
 
@@ -205,6 +205,16 @@ writeScopes:
 - `id` uses lowercase letters, numbers, and hyphens only.
 - Supported `inputs.*.type` values are `text`, `flag`, and `json-file`.
 - `entrypoints.newTask.writes` entries must exactly match declared `writeScopes.*.path` entries.
+- Use `actions` only for task-level post-creation commands exposed through
+  `harness preset action <preset> <action>`.
+- `actions.*.command` must be a package-local `.mjs` file. Non-bundled script
+  actions require explicit `--allow-scripts` trust during install or run.
+- `actions.*.inputs` supports the same schema-only input types as preset
+  inputs: `text`, `flag`, and `json-file`.
+- `actions.*.writes` must be task-local. Prefer `{{task.paths.artifacts}}/**`,
+  `{{task.paths.artifactsIndex}}`, `{{task.paths.progress}}`, and
+  `{{task.paths.visualMap}}`; do not use broad `{{paths.tasksRoot}}/**` for an
+  action.
 - `path` for references must stay under `references/`.
 - `path` for artifacts must stay under `artifacts/`.
 - Do not target `references/INDEX.md`, `artifacts/INDEX.md`, `task_plan.md`, or any canonical task contract file.
@@ -267,7 +277,7 @@ Open `references/INDEX.md`, then read these preset-provided references before im
 
 | Reference | Path | Why |
 | --- | --- | --- |
-| REF-001 | TARGET:docs/coding-agent-harness/planning/tasks/<task-folder>/references/upstream-contract.md | [rendered summary] |
+| REF-001 | TARGET:docs/{{paths.harnessRoot}}/planning/tasks/<task-folder>/references/upstream-contract.md | [rendered summary] |
 ```
 
 The same `REF-001` and exact `TARGET:` path must also appear in `references/INDEX.md`.

package/skills/preset-creator/references/structure-aware-paths.md

@@ -0,0 +1,112 @@
+# Structure-Aware Preset Paths
+
+Use this reference before writing or reviewing any Harness `preset.yaml`.
+
+## Core Rule
+
+Do not hard-code Harness runtime paths such as
+`coding-agent-harness/planning/tasks/**`. Repositories may move the Harness
+folder, so runtime write/read scopes must use `{{paths.*}}`, resolved from the
+target repository's `harness.yaml`.
+
+For ordinary task-creating presets, use this pair:
+
+```yaml
+entrypoints:
+  newTask:
+    writes: [{{paths.tasksRoot}}/**]
+writeScopes:
+  taskDocs:
+    path: {{paths.tasksRoot}}/**
+    access: write
+```
+
+The `entrypoints.newTask.writes` entries must exactly match declared
+`writeScopes.*.path` entries.
+
+## Supported Fields
+
+| Field | Use |
+| --- | --- |
+| `{{paths.harnessRoot}}` | Harness root for the target repository |
+| `{{paths.planningRoot}}` | Planning root |
+| `{{paths.tasksRoot}}` | Task directory root |
+| `{{paths.modulesRoot}}` | Module planning root |
+| `{{paths.externalRoot}}` | External context root |
+| `{{paths.governanceRoot}}` | Governance root |
+| `{{paths.generatedRoot}}` | Generated governance files |
+| `{{paths.regressionRoot}}` | Regression records |
+| `{{paths.ledgerPath}}` | Harness ledger path |
+| `{{paths.closeoutIndexPath}}` | Closeout index path |
+
+## Minimal Manifest
+
+```yaml
+id: custom-review
+version: 1
+purpose: Create a custom review task
+compatibleBudgets: [standard, complex]
+localeSupport: [en-US]
+task:
+  kind: review-task
+  defaultTaskId: custom-review-task
+entrypoints:
+  newTask:
+    type: template
+    writes: [{{paths.tasksRoot}}/**]
+    audit: true
+    templates:
+      taskPlanAppend: templates/task_plan.append.md
+inputs:
+  subject:
+    type: text
+    flag: --subject
+    required: true
+templateValues:
+  subject:
+    from: inputs.subject
+metadata:
+  ReviewSubject:
+    label: Review Subject
+    from: inputs.subject
+evidence:
+  bundleDir: artifacts/preset
+audit:
+  manifestRequired: true
+  evidenceFiles: [preset-audit.json, preset-manifest.json, write-scope.json]
+writeScopes:
+  taskDocs:
+    path: {{paths.tasksRoot}}/**
+    access: write
+```
+
+## Path Boundaries
+
+- Use `{{paths.*}}` only for Harness runtime paths.
