overlord-cli 0.2608210739.0 → 0.2608211014.0

package/dist/connectors/AGENTS.md

@@ -80,11 +80,12 @@ stale. `yarn connectors:check` runs it alongside the connector version check.
 
 ### The rendered agent-session hook
 
-Mechanical observe / decide / inject is **not a live Overlord connector path**. Latch owns
-turns, tools, `awaiting_input`, and PTY answers via `latch events` and `latch send`. New
-adapters must not register `ovld agent-session` hooks. Channel 1 stays on `ovld protocol`
-(follow-up capture, touched-file attribution, Stop delivery reminder). The rest of this
-section documents leftover CLI runtime and historical fixtures.
+Mechanical observe / decide / inject is **not a live Overlord connector path**. Latch v2
+removed the CLI event and send surface that previously backed it, and Overlord does not
+emulate that protocol. New adapters must not register `ovld agent-session` hooks. Channel 1
+stays on `ovld protocol` (follow-up capture, touched-file attribution, Stop delivery reminder).
+A future conversation surface must be a native Latch v2 Conversation Hub client. The rest of
+this section documents leftover CLI runtime and historical fixtures.
 
 Do not write a per-adapter shell script for agent-session traffic. One core script,
 `connectors/core/scripts/agent-session-hook.sh`, is rendered per adapter at `ovld agent-setup`

package/dist/connectors/VERSION

@@ -1 +1 @@
-0.3.32
+0.3.33

package/dist/connectors/adapters/antigravity/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://antigravity.google/schemas/v1/plugin.json",
   "name": "overlord",
-  "version": "0.3.32",
+  "version": "0.3.33",
   "description": "Chat-driven access to Overlord mission protocol workflows through skills and MCP.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/claude/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "overlord",
-  "version": "0.3.32",
+  "version": "0.3.33",
   "description": "Overlord mission protocol workflow for Claude Code — attach/update/ask/deliver, slash commands, and hook-driven activity capture.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/claude/CAPABILITIES.md

@@ -6,7 +6,7 @@
 
 **Harness version verified** `2.1.227` · **range** `>=2.1.0` · **scheme** `semver`
 
-**Descriptor digest** `833a35cecb73aa9b34b40036eb6d3088dd729ca6be465d1ee228dae2c5a8269d`
+**Descriptor digest** `15ea657c37363b74db085466a97f38db044892abf2e28465b1c7bcd6cc4ce927`
 
 > The tier is derived from passing fixtures, never authored. `unsupported` means the harness
 > cannot do it — do not attempt it. `not-implemented` means it is buildable and unbuilt: that is
@@ -63,15 +63,15 @@ Per-command `timeout` in seconds with no documented ceiling. Omitting `decision`
 
