diff --git a/docs/tests/inbox/README.md b/docs/tests/inbox/README.md
new file mode 100644
index 0000000..3721956
--- /dev/null
+++ b/docs/tests/inbox/README.md
@@ -0,0 +1,70 @@
+# Inbox Markdown Test Plan
+
+## Purpose
+
+This directory contains the human-readable Markdown test plan for the `inbox` CLI.
+
+It complements automated Go tests. The goal is not to restate implementation details, but to preserve the user-visible CLI contract in a form that can be reviewed, extended, and executed manually when needed.
+
+## Directory Rules
+
+- one folder per command or shared area
+- one `README.md` per folder
+- no one-file-per-case sprawl
+- no numeric test IDs
+- each case is identified by `path + case slug`
+
+Recommended case heading pattern:
+
+```md
+  ## case: send-rejects-invalid-payload-json
+```
+
+## Authoring Principles
+
+- focus on externally visible behavior of the CLI
+- prefer stable command examples that a new agent can replay against a temp database
+- describe both success shape and failure contract
+- when a case already exists in automated Go tests, reuse its scenario rather than inventing a new one
+- keep terminology consistent with command flags and JSON fields exposed by the CLI
+
+## Common Execution Model
+
+Most cases in this directory assume the same baseline:
+
+1. create an isolated temporary directory
+2. choose a database path such as `TMPDIR/coord.db`
+3. run `inbox --db TMPDIR/coord.db --json init`
+4. run the target command sequence against that database
+
+Unless a case says otherwise:
+
+- commands should use `--json`
+- assertions should check both exit code and JSON payload
+- examples may use explicit `--agent`, or rely on the root `--agent` flag when that is the behavior under test
+
+## Folder Map
+
+- `README.md`: global conventions and glossary
+- `_shared/README.md`: reusable fixtures, JSON assertions, exit codes, payload rules
+- `workflows/README.md`: cross-command end-to-end scenarios
+- per-command folders: command-specific cases and edge conditions
+
+## Glossary
+
+- `thread`: the durable coordination unit tracked by `thread_id`
+- `message`: an event-bearing entry appended to a thread
+- `artifact`: a file attachment associated with a message
+- `read cursor`: the per-agent marker used by unread flows
+- `lease`: the temporary ownership granted by `claim` and extended by `renew`
+- `terminal state`: a thread state such as `done`, `failed`, or `cancelled`
+
+## Relationship To Automated Tests
+
+The current best executable reference is [internal/cli/inbox/integration_test.go](../../../internal/cli/inbox/integration_test.go).
+
+When this Markdown plan is expanded:
+
+- prefer matching an existing automated scenario first
+- record any additional manual-only contract coverage explicitly in the relevant command document
+- keep `docs/tests/inbox/ROADMAP.md` synchronized with authored files and case slugs
diff --git a/docs/tests/inbox/ROADMAP.md b/docs/tests/inbox/ROADMAP.md
index 0b08498..b125c35 100644
--- a/docs/tests/inbox/ROADMAP.md
+++ b/docs/tests/inbox/ROADMAP.md
@@ -25,14 +25,14 @@ Current state:
 - `inbox` CLI is implemented end-to-end
 - automated Go integration tests already exist for the main lifecycle, wait flows, unread behavior, artifacts, and JSON error contracts
 - this roadmap now exists under `docs/tests/inbox/ROADMAP.md`
-- the command, workflow, and shared Markdown test-plan documents have not been authored yet
+- all planned global, shared, workflow, and command-level Markdown test-plan documents have been authored
 
 Progress summary for planned test-plan documents, excluding `ROADMAP.md`:
 
-- planned document files: `16`
-- authored document files: `0`
+- planned document files: `17`
+- authored document files: `17`
 - planned case slugs in this roadmap: `61`
-- authored case slugs in this roadmap: `0`
+- authored case slugs in this roadmap: `61`
 
 ## Scope
 
@@ -105,12 +105,12 @@ Allowed status values in this roadmap:
 
 The Markdown test-plan set starts at zero, but these automated tests already exist and should be used as source material when writing the docs:
 
-- [integration_test.go](/home/kurihada/project/ai-workflow-skill/internal/cli/inbox/integration_test.go#L12) `TestInboxLifecycle`
-- [integration_test.go](/home/kurihada/project/ai-workflow-skill/internal/cli/inbox/integration_test.go#L176) `TestInboxFailLifecycle`
-- [integration_test.go](/home/kurihada/project/ai-workflow-skill/internal/cli/inbox/integration_test.go#L243) `TestInboxRenewWaitReplyAndCancel`
-- [integration_test.go](/home/kurihada/project/ai-workflow-skill/internal/cli/inbox/integration_test.go#L392) `TestInboxWatchListUnreadAndAppend`
-- [integration_test.go](/home/kurihada/project/ai-workflow-skill/internal/cli/inbox/integration_test.go#L549) `TestInboxUnreadReadCursor`
-- [integration_test.go](/home/kurihada/project/ai-workflow-skill/internal/cli/inbox/integration_test.go#L639) `TestInboxJSONErrorsAndExitCodes`
+- [integration_test.go](../../../internal/cli/inbox/integration_test.go#L12) `TestInboxLifecycle`
+- [integration_test.go](../../../internal/cli/inbox/integration_test.go#L176) `TestInboxFailLifecycle`
+- [integration_test.go](../../../internal/cli/inbox/integration_test.go#L243) `TestInboxRenewWaitReplyAndCancel`
+- [integration_test.go](../../../internal/cli/inbox/integration_test.go#L392) `TestInboxWatchListUnreadAndAppend`
+- [integration_test.go](../../../internal/cli/inbox/integration_test.go#L549) `TestInboxUnreadReadCursor`
+- [integration_test.go](../../../internal/cli/inbox/integration_test.go#L639) `TestInboxJSONErrorsAndExitCodes`
 
 These tests do not remove the need for the Markdown plan. They only reduce discovery work.
 
@@ -158,23 +158,23 @@ docs/tests/inbox/
 
 | Path | Purpose | Planned Cases | Authored Cases | Status |
 | --- | --- | ---: | ---: | --- |
-| `docs/tests/inbox/README.md` | Global testing conventions and glossary | 0 | 0 | pending |
-| `docs/tests/inbox/_shared/README.md` | Shared fixtures, JSON assertions, exit-code rules | 0 | 0 | pending |
-| `docs/tests/inbox/workflows/README.md` | Cross-command scenarios | 8 | 0 | pending |
-| `docs/tests/inbox/init/README.md` | `init` command cases | 2 | 0 | pending |
-| `docs/tests/inbox/send/README.md` | `send` command cases | 6 | 0 | pending |
-| `docs/tests/inbox/fetch/README.md` | `fetch` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/claim/README.md` | `claim` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/renew/README.md` | `renew` command cases | 3 | 0 | pending |
-| `docs/tests/inbox/update/README.md` | `update` command cases | 5 | 0 | pending |
-| `docs/tests/inbox/reply/README.md` | `reply` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/done/README.md` | `done` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/fail/README.md` | `fail` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/cancel/README.md` | `cancel` command cases | 3 | 0 | pending |
-| `docs/tests/inbox/list/README.md` | `list` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/show/README.md` | `show` command cases | 4 | 0 | pending |
-| `docs/tests/inbox/watch/README.md` | `watch` command cases | 3 | 0 | pending |
-| `docs/tests/inbox/wait-reply/README.md` | `wait-reply` command cases | 3 | 0 | pending |
+| `docs/tests/inbox/README.md` | Global testing conventions and glossary | 0 | 0 | done |
+| `docs/tests/inbox/_shared/README.md` | Shared fixtures, JSON assertions, exit-code rules | 0 | 0 | done |
+| `docs/tests/inbox/workflows/README.md` | Cross-command scenarios | 8 | 8 | done |
+| `docs/tests/inbox/init/README.md` | `init` command cases | 2 | 2 | done |
+| `docs/tests/inbox/send/README.md` | `send` command cases | 6 | 6 | done |
+| `docs/tests/inbox/fetch/README.md` | `fetch` command cases | 4 | 4 | done |
+| `docs/tests/inbox/claim/README.md` | `claim` command cases | 4 | 4 | done |
+| `docs/tests/inbox/renew/README.md` | `renew` command cases | 3 | 3 | done |
+| `docs/tests/inbox/update/README.md` | `update` command cases | 5 | 5 | done |
+| `docs/tests/inbox/reply/README.md` | `reply` command cases | 4 | 4 | done |
+| `docs/tests/inbox/done/README.md` | `done` command cases | 4 | 4 | done |
+| `docs/tests/inbox/fail/README.md` | `fail` command cases | 4 | 4 | done |
+| `docs/tests/inbox/cancel/README.md` | `cancel` command cases | 3 | 3 | done |
+| `docs/tests/inbox/list/README.md` | `list` command cases | 4 | 4 | done |
+| `docs/tests/inbox/show/README.md` | `show` command cases | 4 | 4 | done |
+| `docs/tests/inbox/watch/README.md` | `watch` command cases | 3 | 3 | done |
+| `docs/tests/inbox/wait-reply/README.md` | `wait-reply` command cases | 3 | 3 | done |
 