+- Keep preset package paths package-relative: `templates/task_plan.append.md`,
+  `resources/runbook.md`, `references/foo.md`.
+- Keep generated task-local resource destinations task-relative:
+  `references/foo.md`, `artifacts/foo.md`.
+- Do not write default-layout examples as runtime instructions. If prose must
+  mention `coding-agent-harness/`, label it as the default layout.
+- Do not overwrite user-customized installed presets unless the user explicitly
+  approved it.
+- Do not overwrite project-authored files or user-modified template projections.
+
+## Validation Commands
+
+```bash
+harness preset check ./my-preset
+harness new-task --title "Smoke" --preset my-preset /path/to/project
+harness check --profile target-project /path/to/project
+harness preset audit --json
+harness preset audit --project --json /path/to/project
+harness templates audit --json /path/to/project
+harness templates refresh --apply --json /path/to/project
+```
+
+`harness preset check` validates token syntax. Creating or installing against a
+target validates resolved scopes against that target's actual structure.
+
+Use `harness preset audit` before replacing installed user/project presets. Use
+`harness templates audit` before refreshing generated project templates.

package/templates/AGENTS.md.template

@@ -2,7 +2,7 @@
 
 This file is the operating entry point for agents working in this repository.
 Keep it as a charter and routing index. Detailed rules belong in
-`coding-agent-harness/governance/standards/`, not in this file.
+`{{paths.harnessRoot}}/governance/standards/`, not in this file.
 
 ## Project Profile
 
@@ -14,13 +14,13 @@ Keep it as a charter and routing index. Detailed rules belong in
 
 ## Non-Negotiable Rules
 
-1. Preserve architecture boundaries documented in `coding-agent-harness/governance/standards/engineering-standard.md`.
+1. Preserve architecture boundaries documented in `{{paths.harnessRoot}}/governance/standards/engineering-standard.md`.
 2. Do not commit secrets, credentials, private endpoints, or user data.
-3. Start non-trivial work with a Harness CLI-created task directory under `coding-agent-harness/planning/tasks/`.
+3. Start non-trivial work with a Harness CLI-created task directory under `{{paths.harnessRoot}}/planning/tasks/`.
 4. Record evidence before claiming completion.
 5. Protect unrelated working-tree changes; never revert files outside the assigned scope.
 6. Commit verified, meaningful work slices proactively. Do not leave completed work only as dirty files unless the user explicitly pauses commits, checks fail, dirty ownership is unclear, or a security boundary prevents a clean commit; in those cases record the no-commit reason, owner, and next step in progress, handoff, or closeout notes. Never mix unrelated dirty changes into a task commit.
-7. Before the final response, inspect the current task `visual_map.md` gate phase. If the current `Exit Command` has `Actor: agent`, run it or record the blocker. Never run a `human` gate such as `review-confirm` unless the user has explicitly performed or delegated that human confirmation.
+7. Before the final response, inspect the current task `visual_map.md` gate phase. If the current `Exit Command` has `Actor: agent`, run it or record the blocker. Never perform a `human` gate; human review confirmation belongs in the local Dashboard workbench.
 
 ## Reading Matrix
 
@@ -28,23 +28,23 @@ Read the smallest context that can answer the task.