 ### Capability notes
 
-- **`observe.prompt`** — Two registrations used to ride UserPromptSubmit. Mechanical agent-session observation moved to Latch `events`. This capability grades the protocol follow-up hook that posts `user_follow_up` mission events; that Channel 1 path stays installed.
+- **`observe.prompt`** — Two registrations used to ride UserPromptSubmit. Latch v2 removed the CLI event stream that previously owned mechanical agent-session observation. This capability grades the protocol follow-up hook that posts `user_follow_up` mission events; that Channel 1 path stays installed.
 - **`observe.toolCall`** — Deliberately unregistered in Phase 1. PreToolUse is decision-capable: registering it for observation alone would put a script in front of every tool call for no benefit a PostToolUse registration does not already provide, while inheriting the blast radius of the decision path. It lands with the codecs that can actually use it.
-- **`observe.toolResult`** — Mechanical tool-result observation moved to Latch `events`. The protocol PostToolUse hook still records touched files for delivery attribution.
+- **`observe.toolResult`** — Latch v2 removed the CLI event stream that previously owned mechanical tool-result observation. The protocol PostToolUse hook still records touched files for delivery attribution.
 - **`observe.fileEdit`** — The shipped PostToolUse hook that records touched files into the per-session log used by `deliver` for change attribution remains registered. Mechanical agent-session observation is no longer a live Overlord path.
-- **`observe.sessionLifecycle`** — SessionStart observation moved to Latch `events`. There is no matching Overlord registration.
-- **`decide.universal`** — Remote permission decisions moved to Latch `awaiting_input` / `latch send --resolve`. The native Claude prompt owns approvals when Latch is absent.
+- **`observe.sessionLifecycle`** — Latch v2 removed the CLI event stream that previously owned session observation. There is no matching Overlord registration.
+- **`decide.universal`** — Latch v2 removed the CLI pending-input and send surface. The native Claude prompt owns approvals; Overlord exposes no remote decision control.
 - **`answer.structuredQuestion`** — Elicitation/ElicitationResult hook events exist in the binary but their request and response shapes were not read. Write the fixture before relying on them.
 - **`answer.persistentAllow`** — Structurally available, deliberately unbuilt. A standing permission grant needs its own design, confirmation, audit language, and RBAC review before any remote surface offers it.
-- **`inject.midTurn`** — Session-input injection moved off Overlord connectors. Latch PTY write is Phase 3.
+- **`inject.midTurn`** — Session-input injection is not a live Overlord connector path. A future conversation surface must use the native Latch v2 Conversation Hub protocol.
 - **`inject.turnBoundary`** — Turn-boundary injection moved off Overlord connectors. The protocol Stop hook still reminds the agent to deliver when needed.
 - **`inject.nextTurn`** — No distinct next-turn-only path yet; turn-boundary Stop covers the portable fallback.
 - **`terminal.statusSurface`** — `statusMessage` is static install-time configuration with no per-request interpolation and no session-binding predicate, so it renders in unbound sessions too. It must be worded as an option, never as an instruction.

package/dist/connectors/adapters/claude/README.md

@@ -87,7 +87,8 @@ This connector is intentionally reviewable against the four connector layers in
 - `commands/*.md` — slash commands for session routing, objective discussion, mission creation, and work recording.
 - `hooks/hooks.json` + `scripts/*.sh` — Channel 1 mission follow-up (`UserPromptSubmit`),
   touched-file capture (`PostToolUse`), and Stop delivery reminder. Mechanical harness
-  observation, permission, and injection are Latch's (`latch events` / PTY).
+  observation, permission, and injection are not live Overlord connector paths under Latch v2;
+  native Claude prompts retain control.
 - `userConfig` for `overlord_url` and `user_token`; the hook scripts use the `ovld protocol` CLI and pass the plugin's USER_TOKEN to that CLI as `Overlord_USER_TOKEN`.
 - `conformance-manifest.yaml` — connector conformance declaration for the Overlord contract.
 - `prompt-wrapper.md` — Claude launch context-file guidance.

package/dist/connectors/adapters/claude/conformance-manifest.yaml

@@ -15,7 +15,7 @@ connector:
   harnessCapabilities:
     path: connectors/adapters/claude/harness-capabilities.yaml
     schemaVersion: 1
-    digest: "833a35cecb73aa9b34b40036eb6d3088dd729ca6be465d1ee228dae2c5a8269d"
+    digest: "15ea657c37363b74db085466a97f38db044892abf2e28465b1c7bcd6cc4ce927"
   capabilities:
     - followUpHook
     - permissionHook