 ## Authoring Order
 
@@ -198,121 +198,79 @@ Reason:
 
 ## Authored Case Register
 
-No Markdown test cases have been authored yet.
-
-When the first case is written, add rows in this format:
-
 | Path | Case Slug | Coverage Note | Status |
 | --- | --- | --- | --- |
-| `docs/tests/inbox/send/README.md` | `send-creates-new-thread` | minimal happy path for new thread creation | done |
+| `docs/tests/inbox/workflows/README.md` | `thread-lifecycle-happy-path` | end-to-end happy path from send to show after done | done |
+| `docs/tests/inbox/workflows/README.md` | `blocked-question-reply-resume-to-done` | blocked thread receives answer and resumes to done | done |
+| `docs/tests/inbox/workflows/README.md` | `fail-lifecycle-from-claim-to-terminal` | claimed thread transitions to failed terminal state | done |
+| `docs/tests/inbox/workflows/README.md` | `cancel-lifecycle-after-worker-claim` | claimed thread can be cancelled by initiator | done |
+| `docs/tests/inbox/workflows/README.md` | `watch-wakes-then-fetch-sees-new-thread` | watch wake-up remains consistent with unread fetch visibility | done |
+| `docs/tests/inbox/workflows/README.md` | `artifact-visible-through-send-and-show` | body-file and artifact data survive send and show | done |
+| `docs/tests/inbox/workflows/README.md` | `unread-clears-after-mark-read-and-reappears-on-new-message` | read cursor clears unread and new message restores it | done |
+| `docs/tests/inbox/workflows/README.md` | `wait-reply-clears-blocked-unread-for-agent` | wait-reply consumes reply and clears blocked unread view | done |
+| `docs/tests/inbox/init/README.md` | `init-creates-schema-on-empty-db` | initializes an empty database path and returns initialized status | done |
+| `docs/tests/inbox/init/README.md` | `init-is-idempotent-on-existing-db` | repeated init succeeds on the same database path | done |
+| `docs/tests/inbox/send/README.md` | `send-creates-new-thread` | creates a pending thread with an initial task message | done |
+| `docs/tests/inbox/send/README.md` | `send-appends-message-to-existing-thread` | appends a message to an existing non-terminal thread | done |
+| `docs/tests/inbox/send/README.md` | `send-reads-body-from-body-file` | reads message body from a file path | done |
+| `docs/tests/inbox/send/README.md` | `send-attaches-artifact-with-metadata` | persists artifact path, kind, and metadata on send | done |
+| `docs/tests/inbox/send/README.md` | `send-rejects-invalid-payload-json` | rejects malformed payload JSON with invalid_input | done |
+| `docs/tests/inbox/send/README.md` | `send-rejects-invalid-artifact-metadata-json` | rejects malformed artifact metadata JSON | done |
+| `docs/tests/inbox/fetch/README.md` | `fetch-returns-pending-thread-for-target-agent` | returns pending candidate work for the target agent | done |
+| `docs/tests/inbox/fetch/README.md` | `fetch-respects-status-and-limit-filters` | enforces status filtering and max row count | done |
+| `docs/tests/inbox/fetch/README.md` | `fetch-unread-uses-read-cursor` | unread filtering depends on per-agent read cursor state | done |
+| `docs/tests/inbox/fetch/README.md` | `fetch-returns-no-matching-work-when-empty` | empty fetch result returns no_matching_work | done |
+| `docs/tests/inbox/claim/README.md` | `claim-acquires-thread-lease` | claims a pending thread and records a claim event message | done |
+| `docs/tests/inbox/claim/README.md` | `claim-rejects-when-thread-missing` | missing thread returns not_found | done |
+| `docs/tests/inbox/claim/README.md` | `claim-rejects-when-thread-already-claimed` | active lease conflict returns lease_conflict | done |
+| `docs/tests/inbox/claim/README.md` | `claim-records-requested-lease-duration` | claim event payload records requested lease duration | done |
+| `docs/tests/inbox/renew/README.md` | `renew-extends-active-lease` | owner renews an active lease and gets a renewal event | done |
+| `docs/tests/inbox/renew/README.md` | `renew-rejects-non-owner` | non-owner renew attempt returns lease_conflict | done |
+| `docs/tests/inbox/renew/README.md` | `renew-rejects-without-active-lease` | missing active lease returns invalid_state | done |
+| `docs/tests/inbox/update/README.md` | `update-moves-thread-to-in-progress` | owner moves a thread into in_progress with a progress message | done |
+| `docs/tests/inbox/update/README.md` | `update-moves-thread-to-blocked-with-payload` | blocked update writes a question message and payload JSON | done |
+| `docs/tests/inbox/update/README.md` | `update-accepts-body-file-and-artifact` | update supports body-file input and artifact attachments | done |
+| `docs/tests/inbox/update/README.md` | `update-rejects-invalid-payload-json` | malformed update payload returns invalid_input | done |
+| `docs/tests/inbox/update/README.md` | `update-rejects-non-owner` | non-owner update attempt returns lease_conflict | done |
+| `docs/tests/inbox/reply/README.md` | `reply-adds-answer-message` | default reply kind is answer and thread stays non-terminal | done |
+| `docs/tests/inbox/reply/README.md` | `reply-supports-control-kind` | reply can explicitly send control messages | done |
+| `docs/tests/inbox/reply/README.md` | `reply-attaches-artifact` | reply persists artifact data on the message | done |
+| `docs/tests/inbox/reply/README.md` | `reply-rejects-invalid-payload-json` | malformed reply payload returns invalid_input | done |
+| `docs/tests/inbox/done/README.md` | `done-marks-thread-terminal` | owner completes a thread into done terminal state | done |
+| `docs/tests/inbox/done/README.md` | `done-persists-result-body-and-artifact` | result body and artifacts remain visible after done | done |
+| `docs/tests/inbox/done/README.md` | `done-rejects-non-owner` | non-owner done attempt returns lease_conflict | done |
+| `docs/tests/inbox/done/README.md` | `done-rejects-on-terminal-thread` | terminal threads reject repeated done calls | done |
+| `docs/tests/inbox/fail/README.md` | `fail-marks-thread-failed` | owner completes a thread into failed terminal state | done |
+| `docs/tests/inbox/fail/README.md` | `fail-persists-failure-body-and-artifact` | failure body and artifacts remain visible after fail | done |
+| `docs/tests/inbox/fail/README.md` | `fail-rejects-non-owner` | non-owner fail attempt returns lease_conflict | done |
+| `docs/tests/inbox/fail/README.md` | `fail-rejects-on-terminal-thread` | terminal threads reject repeated fail calls | done |
+| `docs/tests/inbox/cancel/README.md` | `cancel-marks-thread-cancelled` | cancel moves a non-terminal thread to cancelled | done |
+| `docs/tests/inbox/cancel/README.md` | `cancel-persists-reason-and-artifact` | cancel records its reason text and attachments | done |
+| `docs/tests/inbox/cancel/README.md` | `cancel-rejects-when-thread-missing` | missing thread returns not_found on cancel | done |
+| `docs/tests/inbox/list/README.md` | `list-filters-by-status` | status filter limits listed threads | done |
+| `docs/tests/inbox/list/README.md` | `list-filters-by-created-by` | created-by filter limits listed threads | done |
+| `docs/tests/inbox/list/README.md` | `list-filters-by-assigned-to` | assigned-to filter limits listed threads | done |
+| `docs/tests/inbox/list/README.md` | `list-respects-limit` | list returns at most the requested number of rows | done |
+| `docs/tests/inbox/show/README.md` | `show-returns-thread-and-message-history` | show returns thread metadata and ordered history | done |
+| `docs/tests/inbox/show/README.md` | `show-includes-artifacts-per-message` | show expands message artifacts in detail view | done |
+| `docs/tests/inbox/show/README.md` | `show-mark-read-advances-read-cursor` | mark-read changes subsequent unread visibility | done |
+| `docs/tests/inbox/show/README.md` | `show-rejects-when-thread-missing` | missing thread returns not_found on show | done |
+| `docs/tests/inbox/watch/README.md` | `watch-wakes-on-matching-thread` | watch wakes on a new matching thread event | done |
+| `docs/tests/inbox/watch/README.md` | `watch-respects-status-filter` | watch only wakes on events whose resulting thread status matches | done |
+| `docs/tests/inbox/watch/README.md` | `watch-times-out-with-no-activity` | watch timeout returns no_matching_work | done |
+| `docs/tests/inbox/wait-reply/README.md` | `wait-reply-wakes-on-answer-after-message` | wait-reply resumes from a known message boundary | done |
+| `docs/tests/inbox/wait-reply/README.md` | `wait-reply-can-start-from-after-event` | wait-reply resumes from a known event cursor | done |
+| `docs/tests/inbox/wait-reply/README.md` | `wait-reply-times-out-when-no-reply` | wait-reply timeout returns no_matching_work | done |
 
 ## Pending Case Backlog
 
-### `docs/tests/inbox/workflows/README.md`
+No pending case slugs remain in the current plan.
 