 
 | Work Type | Read First |
 | --- | --- |
-| Architecture or core module change | `coding-agent-harness/governance/standards/engineering-standard.md` |
-| System map, service responsibility, or external system relationship | `coding-agent-harness/context/architecture/README.md`, `coding-agent-harness/context/architecture/service-catalog.md` |
-| Local development, mocks, stubs, or cross-repo debugging | `coding-agent-harness/context/development/README.md`, `coding-agent-harness/context/development/codebase-map.md` |
-| API, event, webhook, SDK, or third-party contract | `coding-agent-harness/context/integrations/README.md` and the related contract file |
-| Tests, smoke, or regression | `coding-agent-harness/governance/standards/testing-standard.md`, `coding-agent-harness/governance/regression/Regression-SSoT.md` |
-| Execution, commit, PR, or release work | `coding-agent-harness/governance/standards/execution-workflow-standard.md`, `coding-agent-harness/governance/standards/repo-governance-standard.md`, `coding-agent-harness/governance/standards/pull-request-standard.md`, `coding-agent-harness/governance/standards/ci-cd-standard.md` |
-| Creating or advancing a task | Current task directory under `coding-agent-harness/planning/tasks/`; use `harness new-task` / lifecycle commands when this project has the Harness CLI configured |
+| Architecture or core module change | `{{paths.harnessRoot}}/governance/standards/engineering-standard.md` |
+| System map, service responsibility, or external system relationship | `{{paths.harnessRoot}}/context/architecture/README.md`, `{{paths.harnessRoot}}/context/architecture/service-catalog.md` |
+| Local development, mocks, stubs, or cross-repo debugging | `{{paths.harnessRoot}}/context/development/README.md`, `{{paths.harnessRoot}}/context/development/codebase-map.md` |
+| API, event, webhook, SDK, or third-party contract | `{{paths.harnessRoot}}/context/integrations/README.md` and the related contract file |
+| Tests, smoke, or regression | `{{paths.harnessRoot}}/governance/standards/testing-standard.md`, `{{paths.harnessRoot}}/governance/regression/Regression-SSoT.md` |
+| Execution, commit, PR, or release work | `{{paths.harnessRoot}}/governance/standards/execution-workflow-standard.md`, `{{paths.harnessRoot}}/governance/standards/repo-governance-standard.md`, `{{paths.harnessRoot}}/governance/standards/pull-request-standard.md`, `{{paths.harnessRoot}}/governance/standards/ci-cd-standard.md` |
+| Creating or advancing a task | Current task directory under `{{paths.harnessRoot}}/planning/tasks/`; use `harness new-task` / lifecycle commands when this project has the Harness CLI configured |
 | Brief, Execution Strategy, or Visual Map | Current task `brief.md`, `execution_strategy.md`, and `visual_map.md` |
 | Long-running task | Current task `long-running-task-contract.md` when present |
-| Reviewer, subagent, or adversarial review | Current task `review.md`, `coding-agent-harness/governance/standards/review-routing-standard.md` |
-| Team, multi-repo, or staged delivery | `coding-agent-harness/governance/standards/delivery-operating-model-standard.md`, `coding-agent-harness/planning/Delivery-SSoT.md` |
-| Module-parallel work | `coding-agent-harness/planning/modules/Module-Registry.md` and the related module plan |
-| Documentation governance or Harness update | `coding-agent-harness/governance/standards/docs-library-standard.md`, `coding-agent-harness/harness.yaml` |
-| External source material intake | `coding-agent-harness/governance/standards/external-source-intake-standard.md` |
-| Regression SSoT maintenance | `coding-agent-harness/governance/standards/regression-ssot-governance.md` |
-| Walkthrough, closeout, or lessons | `coding-agent-harness/governance/standards/walkthrough-standard.md`, current task `walkthrough.md`, current task `lesson_candidates.md`, `coding-agent-harness/governance/lessons/` |
-| Worktree operations | `coding-agent-harness/governance/standards/worktree-standard.md` |
+| Reviewer, subagent, or adversarial review | Current task `review.md`, `{{paths.harnessRoot}}/governance/standards/review-routing-standard.md` |
+| Team, multi-repo, or staged delivery | `{{paths.harnessRoot}}/governance/standards/delivery-operating-model-standard.md`, `{{paths.harnessRoot}}/planning/Delivery-SSoT.md` |
+| Module-parallel work | `{{paths.harnessRoot}}/harness.yaml`, generated `{{paths.harnessRoot}}/planning/modules/Module-Registry.md`, the module `brief.md`, and the related `module_plan.md` |
+| Documentation governance or Harness update | `{{paths.harnessRoot}}/governance/standards/docs-library-standard.md`, `{{paths.harnessRoot}}/harness.yaml` |
+| External source material intake | `{{paths.harnessRoot}}/governance/standards/external-source-intake-standard.md` |
+| Regression SSoT maintenance | `{{paths.harnessRoot}}/governance/standards/regression-ssot-governance.md` |
+| Walkthrough, closeout, or lessons | `{{paths.harnessRoot}}/governance/standards/walkthrough-standard.md`, current task `walkthrough.md`, current task `lesson_candidates.md`, `{{paths.harnessRoot}}/governance/lessons/` |
+| Worktree operations | `{{paths.harnessRoot}}/governance/standards/worktree-standard.md` |
 
 ## Standard Execution Flow
 
@@ -70,17 +70,19 @@ Read the smallest context that can answer the task.