package/dist/connectors/adapters/claude/harness-capabilities.yaml

@@ -50,9 +50,10 @@ capabilities:
     native: UserPromptSubmit
     fixtures: [fixtures/normalize-user-prompt-submit.json]
     notes: >-
-      Two registrations used to ride UserPromptSubmit. Mechanical agent-session observation
-      moved to Latch `events`. This capability grades the protocol follow-up hook that posts
-      `user_follow_up` mission events; that Channel 1 path stays installed.
+      Two registrations used to ride UserPromptSubmit. Latch v2 removed the CLI event stream
+      that previously owned mechanical agent-session observation. This capability grades the
+      protocol follow-up hook that posts `user_follow_up` mission events; that Channel 1 path
+      stays installed.
   observe.toolCall:
     status: not-implemented
     native: PreToolUse
@@ -67,8 +68,9 @@ capabilities:
     native: PostToolUse
     trackedAs: latch-engine
     notes: >-
-      Mechanical tool-result observation moved to Latch `events`. The protocol PostToolUse
-      hook still records touched files for delivery attribution.
+      Latch v2 removed the CLI event stream that previously owned mechanical tool-result
+      observation. The protocol PostToolUse hook still records touched files for delivery
+      attribution.
   observe.fileEdit:
     status: supported
     native: PostToolUse
@@ -84,8 +86,8 @@ capabilities:
     native: SessionStart
     trackedAs: latch-engine
     notes: >-
-      SessionStart observation moved to Latch `events`. There is no matching Overlord
-      registration.
+      Latch v2 removed the CLI event stream that previously owned session observation. There
+      is no matching Overlord registration.
   decide.shell:
     status: not-implemented
     native: PermissionRequest
@@ -107,8 +109,8 @@ capabilities:
     native: PermissionRequest
     trackedAs: latch-engine
     notes: >-
-      Remote permission decisions moved to Latch `awaiting_input` / `latch send --resolve`.
-      The native Claude prompt owns approvals when Latch is absent.
+      Latch v2 removed the CLI pending-input and send surface. The native Claude prompt owns
+      approvals; Overlord exposes no remote decision control.
   answer.structuredQuestion:
     status: unverified
     native: Elicitation
@@ -128,7 +130,8 @@ capabilities:
     native: asyncRewake
     trackedAs: latch-engine
     notes: >-
-      Session-input injection moved off Overlord connectors. Latch PTY write is Phase 3.
+      Session-input injection is not a live Overlord connector path. A future conversation
+      surface must use the native Latch v2 Conversation Hub protocol.
   inject.turnBoundary:
     status: not-implemented
     native: Stop

package/dist/connectors/adapters/codex/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "overlord",
-  "version": "0.3.32",
+  "version": "0.3.33",
   "description": "Chat-driven access to Overlord mission protocol workflows with hook-driven activity capture.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/codex/CAPABILITIES.md

@@ -6,7 +6,7 @@
 
 **Harness version verified** `0.147.0` · **range** `>=0.124.0` · **scheme** `semver`
 
-**Descriptor digest** `ede31118a90f1afc3d8e934b7fd921759faac4afca1a08e4214a1104ee487b1d`
+**Descriptor digest** `9743c504d8d11ffec193751273102d9a1277293e55f43aef8914da4b05ddfa23`
 
 > The tier is derived from passing fixtures, never authored. `unsupported` means the harness
 > cannot do it — do not attempt it. `not-implemented` means it is buildable and unbuilt: that is
@@ -64,9 +64,9 @@ Per-hook timeouts in seconds; the documentation gives a 600-second default. The
 
 ### Capability notes
 
-- **`observe.toolCall`** — Mechanical PreToolUse observation moved to Latch `events`.
+- **`observe.toolCall`** — Latch v2 removed the CLI event stream that previously owned mechanical observation. Overlord exposes no replacement until it has a native Conversation Hub client.
 - **`observe.fileEdit`** — The lifecycle PostToolUse registration pipes Codex's native payload to `ovld protocol record-touched`, which appends normalized changed paths to the per-session attribution log read by `deliver`. It is intentionally separate from mechanical agent-session observation, which remains owned by Latch.
-- **`decide.universal`** — Remote permission decisions moved to Latch `awaiting_input` / `latch send --resolve`.
+- **`decide.universal`** — Latch v2 removed the CLI pending-input and send surface. The native Codex prompt owns approvals; Overlord exposes no remote decision control.
 - **`answer.structuredQuestion`** — `codex app-server` exposes ToolRequestUserInput with per-option label/description and an autoResolutionMs — an ACP-shaped structured question in everything but name. It is unreachable until the single-subscriber question is settled.
 - **`inject.midTurn`** — ThreadInjectItems/TurnSteer exist on the experimental app-server surface only. The binary carries "expected exactly one client subscribed to the thread, found {}", so if the TUI is that one client Overlord cannot also subscribe and this path is closed.
 