-- `pending` `thread-lifecycle-happy-path`
-- `pending` `blocked-question-reply-resume-to-done`
-- `pending` `fail-lifecycle-from-claim-to-terminal`
-- `pending` `cancel-lifecycle-after-worker-claim`
-- `pending` `watch-wakes-then-fetch-sees-new-thread`
-- `pending` `artifact-visible-through-send-and-show`
-- `pending` `unread-clears-after-mark-read-and-reappears-on-new-message`
-- `pending` `wait-reply-clears-blocked-unread-for-agent`
+When a new CLI contract or workflow needs coverage:
 
-### `docs/tests/inbox/init/README.md`
-
-- `pending` `init-creates-schema-on-empty-db`
-- `pending` `init-is-idempotent-on-existing-db`
-
-### `docs/tests/inbox/send/README.md`
-
-- `pending` `send-creates-new-thread`
-- `pending` `send-appends-message-to-existing-thread`
-- `pending` `send-reads-body-from-body-file`
-- `pending` `send-attaches-artifact-with-metadata`
-- `pending` `send-rejects-invalid-payload-json`
-- `pending` `send-rejects-invalid-artifact-metadata-json`
-
-### `docs/tests/inbox/fetch/README.md`
-
-- `pending` `fetch-returns-pending-thread-for-target-agent`
-- `pending` `fetch-respects-status-and-limit-filters`
-- `pending` `fetch-unread-uses-read-cursor`
-- `pending` `fetch-returns-no-matching-work-when-empty`
-
-### `docs/tests/inbox/claim/README.md`
-
-- `pending` `claim-acquires-thread-lease`
-- `pending` `claim-rejects-when-thread-missing`
-- `pending` `claim-rejects-when-thread-already-claimed`
-- `pending` `claim-records-requested-lease-duration`
-
-### `docs/tests/inbox/renew/README.md`
-
-- `pending` `renew-extends-active-lease`
-- `pending` `renew-rejects-non-owner`
-- `pending` `renew-rejects-without-active-lease`
-
-### `docs/tests/inbox/update/README.md`
-
-- `pending` `update-moves-thread-to-in-progress`
-- `pending` `update-moves-thread-to-blocked-with-payload`
-- `pending` `update-accepts-body-file-and-artifact`
-- `pending` `update-rejects-invalid-payload-json`
-- `pending` `update-rejects-non-owner`
-
-### `docs/tests/inbox/reply/README.md`
-
-- `pending` `reply-adds-answer-message`
-- `pending` `reply-supports-control-kind`
-- `pending` `reply-attaches-artifact`
-- `pending` `reply-rejects-invalid-payload-json`
-
-### `docs/tests/inbox/done/README.md`
-
-- `pending` `done-marks-thread-terminal`
-- `pending` `done-persists-result-body-and-artifact`
-- `pending` `done-rejects-non-owner`
-- `pending` `done-rejects-on-terminal-thread`
-
-### `docs/tests/inbox/fail/README.md`
-
-- `pending` `fail-marks-thread-failed`
-- `pending` `fail-persists-failure-body-and-artifact`
-- `pending` `fail-rejects-non-owner`
-- `pending` `fail-rejects-on-terminal-thread`
-
-### `docs/tests/inbox/cancel/README.md`
-
-- `pending` `cancel-marks-thread-cancelled`
-- `pending` `cancel-persists-reason-and-artifact`
-- `pending` `cancel-rejects-when-thread-missing`
-
-### `docs/tests/inbox/list/README.md`
-
-- `pending` `list-filters-by-status`
-- `pending` `list-filters-by-created-by`
-- `pending` `list-filters-by-assigned-to`
-- `pending` `list-respects-limit`
-
-### `docs/tests/inbox/show/README.md`
-
-- `pending` `show-returns-thread-and-message-history`
-- `pending` `show-includes-artifacts-per-message`
-- `pending` `show-mark-read-advances-read-cursor`
-- `pending` `show-rejects-when-thread-missing`
-
-### `docs/tests/inbox/watch/README.md`
-
-- `pending` `watch-wakes-on-matching-thread`
-- `pending` `watch-respects-status-filter`
-- `pending` `watch-times-out-with-no-activity`
-
-### `docs/tests/inbox/wait-reply/README.md`
-
-- `pending` `wait-reply-wakes-on-answer-after-message`
-- `pending` `wait-reply-can-start-from-after-event`
-- `pending` `wait-reply-times-out-when-no-reply`
+1. add the new case to the relevant `README.md`
+2. add the new slug to `Authored Case Register`
+3. update `Current Snapshot` and `Document Progress`
 
