feat(han-core): add the pairing collaborative working mode

CLAUDE.md

@@ -9,7 +9,7 @@ surfaces the first two, the inline `explanation-guidance` skill that surfaces th
 the `readability-editor` agent, and the `han-readability` output style that applies the first two to every turn of a
 session; it depends on nothing and every prose-producing plugin depends on it), `han-core` (the shared foundation: the
 specialist agent roster the rest of the suite dispatches — every shared agent except the `readability-editor`, the
-`research-analyst`, and the `discussion-facilitator` — plus the `project-discovery` skill and the canonical rule files; depends on no other Han plugin),
+`research-analyst`, and the `discussion-facilitator` — plus the `project-discovery` skill, the `pairing` collaborative working mode, and the canonical rule files; depends on no other Han plugin),
 `han-documentation` (the documentation skills: `project-documentation`, `architectural-decision-record`, and `runbook`;
 depends on `han-communication` and `han-core` and is bundled by the `han` meta-plugin), `han-research` (the
 pre-planning knowledge-work skills — `research`, `gap-analysis`, and `issue-triage` — plus the `research-analyst`
@@ -80,14 +80,14 @@ han-plugin-builder skill:
 │   ├── output-styles/  # han-readability.md: the readability rule and writing voice distilled into a selectable output style (auto-discovered default location; no plugin.json field)
 │   ├── docs/           # In-plugin long-form docs: docs/skills/{name}.md + docs/agents/readability-editor.md + docs/output-styles/han-readability.md
 │   └── references/     # Canonical readability-rule.md + writing-voice.md + explanation-rule.md (owned here; no vendored copies elsewhere), beside a vendored config-rule.md
-├── han-core/           # Core plugin: the shared specialist agent roster (all agents except readability-editor and research-analyst) + project-discovery (depends on no other Han plugin)
+├── han-core/           # Core plugin: the shared specialist agent roster (all agents except readability-editor and research-analyst) + project-discovery + pairing (depends on no other Han plugin)
 │   ├── README.md       # Light front door; skills and agents grouped by purpose
 │   ├── .claude-plugin/
 │   │   └── plugin.json
 │   ├── agents/         # Agent definitions (.md with frontmatter)
-│   ├── skills/         # project-discovery skill directory, with SKILL.md + references/
+│   ├── skills/         # project-discovery and pairing skill directories, each with SKILL.md (+ references/ where used)
 │   ├── docs/           # In-plugin long-form docs: docs/skills/{name}.md + docs/agents/{name}.md
-│   └── references/     # Cross-skill reference files (yagni-rule.md, evidence-rule.md, config-rule.md — canonical copies)
+│   └── references/     # Cross-skill reference files (yagni-rule.md, evidence-rule.md, config-rule.md, collaborative-stop-rule.md — canonical copies)
 ├── han-documentation/  # Documentation plugin: project-documentation, architectural-decision-record, runbook (depends on han-communication and han-core; bundled by the han meta-plugin)
 │   ├── README.md       # Light front door + scent-line skills list
 │   ├── .claude-plugin/
@@ -110,14 +110,14 @@ han-plugin-builder skill:
 │   ├── agents/         # discussion-facilitator agent definition
 │   ├── skills/         # Planning skill directories, each with SKILL.md + references/
 │   ├── docs/           # In-plugin long-form docs: docs/skills/{name}.md + docs/agents/discussion-facilitator.md
-│   └── references/     # Both kinds: han-planning-owned canonical files (planning-boundary-rule.md, scope-justification-rule.md, operator-escalation-rule.md) beside vendored copies (yagni-rule.md, evidence-rule.md, config-rule.md). Each owned file opens by saying so; do not overwrite one in a re-sync sweep
+│   └── references/     # Both kinds: han-planning-owned canonical files (planning-boundary-rule.md, scope-justification-rule.md, operator-escalation-rule.md) beside vendored copies (yagni-rule.md, evidence-rule.md, config-rule.md, collaborative-stop-rule.md). Each owned file opens by saying so; do not overwrite one in a re-sync sweep
 ├── han-coding/         # Coding plugin: tdd, refactor, design-an-api, code-review, code-overview, code-walkthrough, architectural-analysis, automated-test-planning, manual-test-planning, investigate, coding-standard (the skills for working in code; depends on han-communication and han-core; bundled by the han meta-plugin)
 │   ├── README.md       # Light front door + scent-line skills list
 │   ├── .claude-plugin/
 │   │   └── plugin.json
 │   ├── skills/         # Coding-facing skill directories, each with SKILL.md + references/ (+ scripts/ where used)
 │   ├── docs/           # In-plugin long-form docs: docs/skills/{name}.md
-│   └── references/     # Cross-skill reference files vendored for han-coding skills (yagni-rule.md, evidence-rule.md, config-rule.md)
+│   └── references/     # Cross-skill reference files vendored for han-coding skills (yagni-rule.md, evidence-rule.md, config-rule.md, collaborative-stop-rule.md)
 ├── han-github/         # GitHub plugin: post-code-review-to-pr, update-pr-description, work-items-to-issues (depends on han-communication, han-core, and han-coding, whose code-review skill post-code-review-to-pr wraps)
 │   ├── README.md       # Light front door + scent-line skills list
 │   ├── .claude-plugin/
@@ -238,6 +238,15 @@ such as Claude, should be referenced here.
   (schema tokens, precedence, containment, pool-join, degradation) every participating skill applies. Vendored
   byte-identical into every skill-carrying plugin's `references/`; edit the canonical copy and re-sync the others.
 