package/dist/connectors/adapters/codex/README.md

@@ -72,7 +72,7 @@ This connector is intentionally reviewable against the four connector layers in
 - `.codex-plugin/hooks.json` — Channel 1 `UserPromptSubmit` follow-up capture through
   `ovld protocol hook-event` and `PostToolUse` delivery attribution through
   `ovld protocol record-touched`. Mechanical harness observation, permission, and injection
-  are Latch's (`latch events` / PTY), not Overlord connector registrations.
+  are not live Overlord connector paths under Latch v2; native Codex prompts retain control.
 - `.mcp.json` + `scripts/overlord-mcp.mjs` — MCP bridge to common `ovld protocol` operations. The shim is **generated**: it is rendered from `connectors/core/scripts/overlord-mcp.mjs` at setup time with the adapter key substituted, and there is no copy in this directory to edit.
 - `assets/` — branded plugin assets for Codex install surfaces.
 - `conformance-manifest.yaml` — connector conformance declaration for the Overlord contract.

package/dist/connectors/adapters/codex/conformance-manifest.yaml

@@ -15,7 +15,7 @@ connector:
   harnessCapabilities:
     path: connectors/adapters/codex/harness-capabilities.yaml
     schemaVersion: 1
-    digest: "ede31118a90f1afc3d8e934b7fd921759faac4afca1a08e4214a1104ee487b1d"
+    digest: "9743c504d8d11ffec193751273102d9a1277293e55f43aef8914da4b05ddfa23"
   capabilities:
     - followUpHook
     - permissionHook

package/dist/connectors/adapters/codex/harness-capabilities.yaml

@@ -56,7 +56,8 @@ capabilities:
     native: PreToolUse
     trackedAs: latch-engine
     notes: >-
-      Mechanical PreToolUse observation moved to Latch `events`.
+      Latch v2 removed the CLI event stream that previously owned mechanical observation.
+      Overlord exposes no replacement until it has a native Conversation Hub client.
   observe.toolResult:
     status: not-implemented
     native: PostToolUse
@@ -98,7 +99,8 @@ capabilities:
     native: PermissionRequest
     trackedAs: latch-engine
     notes: >-
-      Remote permission decisions moved to Latch `awaiting_input` / `latch send --resolve`.
+      Latch v2 removed the CLI pending-input and send surface. The native Codex prompt owns
+      approvals; Overlord exposes no remote decision control.
   answer.structuredQuestion:
     status: unverified
     native: ToolRequestUserInput

package/dist/connectors/adapters/cursor/.cursor-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "overlord",
-  "version": "0.3.32",
+  "version": "0.3.33",
   "description": "Overlord mission protocol workflow for Cursor — attach/update/ask/deliver, slash commands, MCP bridge, and follow-up capture.",
   "author": {
     "name": "Cooperativ",

package/dist/connectors/adapters/cursor/CAPABILITIES.md

@@ -6,7 +6,7 @@
 
 **Harness version verified** `2026.08.04-aaa8809` · **range** `>=2026.07.01` · **scheme** `calendar`
 
-**Descriptor digest** `7085c25291f5c96cf3c7fb9c280281fa698b5c026e9859377ad8c53d1cd82537`
+**Descriptor digest** `48b805b1d1cd28ccde10a839b0eb2a85eb237d82b46e3d54b21f671997b0c870`
 
 > The tier is derived from passing fixtures, never authored. `unsupported` means the harness
 > cannot do it — do not attempt it. `not-implemented` means it is buildable and unbuilt: that is
@@ -65,7 +65,7 @@ The 60-second per-script default is an order of magnitude below any window sized
 
 ### Capability notes
 
-- **`observe.toolCall`** — Mechanical preToolUse observation moved to Latch `events`. Native Cursor permission prompts own approvals when Latch is absent.
+- **`observe.toolCall`** — Latch v2 removed the CLI event stream that previously owned mechanical observation. Native Cursor permission prompts own approvals; Overlord exposes no replacement until it has a native Conversation Hub client.
 - **`observe.fileEdit`** — Carried on postToolUse rather than on Cursor's dedicated `afterFileEdit`, for the same reason Claude carries it on PostToolUse: one registration then covers every mutation, including the ones a shell command makes, and the split into `file.edited` happens where the normalized tool name is already known. `afterFileEdit` remains available if a future need appears for edit-level granularity the tool-level event cannot express.
 - **`observe.sessionLifecycle`** — Cursor fires the event, but no shipped code in this repo has ever read one of its payloads, so the field carrying the equivalent of Claude's `source` is unknown. A codec rule on a guessed path would not fail loudly — an unresolved path yields a card with no detail — so this stays unbuilt until a real payload is recorded, rather than being claimed on an assumed field name.
 - **`inject.turnBoundary`** — `stop` returns `followup_message` for the protocol delivery reminder. Session-input injection is no longer a live Overlord path.