 ## Definition Of Done
 
diff --git a/docs/tests/inbox/_shared/README.md b/docs/tests/inbox/_shared/README.md
new file mode 100644
index 0000000..80ca409
--- /dev/null
+++ b/docs/tests/inbox/_shared/README.md
@@ -0,0 +1,130 @@
+# Inbox Shared Test Conventions
+
+## Purpose
+
+This document captures shared assumptions used by multiple `inbox` test-plan documents so command and workflow files can stay focused on behavior rather than repeating setup boilerplate.
+
+## Recommended Fixture Shape
+
+Use an isolated temp workspace per case:
+
+- database path: `TMPDIR/coord.db`
+- optional body file: `TMPDIR/body.md`
+- optional artifact file: `TMPDIR/artifact.txt`
+
+Recommended bootstrap command:
+
+```bash
+inbox --db TMPDIR/coord.db --json init
+```
+
+## Global Flags
+
+Root-level flags apply to every subcommand:
+
+- `--db`: SQLite database path, default `.agents/coord.db`
+- `--json`: emit machine-readable JSON
+- `--agent`: acting agent identity shortcut used by commands that accept agent context
+
+When a command-specific `--agent` or `--from` flag is omitted, the root `--agent` value may be used instead. Cases that verify fallback behavior should state that explicitly.
+
+## Success JSON Contract
+
+Successful JSON output uses this shape:
+
+```json
+{
+  "ok": true,
+  "command": "send",
+  "data": {}
+}
+```
+
+Shared assertion points:
+
+- `ok` is `true`
+- `command` matches the invoked subcommand
+- `data` contains the command-specific payload
+
+## Error JSON Contract
+
+Failure JSON output uses this shape:
+
+```json
+{
+  "ok": false,
+  "error": {
+    "code": "invalid_input",
+    "message": "..."
+  }
+}
+```
+
+Shared assertion points:
+
+- `ok` is `false`
+- `error.code` matches the stable contract
+- `error.message` is present and human-readable
+
+## Exit Code Contract
+
+The current CLI contract uses these exit codes:
+
+| Exit Code | Meaning | Typical Error Code |
+| --- | --- | --- |
+| `0` | success | none |
+| `10` | no matching work / timeout without match | `no_matching_work` |
+| `20` | lease conflict | `lease_conflict` |
+| `30` | invalid input, invalid state, usage-style error | `invalid_input` or `invalid_state` |
+| `40` | referenced thread or message missing | `not_found` |
+| `50` | unexpected internal failure | `internal_error` |
+
+When a case expects no result, assert both the exit code and the JSON error code.
+
+## Body Input Rules
+
+Commands that support `--body` and `--body-file` follow these rules:
+
+- `--body` and `--body-file` are mutually exclusive
+- `--body-file` content is read verbatim into the message body
+- unreadable `--body-file` should be treated as `invalid_input`
+
+Relevant commands:
+
+- `send`
+- `update`
+- `reply`
+- `done`
+- `fail`
+
+## Artifact Rules
+
+Commands with artifact support use these shared rules:
+
+- `--artifact` may be repeated
+- `--artifact-kind` may be specified once for all artifacts, or once per artifact
+- `--artifact-metadata-json` may be specified once for all artifacts, or once per artifact
+- `--artifact-kind` and `--artifact-metadata-json` are invalid without at least one `--artifact`
+- an empty artifact path is invalid input
+
+When artifact behavior is under test, assert at least:
+
+- artifact count
+- artifact `path`
+- artifact `kind`
+- metadata presence when supplied
+
+## Read And Unread Assertions
+
+Unread-related cases should verify behavior from the agent's point of view, not only raw message existence.
+
+Recommended checks:
+
+- `fetch --unread` returns a thread before read acknowledgement
+- `show --mark-read` clears unread state for that agent
+- a new message to the same thread makes the thread unread again
+- `wait-reply` may clear blocked unread state for the waiting agent when the reply is consumed
+
+## Workflow Authoring Rule
+
+If a case spans multiple commands, place the end-to-end narrative in `workflows/README.md` first, then add narrower command-level cases only when they introduce behavior that is easier to reason about in isolation.
diff --git a/docs/tests/inbox/cancel/README.md b/docs/tests/inbox/cancel/README.md
new file mode 100644
index 0000000..d2c00a2
--- /dev/null
+++ b/docs/tests/inbox/cancel/README.md
@@ -0,0 +1,87 @@
+# Inbox `cancel` Test Plan
+
+## Scope
+
+This document covers thread cancellation via `inbox cancel`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: cancel-marks-thread-cancelled
+
+### 用例意义
+
+验证 `cancel` 可以把非终态线程推进到 `cancelled` 终态，并生成控制消息。
+
+### 前置条件
+
+- 已存在一个非终态线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json cancel --agent leader --thread THREAD_ID --reason "Task superseded by a larger refactor"
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status == "cancelled"`
+- `message.kind == "control"`
+
+### 断言结论
+
+- `cancel` 是线程级终态转换
+- 取消时会释放活跃 lease
+
+## case: cancel-persists-reason-and-artifact
+
+### 用例意义
+
+验证 `cancel` 的原因文本与附件会被完整持久化。
+
+### 前置条件
+
+- 已存在一个非终态线程 `THREAD_ID`
+- `TMPDIR/cancel.md` 已存在
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json cancel --agent leader --thread THREAD_ID --reason "Task superseded by a larger refactor" --artifact TMPDIR/cancel.md --artifact-kind brief
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `cancel` 成功
+- 取消消息 `summary` 与 `body` 都保留取消原因
+- 取消消息包含 1 个 artifact
+
+### 断言结论
+
+- `cancel` 既保留人类可读原因，也支持附带上下文材料
+
+## case: cancel-rejects-when-thread-missing
+
+### 用例意义
+
+验证 `cancel` 对不存在线程返回稳定的 not-found 错误契约。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json cancel --agent leader --thread thr_missing
+```
+
+### 预期输出
+
+- 退出码为 `40`
+- JSON 错误码为 `not_found`
+
+### 断言结论
+
+- `cancel` 不会为缺失线程隐式创建控制消息
diff --git a/docs/tests/inbox/claim/README.md b/docs/tests/inbox/claim/README.md
new file mode 100644
index 0000000..a69f359
--- /dev/null
+++ b/docs/tests/inbox/claim/README.md
@@ -0,0 +1,112 @@
+# Inbox `claim` Test Plan
+
+## Scope
+
+This document covers lease acquisition via `inbox claim`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: claim-acquires-thread-lease
+
+### 用例意义
+
+验证 `claim` 可以把 `pending` 线程切换到 `claimed`，并生成租约事件消息。
+
+### 前置条件
+
+- 已存在一个 `pending` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json claim --agent worker-a --thread THREAD_ID --lease-seconds 300
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status == "claimed"`
+- `thread.assigned_to == "worker-a"`
+- `message.kind == "event"`
+- `message.summary == "thread claimed"`
+
+### 断言结论
+
+- `claim` 同时更新线程状态与活跃租约
+- 成功领取会附带一条事件消息，而不是静默改状态
+
+## case: claim-rejects-when-thread-missing
+
+### 用例意义
+
+验证 `claim` 对不存在的线程返回稳定的 not-found 错误契约。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json claim --agent worker-z --thread thr_missing
+```
+
+### 预期输出
+
+- 退出码为 `40`
+- JSON 错误码为 `not_found`
+
+### 断言结论
+
+- 缺失线程会被明确区分为引用错误，而不是 lease 冲突
+
+## case: claim-rejects-when-thread-already-claimed
+
+### 用例意义
+
+验证同一线程在已有活跃租约时，其他执行者无法重复领取。
+
+### 前置条件
+
+- `worker-z` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json claim --agent worker-y --thread THREAD_ID
+```
+
+### 预期输出
+
+- 退出码为 `20`
+- JSON 错误码为 `lease_conflict`
+
+### 断言结论
+
+- 活跃 lease 是 `claim` 的排他条件
+
+## case: claim-records-requested-lease-duration
+
+### 用例意义
+
+验证 `claim --lease-seconds` 的请求值会进入事件消息 payload，便于后续审计。
+
+### 前置条件
+
+- 已存在一个 `pending` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json claim --agent worker-a --thread THREAD_ID --lease-seconds 300
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `message.payload_json.lease_seconds == 300`
+- `message.payload_json.lease_token` 存在
+
+### 断言结论
+
+- 请求的租约时长不是仅用于内部计算，也会被持久化到事件消息中
diff --git a/docs/tests/inbox/done/README.md b/docs/tests/inbox/done/README.md
new file mode 100644
index 0000000..02b8f20
--- /dev/null
+++ b/docs/tests/inbox/done/README.md
@@ -0,0 +1,112 @@
+# Inbox `done` Test Plan
+
+## Scope
+
+This document covers successful terminal completion via `inbox done`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: done-marks-thread-terminal
+
+### 用例意义
+
+验证租约拥有者可以将线程推进到 `done` 终态，并生成结果消息。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json done --agent worker-a --thread THREAD_ID --summary "Retry policy implemented" --body "The HTTP client now retries the selected transient failures."