+### Collaborative working mode
+
+- **[han-core/references/collaborative-stop-rule.md](./han-core/references/collaborative-stop-rule.md).** The shared
+  contract behind `/pairing`: how a skill detects the collaborative flag, what a stop presents and in what order, when
+  the pre-build ask fires and what makes a choice expensive to walk back, and what to do with the answer. Consumed by
+  `pairing` and by the skills that hand control back at their own boundaries — `tdd`, `refactor`, `design-an-api`,
+  `iterative-plan-review`, and `plan-implementation`. Vendored byte-identical into `han-coding/references/` and
+  `han-planning/references/`; edit the canonical copy and re-sync the others.
+
 ### Writing voice
 
 - **[han-communication/references/writing-voice.md](./han-communication/references/writing-voice.md).** Voice profile

README.md

@@ -45,7 +45,7 @@ Han ships as multiple plugins:
 | -------------------- | ------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
 | **`han`**            | parent  | the parent plugin that brings in `han-communication`, `han-core`, `han-documentation`, `han-research`, `han-planning`, `han-coding`, `han-github`, and `han-reporting` |
 | `han-communication`  | bundled | the foundational plugin beneath every other: the shared readability standard and writing-voice profile, plus the skills and agent that apply them                      |
-| `han-core`           | bundled | the shared specialist agent roster, the project-discovery skill, and the canonical rule files                                                                          |
+| `han-core`           | bundled | the shared specialist agent roster, the project-discovery skill, the `/pairing` collaborative working mode, and the canonical rule files                               |
 | `han-documentation`  | bundled | documentation skills: project docs, architectural decision records, and runbooks                                                                                       |
 | `han-research`       | bundled | pre-planning knowledge-work skills: research, gap analysis, and issue triage, plus the research-analyst agent                                                          |
 | `han-planning`       | bundled | planning skills you reach for before implementation                                                                                                                    |