package/dist/connectors/adapters/cursor/README.md

@@ -77,7 +77,7 @@ This connector is intentionally reviewable against the four connector layers in
 - `commands/*.md` — slash commands for session routing, objective discussion, mission creation, and work recording.
 - `hooks/overlord-user-prompt-submit.sh` — `beforeSubmitPrompt` follow-up capture through `ovld protocol hook-event`.
 - `hooks/overlord-post-tool-use.sh` — records `postToolUse` file edits and shell-mediated changes for exact per-session delivery attribution.
-- `hooks/overlord-stop.sh` — checks pending delivery through `ovld protocol hook-event`; when needed, it asks Cursor to continue once so the agent can finish delivery. Mechanical harness observation, permission, and injection are Latch's (`latch events` / PTY), not Overlord connector registrations.
+- `hooks/overlord-stop.sh` — checks pending delivery through `ovld protocol hook-event`; when needed, it asks Cursor to continue once so the agent can finish delivery. Mechanical harness observation, permission, and injection are not live Overlord connector paths under Latch v2; native Cursor prompts retain control.
 - `rules/overlord-local.mdc` — always-on workflow rules for Cursor sessions.
 - `mcp.json` + `scripts/overlord-mcp.mjs` — MCP bridge to common `ovld protocol` operations. The shim is **generated**: it is rendered from `connectors/core/scripts/overlord-mcp.mjs` at setup time with the adapter key substituted, and there is no copy in this directory to edit.
 - `conformance-manifest.yaml` — connector conformance declaration for the Overlord contract.
@@ -92,4 +92,4 @@ The shared Connector Core source lives at `connectors/core/overlord-mission`. `o
 
 See the [adapter capability matrix](../../README.md#adapter-capability-matrix) for how this compares across adapters.
 
-Cursor hook execution is fail-open by default, so Overlord protocol-hook failures do not strand the agent. Permission and question prompts are Latch `awaiting_input` when Latch hosts the session.
+Cursor hook execution is fail-open by default, so Overlord protocol-hook failures do not strand the agent. Permission and question prompts remain with Cursor; Overlord has no Latch v2 conversation client yet.

package/dist/connectors/adapters/cursor/conformance-manifest.yaml

@@ -15,7 +15,7 @@ connector:
   harnessCapabilities:
     path: connectors/adapters/cursor/harness-capabilities.yaml
     schemaVersion: 1
-    digest: "7085c25291f5c96cf3c7fb9c280281fa698b5c026e9859377ad8c53d1cd82537"
+    digest: "48b805b1d1cd28ccde10a839b0eb2a85eb237d82b46e3d54b21f671997b0c870"
   capabilities:
     - followUpHook
     - stopHook

package/dist/connectors/adapters/cursor/harness-capabilities.yaml

@@ -58,8 +58,9 @@ capabilities:
     native: preToolUse
     trackedAs: latch-engine
     notes: >-
-      Mechanical preToolUse observation moved to Latch `events`. Native Cursor permission
-      prompts own approvals when Latch is absent.
+      Latch v2 removed the CLI event stream that previously owned mechanical observation.
+      Native Cursor permission prompts own approvals; Overlord exposes no replacement until it
+      has a native Conversation Hub client.
   observe.toolResult:
     status: supported
     native: postToolUse

package/dist/connectors/adapters/pi/CAPABILITIES.md

@@ -6,7 +6,7 @@
 
 **Harness version verified** `0.83.0` · **range** `>=0.83.0 <0.84.0` · **scheme** `semver`
 
-**Descriptor digest** `d083b9610578e1af2e73b273d0661430213fbbe5ff24550e8f0815c284733ae1`
+**Descriptor digest** `0166eb18d8913ea41704d3aa9afbf9164d994ae16089927e38f681be5dd005ff`
 
 > The tier is derived from passing fixtures, never authored. `unsupported` means the harness
 > cannot do it — do not attempt it. `not-implemented` means it is buildable and unbuilt: that is
@@ -64,7 +64,7 @@ Remote decision hold moved to Latch. Pi has no native prompt; Overlord no longer
 
 - **`observe.prompt`** — The shipped extension captures `input` and posts follow-up activity through `ovld protocol hook-event`. Mechanical agent-session observation is not a live path.
 - **`decide.universal`** — Remote tool-call decisions moved to Latch. Pi no longer intercepts tools on the Overlord connector path.
-- **`inject.midTurn`** — Session-input injection moved off the Overlord connector. Latch PTY write is Phase 3.
+- **`inject.midTurn`** — Session-input injection is not a live Overlord connector path. A future conversation surface must use the native Latch v2 Conversation Hub protocol.
 