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status == "done"`
+- `message.kind == "result"`
+
+### 断言结论
+
+- `done` 会把线程推进到成功终态
+- 完成时会释放活跃 lease
+
+## case: done-persists-result-body-and-artifact
+
+### 用例意义
+
+验证 `done` 能持久化结果正文与附件，并被后续 `show` 读取。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+- `TMPDIR/result.md` 已存在
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json done --agent worker-a --thread THREAD_ID --summary "Retry policy implemented" --body-file TMPDIR/result.md --artifact TMPDIR/result.md --artifact-kind report
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `done` 成功
+- 最终结果消息 `body` 等于文件内容
+- 结果消息包含 1 个 `report` artifact
+
+### 断言结论
+
+- `done` 是结果交付命令，不只是状态切换命令
+
+## case: done-rejects-non-owner
+
+### 用例意义
+
+验证非租约拥有者不能代替执行者完成线程。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json done --agent worker-b --thread THREAD_ID --summary "Retry policy implemented"
+```
+
+### 预期输出
+
+- 退出码为 `20`
+- JSON 错误码为 `lease_conflict`
+
+### 断言结论
+
+- `done` 受活跃 lease 所属者约束
+
+## case: done-rejects-on-terminal-thread
+
+### 用例意义
+
+验证已进入终态的线程不能再次执行 `done`。
+
+### 前置条件
+
+- 线程 `THREAD_ID` 已经是 `done`、`failed` 或 `cancelled`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json done --agent worker-a --thread THREAD_ID --summary "Retry policy implemented"
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_state`
+
+### 断言结论
+
+- `done` 对终态线程是幂等失败，而不是重复成功
diff --git a/docs/tests/inbox/fail/README.md b/docs/tests/inbox/fail/README.md
new file mode 100644
index 0000000..da013de
--- /dev/null
+++ b/docs/tests/inbox/fail/README.md
@@ -0,0 +1,111 @@
+# Inbox `fail` Test Plan
+
+## Scope
+
+This document covers failure terminal completion via `inbox fail`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: fail-marks-thread-failed
+
+### 用例意义
+
+验证租约拥有者可以把线程推进到 `failed` 终态，并生成失败结果消息。
+
+### 前置条件
+
+- `worker-b` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fail --agent worker-b --thread THREAD_ID --summary "Migration failed" --body "The migration cannot proceed because the prior schema is inconsistent."
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status == "failed"`
+- `message.kind == "result"`
+
+### 断言结论
+
+- `fail` 与 `done` 共享结果消息模型，但进入的是失败终态
+
+## case: fail-persists-failure-body-and-artifact
+
+### 用例意义
+
+验证 `fail` 能持久化失败说明与附件。
+
+### 前置条件
+
+- `worker-b` 已成功 `claim` 线程 `THREAD_ID`
+- `TMPDIR/failure.md` 已存在
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fail --agent worker-b --thread THREAD_ID --summary "Migration failed" --body-file TMPDIR/failure.md --artifact TMPDIR/failure.md --artifact-kind report
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `fail` 成功
+- 最终结果消息 `body` 等于文件内容
+- 结果消息包含 1 个 `report` artifact
+
+### 断言结论
+
+- 失败终态同样要能完整交付排障材料
+
+## case: fail-rejects-non-owner
+
+### 用例意义
+
+验证非租约拥有者不能把线程标记为失败。
+
+### 前置条件
+
+- `worker-b` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fail --agent worker-x --thread THREAD_ID --summary "Migration failed"
+```
+
+### 预期输出
+
+- 退出码为 `20`
+- JSON 错误码为 `lease_conflict`
+
+### 断言结论
+
+- `fail` 与 `done` 一样受 lease owner 约束
+
+## case: fail-rejects-on-terminal-thread
+
+### 用例意义
+
+验证已进入终态的线程不能再次执行 `fail`。
+
+### 前置条件
+
+- 线程 `THREAD_ID` 已经是 `done`、`failed` 或 `cancelled`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fail --agent worker-b --thread THREAD_ID --summary "Migration failed"
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_state`
+
+### 断言结论
+
+- `fail` 对终态线程不会重复成功
diff --git a/docs/tests/inbox/fetch/README.md b/docs/tests/inbox/fetch/README.md
new file mode 100644
index 0000000..1f738de
--- /dev/null
+++ b/docs/tests/inbox/fetch/README.md
@@ -0,0 +1,117 @@
+# Inbox `fetch` Test Plan
+
+## Scope
+
+This document covers agent-scoped candidate thread retrieval via `inbox fetch`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: fetch-returns-pending-thread-for-target-agent
+
+### 用例意义
+
+验证 `fetch` 能按目标执行者拉取待处理线程。
+
+### 前置条件
+
+- `leader` 已向 `worker-a` 发送至少一个 `pending` 线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fetch --agent worker-a --status pending
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回 `data.threads`
+- 至少包含一个 `assigned_to == "worker-a"` 且 `status == "pending"` 的线程
+
+### 断言结论
+
+- `fetch` 默认是执行者视角的候选工作列表，不是全局线程扫描
+
+## case: fetch-respects-status-and-limit-filters
+
+### 用例意义
+
+验证 `fetch` 同时遵守状态过滤与返回上限。
+
+### 前置条件
+
+- `worker-a` 拥有多个不同状态的线程
+- 其中至少两个线程满足目标状态过滤条件
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fetch --agent worker-a --status pending,blocked --limit 1
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回线程数不超过 `1`
+- 返回的每条线程都满足 `status in ["pending","blocked"]`
+
+### 断言结论
+
+- `fetch` 的 `status` 与 `limit` 会同时生效
+- 返回顺序按 `updated_at` 倒序，优先暴露最新线程
+
+## case: fetch-unread-uses-read-cursor
+
+### 用例意义
+
+验证 `fetch --unread` 基于 agent 的 read cursor 计算未读，而不是仅按线程是否存在新消息。
+
+### 前置条件
+
+- `leader` 已向 `worker-e` 发送一个 `pending` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+inbox --db TMPDIR/coord.db --agent worker-e --json show --thread THREAD_ID --mark-read
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-e --thread THREAD_ID --summary "Use sentence case" --body "Keep the nav labels in sentence case."
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+```
+
+### 预期输出
+
+- 第一次 `fetch --unread` 返回该线程
+- `show --mark-read` 后，第二次 `fetch --unread` 无匹配结果
+- 新消息追加后，第三次 `fetch --unread` 再次返回该线程
+
+### 断言结论
+
+- 未读判断依赖 `thread_reads.last_read_message_id`
+- 新消息到达会让同线程重新进入未读结果集
+
+## case: fetch-returns-no-matching-work-when-empty
+
+### 用例意义
+
+验证 `fetch` 在没有匹配线程时返回稳定的“无工作”错误契约。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fetch --agent worker-z --status pending
+```
+
+### 预期输出
+
+- 退出码为 `10`
+- JSON 错误码为 `no_matching_work`
+
+### 断言结论
+
+- 空结果不是成功空数组，而是显式的“无匹配工作”信号
diff --git a/docs/tests/inbox/init/README.md b/docs/tests/inbox/init/README.md
new file mode 100644
index 0000000..526a4fb
--- /dev/null
+++ b/docs/tests/inbox/init/README.md
@@ -0,0 +1,64 @@
+# Inbox `init` Test Plan
+
+## Scope
+
+This document covers the externally visible behavior of `inbox init`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: init-creates-schema-on-empty-db
+
+### 用例意义
+
+验证在空数据库路径上执行 `init` 会创建可用的 inbox schema，并返回稳定的初始化响应。
+
+### 前置条件
+
+- 选择一个尚不存在的数据库路径 `TMPDIR/coord.db`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json init
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回 `ok=true`
+- `command` 为 `init`
+- `data.db_path` 等于传入路径
+- `data.status` 为 `initialized`
+
+### 断言结论
+
+- `init` 在空路径上可以直接完成 schema 初始化
+- 初始化结果足以让后续 `send`、`fetch` 等命令继续使用同一数据库
+
+## case: init-is-idempotent-on-existing-db
+
+### 用例意义
+
+验证 `init` 可以对已初始化过的数据库重复执行，而不会报错或破坏已有 schema。
+
+### 前置条件
+
+- `TMPDIR/coord.db` 已经执行过一次 `inbox --db TMPDIR/coord.db --json init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json init
+inbox --db TMPDIR/coord.db --json init
+```
+
+### 预期输出
+
+- 两次命令都退出码为 `0`
+- 两次响应都返回 `data.status == "initialized"`
+- 两次响应都返回相同的 `data.db_path`
+
+### 断言结论
+
+- `init` 是幂等操作
+- 对已存在 schema 的重复初始化不应引入额外迁移失败或状态漂移
diff --git a/docs/tests/inbox/list/README.md b/docs/tests/inbox/list/README.md
new file mode 100644
index 0000000..aec2382
--- /dev/null
+++ b/docs/tests/inbox/list/README.md
@@ -0,0 +1,107 @@
+# Inbox `list` Test Plan
+
+## Scope
+
+This document covers thread listing behavior via `inbox list`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: list-filters-by-status
+
+### 用例意义
+
+验证 `list --status` 只返回指定状态集合内的线程。
+
+### 前置条件
+
+- 数据库中存在多个不同状态的线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json list --status pending,blocked
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回的每条线程都满足 `status in ["pending","blocked"]`
+
+### 断言结论
+
+- `list` 是全局筛选视角，状态过滤不会被忽略
+
+## case: list-filters-by-created-by
+
+### 用例意义
+
+验证 `list --created-by` 能按线程创建者筛选结果。
+
+### 前置条件
+
+- 至少有两位不同创建者产生的线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json list --created-by leader
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回的每条线程都满足 `created_by == "leader"`
+
+### 断言结论
+
+- `created-by` 过滤条件直接作用在线程元数据上
+
+## case: list-filters-by-assigned-to
+
+### 用例意义
+
+验证 `list --assigned-to` 能按当前指派执行者筛选线程。
+
+### 前置条件
+