docs/choosing-a-han-plugin.md

@@ -34,7 +34,8 @@ description of what it does. The `han` meta-plugin is a convenience wrapper that
   plus the skills and agent that apply them and a `Han Readability` output style that applies the readability standard
   to a whole session. Bundled; depends on nothing.
 - **[`han-core`](../han-core/README.md).** The shared foundation: the specialist agent roster the other plugins
-  dispatch, the project-discovery skill, and the canonical rule files. Bundled; depends on no other Han plugin.
+  dispatch, the project-discovery skill, the canonical rule files, and the `/pairing` working mode that builds any kind
+  of work in reviewable pieces. Bundled; depends on no other Han plugin.
 - **[`han-documentation`](../han-documentation/README.md).** The documentation layer: feature and system docs,
   architectural decision records, and runbooks. Bundled; depends on `han-communication` and `han-core`.
 - **[`han-research`](../han-research/README.md).** The pre-planning knowledge-work layer: open-ended research, gap
@@ -73,13 +74,14 @@ So you might expect installing one to give you that slice of Han with nothing el
 their skills dispatch the shared specialist agents that live there, and on `han-communication`, because their skills
 produce prose and source the readability standard from it. When you install a plugin that declares a dependency, Claude
 Code resolves and installs the dependency for you automatically and tells you what it added. So installing any of them
-installs both alongside it, and you get the shared agent roster, project discovery, and the readability standard either
-way. (`han-reporting` is the exception: it depends on `han-communication` alone.)
+installs both alongside it, and you get the shared agent roster, project discovery, the pairing working mode, and the
+readability standard either way. (`han-reporting` is the exception: it depends on `han-communication` alone.)
 
 That means **every layer install comes with the shared agents.** The real choice comes down to:
 
 - **A layer plus the foundations** (for example `han-documentation` or `han-coding`): that layer's skills, plus the
-  shared agent roster and project discovery from `han-core` and the readability standard from `han-communication`.
+  shared agent roster, project discovery, and the pairing working mode from `han-core`, and the readability standard
+  from `han-communication`.
 - **The bundled suite** (`han`): every layer at once.
 
 The opt-in plugins (`han-feedback`, `han-atlassian`, `han-linear`, `han-plugin-builder`) sit outside that choice. The
@@ -98,7 +100,7 @@ to.
 | You work with GitHub from Claude Code (review PRs, write PR descriptions, publish work items as issues) | `han` (the bundled suite includes the GitHub skills)                                 | `/plugin install han@han`                |
 | You want only the documentation skills (project docs, ADRs, runbooks)                                   | `han-documentation` (brings the `han-core` agents along)                             | `/plugin install han-documentation@han`  |
 | You want only the pre-planning research skills (research, gap analysis, issue triage)                   | `han-research` (brings the `han-core` agents along)                                  | `/plugin install han-research@han`       |
-| You want only the shared agents and project discovery, with no other skills                             | `han-core`                                                                           | `/plugin install han-core@han`           |
+| You want the shared agents, project discovery, and the pairing working mode                             | `han-core`                                                                           | `/plugin install han-core@han`           |
 | You installed a single layer and now want the planning skills                                           | `han-planning` (alongside what you already have)                                     | `/plugin install han-planning@han`       |
 | You installed a single layer and now want the coding skills                                             | `han-coding` (alongside what you already have)                                       | `/plugin install han-coding@han`         |
 | You want to send post-session feedback on Han skills to the maintainers                                 | `han-feedback` (alongside whatever you already have)                                 | `/plugin install han-feedback@han`       |

docs/concepts.md

@@ -232,7 +232,7 @@ Direct invocation uses the `Agent` tool with `subagent_type: {plugin}:{agent-nam
 ## How Han is packaged
 
 Han ships as a family of plugins in one marketplace. `han-core` carries the shared specialist agent roster the other
-plugins dispatch, the project-discovery skill, and the canonical rule files.
+plugins dispatch, the project-discovery skill, the `/pairing` collaborative working mode, and the canonical rule files.
 
 `han-documentation` adds the documentation skills (`/project-documentation`, `/architectural-decision-record`, and
 `/runbook`). `han-research` adds the pre-planning knowledge-work skills (`/research`, `/gap-analysis`, and