 - Worker subagents require one user authorization recorded in `execution_strategy.md`; after that, reuse is allowed only within the same task, scope, and worktree/branch.
 - Shared ledgers, registries, and SSoT files are coordinator-owned unless a lock is explicitly assigned.
 - Worker handoff must include branch or worktree, changed files, checks, evidence, and residual risks.
+- Modules are registered in `{{paths.harnessRoot}}/harness.yaml` under `modules.items`. `Module-Registry.md` is generated; do not hand-edit it as the source of truth.
+- Module roots own only `brief.md` and `module_plan.md` by default. Task execution contracts such as `execution_strategy.md`, `visual_map.md`, `review.md`, and `walkthrough.md` belong under concrete task directories, including `{{paths.harnessRoot}}/planning/modules/<key>/tasks/<task-id>/`.
 
 ## Single Sources Of Truth
 
-- Delivery SSoT: `coding-agent-harness/planning/Delivery-SSoT.md`
-- Module Registry: `coding-agent-harness/planning/modules/Module-Registry.md`
-- Regression SSoT: `coding-agent-harness/governance/regression/Regression-SSoT.md`
-- Cadence Ledger: `coding-agent-harness/governance/regression/Cadence-Ledger.md`
+- Delivery SSoT: `{{paths.harnessRoot}}/planning/Delivery-SSoT.md`
+- Module Registry: `{{paths.harnessRoot}}/harness.yaml` `modules.items`; generated view at `{{paths.harnessRoot}}/planning/modules/Module-Registry.md`
+- Regression SSoT: `{{paths.harnessRoot}}/governance/regression/Regression-SSoT.md`
+- Cadence Ledger: `{{paths.harnessRoot}}/governance/regression/Cadence-Ledger.md`
 - Lesson Candidates: current task `lesson_candidates.md`
-- Lesson Detail Docs: `coding-agent-harness/governance/lessons/`
-- Generated Closeout Index: `coding-agent-harness/governance/generated/Closeout-Index.md`
-- Harness Ledger: `coding-agent-harness/governance/generated/Harness-Ledger.md`
+- Lesson Detail Docs: `{{paths.harnessRoot}}/governance/lessons/`
+- Generated Closeout Index: `{{paths.harnessRoot}}/governance/generated/Closeout-Index.md`
+- Harness Ledger: `{{paths.harnessRoot}}/governance/generated/Harness-Ledger.md`
 
 ## Local Commands
 

package/templates/architecture/README.md

@@ -22,8 +22,8 @@ This folder is the system-structure source of truth. It explains what the curren
 ## Boundary
 
 - Put system structure, service responsibility, ownership, and critical flows here.
-- Put payload bodies, endpoint parameters, event schemas, and SDK details in `coding-agent-harness/context/integrations/`.
-- Put local setup, mocks, stubs, and cross-repo debugging notes in `coding-agent-harness/context/development/`.
+- Put payload bodies, endpoint parameters, event schemas, and SDK details in `{{paths.harnessRoot}}/context/integrations/`.
+- Put local setup, mocks, stubs, and cross-repo debugging notes in `{{paths.harnessRoot}}/context/development/`.
 
 ## Structure Contract