+- 数据库中存在多个不同 `assigned_to` 的线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json list --assigned-to worker-d --status pending
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回的每条线程都满足 `assigned_to == "worker-d"`
+
+### 断言结论
+
+- `list` 可用于管理侧查看某位执行者当前承担的线程集合
+
+## case: list-respects-limit
+
+### 用例意义
+
+验证 `list --limit` 会约束返回条数，并按更新时间倒序返回最新线程。
+
+### 前置条件
+
+- 存在多个满足过滤条件的线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json list --assigned-to worker-d --limit 1
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回线程数不超过 `1`
+
+### 断言结论
+
+- `list` 的 limit 是硬上限，不会返回超量结果
diff --git a/docs/tests/inbox/renew/README.md b/docs/tests/inbox/renew/README.md
new file mode 100644
index 0000000..fea88bf
--- /dev/null
+++ b/docs/tests/inbox/renew/README.md
@@ -0,0 +1,87 @@
+# Inbox `renew` Test Plan
+
+## Scope
+
+This document covers lease renewal behavior via `inbox renew`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: renew-extends-active-lease
+
+### 用例意义
+
+验证租约拥有者可以对活跃 lease 执行续租，并生成续租事件消息。
+
+### 前置条件
+
+- `worker-c` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json renew --agent worker-c --thread THREAD_ID --lease-seconds 600
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status` 保持原状态
+- `message.kind == "event"`
+- `message.summary == "lease renewed"`
+- `message.payload_json.lease_seconds == 600`
+
+### 断言结论
+
+- `renew` 是在原线程上追加续租事件，而不是重新 claim
+
+## case: renew-rejects-non-owner
+
+### 用例意义
+
+验证非租约拥有者不能续租别人的活跃 lease。
+
+### 前置条件
+
+- `worker-c` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json renew --agent worker-x --thread THREAD_ID --lease-seconds 600
+```
+
+### 预期输出
+
+- 退出码为 `20`
+- JSON 错误码为 `lease_conflict`
+
+### 断言结论
+
+- `renew` 与 `claim` 一样受 lease owner 约束
+
+## case: renew-rejects-without-active-lease
+
+### 用例意义
+
+验证线程没有活跃租约时，`renew` 会明确失败。
+
+### 前置条件
+
+- 已存在线程 `THREAD_ID`
+- 该线程当前没有活跃 lease
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json renew --agent worker-c --thread THREAD_ID --lease-seconds 600
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_state`
+
+### 断言结论
+
+- `renew` 依赖已有活跃租约
+- 没有 lease 属于状态错误，不是 not-found
diff --git a/docs/tests/inbox/reply/README.md b/docs/tests/inbox/reply/README.md
new file mode 100644
index 0000000..20e5ad9
--- /dev/null
+++ b/docs/tests/inbox/reply/README.md
@@ -0,0 +1,111 @@
+# Inbox `reply` Test Plan
+
+## Scope
+
+This document covers reply behavior within an existing thread via `inbox reply`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: reply-adds-answer-message
+
+### 用例意义
+
+验证 `reply` 默认会向现有线程追加一条 `answer` 消息，并保持线程状态不变。
+
+### 前置条件
+
+- 已存在一个非终态线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-a --thread THREAD_ID --summary "Retry read timeouts" --body "Yes, include read timeouts in the retry policy."
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `message.kind == "answer"`
+- `thread.thread_id == THREAD_ID`
+- 线程状态保持原值
+
+### 断言结论
+
+- `reply` 是线程内追加消息，而不是状态转换命令
+
+## case: reply-supports-control-kind
+
+### 用例意义
+
+验证 `reply --kind control` 可以发送控制类消息，而不局限于默认 `answer`。
+
+### 前置条件
+
+- 已存在一个非终态线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-a --thread THREAD_ID --kind control --summary "Pause rollout" --body "Pause rollout until QA confirms the fix."
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `message.kind == "control"`
+
+### 断言结论
+
+- `reply` 的消息种类可由调用方显式指定
+
+## case: reply-attaches-artifact
+
+### 用例意义
+
+验证 `reply` 支持追加带附件的答复消息。
+
+### 前置条件
+
+- 已存在一个非终态线程 `THREAD_ID`
+- `TMPDIR/decision.md` 已存在
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-a --thread THREAD_ID --summary "Retry read timeouts" --artifact TMPDIR/decision.md --artifact-kind brief --artifact-metadata-json '{"label":"decision"}'
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `message.artifacts` 长度为 `1`
+- artifact 路径、kind、metadata 都可读
+
+### 断言结论
+
+- `reply` 与 `send/update/done/fail` 共享附件写入契约
+
+## case: reply-rejects-invalid-payload-json
+
+### 用例意义
+
+验证 `reply` 对非法 `--payload-json` 输入返回稳定错误契约。
+
+### 前置条件
+
+- 已存在一个非终态线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-a --thread THREAD_ID --summary "Retry read timeouts" --payload-json not-json
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_input`
+
+### 断言结论
+
+- `reply` 的 payload 与其他消息写入命令一样需要通过 JSON 校验
diff --git a/docs/tests/inbox/send/README.md b/docs/tests/inbox/send/README.md
new file mode 100644
index 0000000..9b8faff
--- /dev/null
+++ b/docs/tests/inbox/send/README.md
@@ -0,0 +1,176 @@
+# Inbox `send` Test Plan
+
+## Scope
+
+This document covers thread creation and message append behavior exposed by `inbox send`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: send-creates-new-thread
+
+### 用例意义
+
+验证 `send` 在未指定既有线程时会创建新线程，并写入首条任务消息。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-a --subject "Implement feature X" --summary "Add retry policy" --body "Implement retry handling for the HTTP client." --run run_blog_001 --task T1
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回 `thread.thread_id`
+- `thread.status == "pending"`
+- `thread.created_by == "leader"`
+- `thread.assigned_to == "worker-a"`
+- `message.kind == "task"`
+
+### 断言结论
+
+- `send` 会新建线程而不是只插入孤立消息
+- 新线程的默认初始状态是 `pending`
+
+## case: send-appends-message-to-existing-thread
+
+### 用例意义
+
+验证 `send` 在指定既有 `--thread` 时会向原线程追加消息，而不是重建线程。
+
+### 前置条件
+
+- 已存在一个由 `leader` 发给 `worker-d` 的线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-d --thread THREAD_ID --summary "Use a markdown editor" --body "Prefer a textarea-based markdown editor for v1."
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `send` 成功，返回的 `thread.thread_id` 仍为 `THREAD_ID`
+- 线程状态保持原值，不被强制改写为新状态
+- `show` 可见消息数增加
+
+### 断言结论
+
+- 追加消息不会重置线程生命周期
+- 线程历史按时间顺序保留旧消息与新消息
+
+## case: send-reads-body-from-body-file
+
+### 用例意义
+
+验证 `send --body-file` 会把文件内容写入消息正文。
+
+### 前置条件
+
+- `TMPDIR/task.md` 已存在，内容为测试正文
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-d --subject "Build admin editor" --summary "Create the first editor screen" --body-file TMPDIR/task.md
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `send` 成功
+- `show` 首条消息的 `body` 与文件内容一致
+
+### 断言结论
+
+- `body-file` 内容会被原样读取
+- 该行为与直接传 `--body` 的最终存储结果等价
+
+## case: send-attaches-artifact-with-metadata
+
+### 用例意义
+
+验证 `send` 支持附带 artifact、kind 和 metadata，并可在返回值或后续 `show` 中读取。
+
+### 前置条件
+
+- `TMPDIR/task.md` 已存在
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-d --subject "Build admin editor" --summary "Create the first editor screen" --artifact TMPDIR/task.md --artifact-kind brief --artifact-metadata-json '{"label":"task-brief"}'
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `message.artifacts` 长度为 `1`
+- artifact `path == "TMPDIR/task.md"`
+- artifact `kind == "brief"`
+- artifact `metadata_json.label == "task-brief"`
+
+### 断言结论
+
+- `send` 可以在创建消息时持久化附件及其结构化元数据
+
+## case: send-rejects-invalid-payload-json
+
+### 用例意义
+
+验证 `send` 对非法 `--payload-json` 输入给出稳定错误契约。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-z --subject "Invalid payload json" --payload-json not-json
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_input`
+
+### 断言结论
+
+- 非法 payload 在写库前就会被拒绝
+- 错误归类为输入问题，而不是内部错误
+
+## case: send-rejects-invalid-artifact-metadata-json
+
+### 用例意义
+
+验证 `send` 对非法 artifact metadata JSON 给出稳定错误契约。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-z --subject "Invalid artifact json" --artifact TMPDIR/report.md --artifact-metadata-json not-json
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_input`
+
+### 断言结论
+
+- artifact metadata 会在写入前校验 JSON 合法性
+
+## Notes
+
+- 新建线程时未显式传 `--summary`，会回退到 `--subject`
+- `--body` 与 `--body-file` 互斥；该约束由 shared 文档统一说明
diff --git a/docs/tests/inbox/show/README.md b/docs/tests/inbox/show/README.md
new file mode 100644
index 0000000..913f9ff
--- /dev/null
+++ b/docs/tests/inbox/show/README.md
@@ -0,0 +1,115 @@
+# Inbox `show` Test Plan
+
+## Scope
+
+This document covers per-thread detail retrieval via `inbox show`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: show-returns-thread-and-message-history
+
+### 用例意义
+
+验证 `show` 会返回线程详情和完整消息历史。
+
+### 前置条件
+
+- 已存在一个含多条消息的线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 返回 `data.thread`
+- 返回 `data.messages`
+- 消息按创建时间升序排列
+
+### 断言结论
+
+- `show` 是线程详情与时间序历史的读取入口
+
+## case: show-includes-artifacts-per-message
+
+### 用例意义
+
+验证 `show` 返回的每条消息都包含其关联 artifact 列表。
+
+### 前置条件
+
+- 线程 `THREAD_ID` 中至少一条消息附带 artifact
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- 相关消息节点包含 `artifacts`
+- artifact 的 `path`、`kind`、`metadata_json` 可读
+
+### 断言结论
+
+- `show` 需要把附件一并展开，而不是只返回 message 基本字段
+
+## case: show-mark-read-advances-read-cursor
+
+### 用例意义
+
+验证 `show --mark-read` 会推进调用 agent 的 read cursor，并影响后续 unread 查询。
+
+### 前置条件
+
+- `worker-e` 有一个未读线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --agent worker-e --json show --thread THREAD_ID --mark-read
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+```
+
+### 预期输出
+
+- `show` 成功
+- 随后的 `fetch --unread` 对该线程不再返回结果
+
+### 断言结论
+
+- `mark-read` 的副作用是推进该 agent 的 `last_read_message_id`
+
+## case: show-rejects-when-thread-missing
+
+### 用例意义
+
+验证 `show` 对不存在线程返回稳定的 not-found 错误契约。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json show --thread thr_missing
+```
+
+### 预期输出
+
+- 退出码为 `40`
+- JSON 错误码为 `not_found`
+
+### 断言结论
+
+- `show` 不会对缺失线程返回空对象
+
+## Notes
+
+- 使用 `--mark-read` 时必须提供 agent 身份，可通过根级 `--agent` 或命令参数传入
diff --git a/docs/tests/inbox/update/README.md b/docs/tests/inbox/update/README.md
new file mode 100644
index 0000000..e9bd049
--- /dev/null
+++ b/docs/tests/inbox/update/README.md
@@ -0,0 +1,139 @@
+# Inbox `update` Test Plan
+
+## Scope
+
+This document covers progress and blocked transitions via `inbox update`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: update-moves-thread-to-in-progress
+
+### 用例意义
+
+验证租约拥有者可以把线程推进到 `in_progress`，并生成进度消息。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json update --agent worker-a --thread THREAD_ID --status in_progress --summary "Implementation started" --body "Scanning current HTTP client usage."
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status == "in_progress"`
+- `message.kind == "progress"`
+- `message.to_agent` 指向线程创建者
+
+### 断言结论
+
+- `update` 会把状态推进和消息追加合并为同一次事务
+
+## case: update-moves-thread-to-blocked-with-payload
+
+### 用例意义
+
+验证 `update --status blocked` 会写入阻塞问题消息，并保留结构化 payload。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json update --agent worker-a --thread THREAD_ID --status blocked --summary "Need timeout decision" --payload-json '{"question":"Should retries apply to read timeouts?"}'
+```
+
+### 预期输出
+
+- 命令退出码为 `0`
+- `thread.status == "blocked"`
+- `message.kind == "question"`
+- `message.payload_json.question` 保存提问内容
+
+### 断言结论
+
+- `blocked` 更新会生成面向创建者的问题消息
+
+## case: update-accepts-body-file-and-artifact
+
+### 用例意义
+
+验证 `update` 支持通过 `body-file` 与 artifact 发送结构化进度材料。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+- `TMPDIR/progress.md` 已存在
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json update --agent worker-a --thread THREAD_ID --status in_progress --summary "Implementation started" --body-file TMPDIR/progress.md --artifact TMPDIR/progress.md --artifact-kind note
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `update` 成功
+- 对应消息 `body` 等于文件内容
+- 对应消息包含 1 个 artifact，kind 为 `note`
+
+### 断言结论
+
+- `update` 的正文与 artifact 支持与 `send/reply/done/fail` 保持一致
+
+## case: update-rejects-invalid-payload-json
+
+### 用例意义
+
+验证 `update` 对非法 `--payload-json` 输入返回稳定错误契约。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json update --agent worker-a --thread THREAD_ID --status blocked --summary "Need timeout decision" --payload-json not-json
+```
+
+### 预期输出
+
+- 退出码为 `30`
+- JSON 错误码为 `invalid_input`
+
+### 断言结论
+
+- 阻塞问题的 payload 需要满足合法 JSON 约束
+
+## case: update-rejects-non-owner
+
+### 用例意义
+
+验证非租约拥有者不能更新线程状态。
+
+### 前置条件
+
+- `worker-a` 已成功 `claim` 线程 `THREAD_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json update --agent worker-b --thread THREAD_ID --status in_progress --summary "Implementation started"
+```
+
+### 预期输出
+
+- 退出码为 `20`
+- JSON 错误码为 `lease_conflict`
+
+### 断言结论
+
+- `update` 明确依赖活跃 lease 所属者
diff --git a/docs/tests/inbox/wait-reply/README.md b/docs/tests/inbox/wait-reply/README.md
new file mode 100644
index 0000000..4c65a16
--- /dev/null
+++ b/docs/tests/inbox/wait-reply/README.md
@@ -0,0 +1,93 @@
+# Inbox `wait-reply` Test Plan
+
+## Scope
+
+This document covers blocking reply wait behavior within one thread via `inbox wait-reply`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: wait-reply-wakes-on-answer-after-message
+
+### 用例意义
+
+验证 `wait-reply` 可以从某条已知消息之后开始等待，并在答复到达后唤醒。
+
+### 前置条件
+
+- `worker-c` 已拥有一个 `blocked` 线程 `THREAD_ID`
+- 阻塞消息的 `message_id` 为 `BLOCKED_MESSAGE_ID`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --agent worker-c --json wait-reply --thread THREAD_ID --after-message BLOCKED_MESSAGE_ID --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-c --thread THREAD_ID --summary "Redirect to login" --body "Redirect guests to login for the MVP."
+```
+
+### 预期输出
+
+- `wait-reply` 退出码为 `0`
+- `wait-reply.data.woke == true`
+- 返回的 `message.kind == "answer"`
+
+### 断言结论
+
+- `wait-reply` 可以可靠地从既知消息边界之后等待后续答复
+
+## case: wait-reply-can-start-from-after-event
+
+### 用例意义
+
+验证 `wait-reply --after-event` 支持从既知事件游标之后恢复等待。
+
+### 前置条件
+
+- 已通过先前的 `watch` 或 `wait-reply` 结果拿到某个 `NEXT_EVENT_ID`
+- 线程 `THREAD_ID` 后续还会收到新的回复类消息
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --agent worker-c --json wait-reply --thread THREAD_ID --after-event NEXT_EVENT_ID --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-c --thread THREAD_ID --summary "Redirect to login" --body "Redirect guests to login for the MVP."
+```
+
+### 预期输出
+
+- `wait-reply` 在事件游标之后的新回复出现时被唤醒
+- 返回新的 `next_event_id`
+
+### 断言结论
+
+- `after-event` 允许等待逻辑在断点之后继续，而不会重复消费旧回复
+
+## case: wait-reply-times-out-when-no-reply
+
+### 用例意义
+
+验证在超时时间内没有匹配回复出现时，`wait-reply` 返回稳定超时契约。
+
+### 前置条件
+
+- 存在一个线程 `THREAD_ID`
+- 不会有新的 `answer/control/result` 消息到达
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --agent worker-c --json wait-reply --thread THREAD_ID --timeout-seconds 1
+```
+
+### 预期输出
+
+- 退出码为 `10`
+- JSON 错误码为 `no_matching_work`
+
+### 断言结论
+
+- `wait-reply` 超时被视为“没有等到匹配回复”
+
+## Notes
+
+- 默认唤醒 kinds 为 `answer,control,result`
+- 当返回消息是发给等待 agent 的外来消息时，`wait-reply` 会顺带推进该 agent 的 read cursor
diff --git a/docs/tests/inbox/watch/README.md b/docs/tests/inbox/watch/README.md
new file mode 100644
index 0000000..170a7d7
--- /dev/null
+++ b/docs/tests/inbox/watch/README.md
@@ -0,0 +1,91 @@
+# Inbox `watch` Test Plan
+
+## Scope
+
+This document covers blocking thread-activity wait behavior via `inbox watch`.
+
+Shared conventions live in [../_shared/README.md](../_shared/README.md).
+
+## case: watch-wakes-on-matching-thread
+
+### 用例意义
+
+验证 `watch` 在新匹配线程到达时会被唤醒，并返回线程、消息与事件信息。
+
+### 前置条件
+
+- `worker-d` 当前没有匹配 `pending` 线程
+- `watch` 先于 `send` 启动
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json watch --agent worker-d --status pending --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-d --subject "Build admin editor" --summary "Create the first editor screen"
+```
+
+### 预期输出
+
+- `watch` 退出码为 `0`
+- `watch.data.woke == true`
+- 返回 `thread`、`message`、`event`
+
+### 断言结论
+
+- `watch` 唤醒结果不仅说明“醒了”，还提供触发该唤醒的具体事件上下文
+
+## case: watch-respects-status-filter
+
+### 用例意义
+
+验证 `watch --status` 只会对匹配状态的后续事件唤醒。
+
+### 前置条件
+
+- 存在一个会被推进到 `blocked` 的线程 `THREAD_ID`
+- `watch` 以 `--status blocked` 先启动
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json watch --agent worker-c --status blocked --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json update --agent worker-c --thread THREAD_ID --status blocked --summary "Need policy decision"
+```
+
+### 预期输出
+
+- `watch` 只在线程进入 `blocked` 后返回
+- 返回的 `thread.status == "blocked"`
+
+### 断言结论
+
+- `watch` 的状态过滤作用在“事件发生后的线程状态”上
+
+## case: watch-times-out-with-no-activity
+
+### 用例意义
+
+验证在超时时间内没有匹配活动时，`watch` 返回稳定超时契约。
+
+### 前置条件
+
+- 没有新匹配事件会发生
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json watch --agent worker-d --status pending --timeout-seconds 1
+```
+
+### 预期输出
+
+- 退出码为 `10`
+- JSON 错误码为 `no_matching_work`
+
+### 断言结论
+
+- `watch` 超时被归类为“无匹配工作”，而不是内部错误
+
+## Notes
+
+- 未传 `--after-event` 时，`watch` 默认从“当前时刻之后”开始等待，不会回放既有事件
diff --git a/docs/tests/inbox/workflows/README.md b/docs/tests/inbox/workflows/README.md
new file mode 100644
index 0000000..11faddd
--- /dev/null
+++ b/docs/tests/inbox/workflows/README.md
@@ -0,0 +1,276 @@
+# Inbox Workflow Test Plan
+
+## Scope
+
+This document tracks cross-command scenarios where the main value is the interaction between multiple `inbox` subcommands.
+
+All examples assume:
+
+- isolated temp database
+- `inbox --db TMPDIR/coord.db --json init` already executed
+- assertions follow the shared rules in [../_shared/README.md](../_shared/README.md)
+
+## case: thread-lifecycle-happy-path
+
+### 用例意义
+
+验证 `send -> fetch -> claim -> update(in_progress) -> update(blocked) -> reply -> done -> show` 的主干链路可用，且线程与消息历史一致。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+- 发送方为 `leader`
+- 执行方为 `worker-a`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-a --subject "Implement feature X" --summary "Add retry policy" --body "Implement retry handling for the HTTP client." --run run_blog_001 --task T1
+inbox --db TMPDIR/coord.db --json fetch --agent worker-a --status pending
+inbox --db TMPDIR/coord.db --json claim --agent worker-a --thread THREAD_ID --lease-seconds 300
+inbox --db TMPDIR/coord.db --json update --agent worker-a --thread THREAD_ID --status in_progress --summary "Implementation started" --body "Scanning current HTTP client usage."
+inbox --db TMPDIR/coord.db --json update --agent worker-a --thread THREAD_ID --status blocked --summary "Need timeout decision" --payload-json '{"question":"Should retries apply to read timeouts?"}'
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-a --thread THREAD_ID --summary "Retry read timeouts" --body "Yes, include read timeouts in the retry policy."
+inbox --db TMPDIR/coord.db --json done --agent worker-a --thread THREAD_ID --summary "Retry policy implemented" --body "The HTTP client now retries the selected transient failures."
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `send` 返回新建线程，线程状态为 `pending`
+- `fetch` 返回唯一匹配线程
+- `claim` 后线程状态为 `claimed`
+- 第一次 `update` 后线程状态为 `in_progress`
+- 第二次 `update` 后线程状态为 `blocked`
+- `reply` 返回一条 `kind=answer` 的消息
+- `done` 后线程状态为 `done`
+- `show` 返回线程状态 `done`，并包含完整消息历史
+
+### 断言结论
+
+- 全链路所有命令退出码为 `0`
+- `show.data.thread.status == "done"`
+- `show.data.messages` 长度为 `6`
+- 历史中的状态推进顺序与执行顺序一致，不出现丢消息或状态回退
+
+## case: blocked-question-reply-resume-to-done
+
+### 用例意义
+
+验证被阻塞线程在收到答复后可以继续推进，并最终进入完成态。
+
+### 前置条件
+
+- 已存在由 `leader` 发给 `worker-c` 的线程
+- `worker-c` 已经成功 `claim`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json update --agent worker-c --thread THREAD_ID --status blocked --summary "Need policy decision" --body "Should guest users be redirected to login or shown a 403 page?"
+inbox --db TMPDIR/coord.db --agent worker-c --json wait-reply --thread THREAD_ID --after-message BLOCKED_MESSAGE_ID --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-c --thread THREAD_ID --summary "Redirect to login" --body "Redirect guests to login for the MVP."
+inbox --db TMPDIR/coord.db --json done --agent worker-c --thread THREAD_ID --summary "Policy applied" --body "The flow now redirects guests to login."
+```
+
+### 预期输出
+
+- `update` 将线程推进到 `blocked`
+- `wait-reply` 在答复出现后唤醒
+- 唤醒结果包含答复消息
+- `done` 成功将线程推进到 `done`
+
+### 断言结论
+
+- `wait-reply.data.woke == true`
+- `wait-reply.data.message.kind == "answer"`
+- 最终 `done.data.thread.status == "done"`
+- 该用例强调“阻塞后可恢复”，不是单纯验证 reply 本身
+
+## case: fail-lifecycle-from-claim-to-terminal
+
+### 用例意义
+
+验证线程在被领取后可以直接进入失败终态，并且 `show` 对终态读取一致。
+
+### 前置条件
+
+- 空数据库已完成 `init`
+- `leader` 已向 `worker-b` 发送任务
+- `worker-b` 已 `claim` 该线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fail --agent worker-b --thread THREAD_ID --summary "Migration failed" --body "The migration cannot proceed because the prior schema is inconsistent."
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `fail` 返回线程状态 `failed`
+- `show` 返回相同终态
+
+### 断言结论
+
+- `fail.data.thread.status == "failed"`
+- `show.data.thread.status == "failed"`
+- 失败消息保留在线程历史中，可被后续排障读取
+
+## case: cancel-lifecycle-after-worker-claim
+
+### 用例意义
+
+验证线程在执行者已领取后，发起方仍可以取消任务，并进入 `cancelled` 终态。
+
+### 前置条件
+
+- `leader` 已向 `worker-c` 发送任务
+- `worker-c` 已成功 `claim`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json cancel --agent leader --thread THREAD_ID --reason "Task superseded by a larger refactor"
+```
+
+### 预期输出
+
+- `cancel` 成功
+- 返回线程状态 `cancelled`
+- 返回的消息记录取消原因
+
+### 断言结论
+
+- `cancel.data.thread.status == "cancelled"`
+- 取消属于终态转换，不要求执行者先主动释放 lease
+- 原因字段可被后续 `show` 或审计场景消费
+
+## case: watch-wakes-then-fetch-sees-new-thread
+
+### 用例意义
+
+验证 `watch` 的等待语义与 `fetch --unread` 的可见性一致，确保新线程到达时执行者既会被唤醒，也能随后拉到未读任务。
+
+### 前置条件
+
+- `worker-d` 尚无匹配 `pending` 线程
+- `watch` 先于 `send` 启动
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json watch --agent worker-d --status pending --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-d --subject "Build admin editor" --summary "Create the first editor screen" --body-file TMPDIR/task.md --artifact TMPDIR/task.md --artifact-kind brief --artifact-metadata-json '{"label":"task-brief"}' --run run_blog_004 --task T4
+inbox --db TMPDIR/coord.db --json fetch --agent worker-d --status pending --unread
+```
+
+### 预期输出
+
+- `watch` 因新线程到达而唤醒
+- 唤醒结果中的 `thread_id` 与 `send` 返回值一致
+- 随后 `fetch --unread` 仍能看到该 `pending` 线程
+
+### 断言结论
+
+- `watch.data.woke == true`
+- `watch.data.thread.thread_id == send.data.thread.thread_id`
+- `fetch.data.threads` 长度为 `1`
+- `watch` 唤醒不应提前消费掉线程的未读可见性
+
+## case: artifact-visible-through-send-and-show
+
+### 用例意义
+
+验证 `send` 写入的 body-file 与 artifact 信息能被后续 `show` 完整读回。
+
+### 前置条件
+
+- `TMPDIR/task.md` 已存在，内容为测试任务正文
+- 空数据库已完成 `init`
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-d --subject "Build admin editor" --summary "Create the first editor screen" --body-file TMPDIR/task.md --artifact TMPDIR/task.md --artifact-kind brief --artifact-metadata-json '{"label":"task-brief"}' --run run_blog_004 --task T4
+inbox --db TMPDIR/coord.db --json show --thread THREAD_ID
+```
+
+### 预期输出
+
+- `send` 成功创建线程并附带一条 artifact
+- `show` 的首条消息包含从文件读取的正文与 artifact 列表
+
+### 断言结论
+
+- 首条消息 `body` 等于 `TMPDIR/task.md` 的文件内容
+- 首条消息 `artifacts` 长度为 `1`
+- 首个 artifact 的 `path` 等于 `TMPDIR/task.md`
+- 首个 artifact 的 `kind` 等于 `brief`
+
+## case: unread-clears-after-mark-read-and-reappears-on-new-message
+
+### 用例意义
+
+验证 read cursor 的最关键用户感知行为：未读任务可被显式清空，并会在同线程新消息到达后重新出现。
+
+### 前置条件
+
+- `leader` 已向 `worker-e` 发送一个 `pending` 线程
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+inbox --db TMPDIR/coord.db --agent worker-e --json show --thread THREAD_ID --mark-read
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+inbox --db TMPDIR/coord.db --json send --from leader --to worker-e --thread THREAD_ID --summary "Use sentence case" --body "Keep the nav labels in sentence case."
+inbox --db TMPDIR/coord.db --json fetch --agent worker-e --status pending --unread
+```
+
+### 预期输出
+
+- 第一次 `fetch --unread` 返回该线程
+- `show --mark-read` 成功推进 `worker-e` 的 read cursor
+- 第二次 `fetch --unread` 无匹配结果
+- 新消息追加后，第三次 `fetch --unread` 再次返回该线程
+
+### 断言结论
+
+- 第一次 `fetch` 返回 1 条线程
+- 第二次 `fetch` 退出码为 `10`，错误码为 `no_matching_work`
+- 追加消息后第三次 `fetch` 再次返回 1 条线程
+- 未读状态是按 agent 视角计算，而不是线程级布尔值
+
+## case: wait-reply-clears-blocked-unread-for-agent
+
+### 用例意义
+
+验证等待答复的消费者在收到答复后，其阻塞线程未读状态会被消费，避免“已经处理过回复但列表仍显示未读”的错觉。
+
+### 前置条件
+
+- `worker-c` 已拥有一个 `blocked` 线程
+- 该线程阻塞消息对应的 `message_id` 已知
+- `worker-c` 使用 `wait-reply` 等待答复
+
+### 输入
+
+```bash
+inbox --db TMPDIR/coord.db --agent worker-c --json wait-reply --thread THREAD_ID --after-message BLOCKED_MESSAGE_ID --timeout-seconds 2
+inbox --db TMPDIR/coord.db --json reply --from leader --to worker-c --thread THREAD_ID --summary "Redirect to login" --body "Redirect guests to login for the MVP."
+inbox --db TMPDIR/coord.db --agent worker-c --json fetch --status blocked --unread
+```
+
+### 预期输出
+
+- `wait-reply` 在答复后唤醒
+- 唤醒结果携带 `answer` 消息
+- 随后的 `fetch --status blocked --unread` 不再返回该线程
+
+### 断言结论
+
+- `wait-reply.data.woke == true`
+- `wait-reply.data.message.kind == "answer"`
+- 后续 `fetch` 退出码为 `10`
+- 对等待中的 agent 来说，答复消费与未读清理是同一条用户契约链路