 ## Hazards
 

package/dist/connectors/adapters/pi/conformance-manifest.yaml

@@ -15,7 +15,7 @@ connector:
   harnessCapabilities:
     path: connectors/adapters/pi/harness-capabilities.yaml
     schemaVersion: 1
-    digest: "d083b9610578e1af2e73b273d0661430213fbbe5ff24550e8f0815c284733ae1"
+    digest: "0166eb18d8913ea41704d3aa9afbf9164d994ae16089927e38f681be5dd005ff"
   capabilities:
     - followUpHook
     - nativeResume

package/dist/connectors/adapters/pi/harness-capabilities.yaml

@@ -103,7 +103,8 @@ capabilities:
     native: sendUserMessage(deliverAs=steer)
     trackedAs: latch-engine
     notes: >-
-      Session-input injection moved off the Overlord connector. Latch PTY write is Phase 3.
+      Session-input injection is not a live Overlord connector path. A future conversation
+      surface must use the native Latch v2 Conversation Hub protocol.
   inject.turnBoundary:
     status: not-implemented
     native: sendUserMessage(deliverAs=followUp)

package/dist/connectors/core/overlord-mission/reference/cli.md

@@ -181,7 +181,7 @@ When creating missions from within a repository:
 - Add objectives to the same mission when each prompt is a sequential step toward the same feature or goal; use `ovld protocol add-objectives --mission-id <mission_id> --objectives-json '[{"objective":"..."}]'`.
 - `create` and `prompt` require `--objectives-json` or `--objectives-file` with an ordered array of `{ "objective": "...", "title": "...", "autoAdvance": true }` objects. A single objective is just an array with one item. `--auto-advance` / `--no-auto-advance` map each opted-in item to authoritative Run Queue membership (default off).
 - `add-objectives` uses the same per-item `autoAdvance` field and `--auto-advance` / `--no-auto-advance` mapping.
-- Use `ovld protocol queue-objective --objective-id <id> [--queue <id|name>] [--after <queued-entry|objective>|--front|--position <rank>]` to explicitly enqueue or move work, `dequeue-objective --objective-id <id>` to remove it, and `run-queue --project-id <id|slug|name>` to inspect every queue.
+- Use `ovld protocol run-queue [--project-id <id|slug|name>|--objective-id <id>|--mission-id <id>] [--queue <id|name>]` to inspect live queues. Entry operations use `queue-objective --objective-id <id> [--queue <id|name>] [--after <queued-entry|objective>|--front|--position <rank>]`, `dequeue-objective --objective-id <id>`, and `reorder-run-queue --queue <id|name> --ordered-entries-json <json>`; they require `execution_request:create`. Queue-definition operations use `create-run-queue`, `update-run-queue`, `delete-run-queue`, and `reorder-project-run-queues`; they require `project:update` and therefore a full-scope token.
 - `ovld protocol update-objective --objective-id <id> --auto-advance` / `--no-auto-advance` is retained as an enqueue/dequeue compatibility mapping. `autoAdvance` is deprecated and derived from `queueEntry`; legacy writes retire next release and column removal requires a later contract bump.
 - `record-work` creates exactly one **completed** objective from a single `--objective` (or positional / an `objective` field in `--payload-json`) plus a `--summary` and file-change data — it does not take `--objectives-json`. See [record-work.md](record-work.md).
 

package/dist/connectors/core/overlord-mission/reference/mcp.md

@@ -1,6 +1,6 @@
 # MCP And API Defaults
 
-- Every mission-scoped tool accepts `objectiveId` (hosted MCP) / `objective_id` (local shim) alongside `missionId`, taking an objective UUID or a display id such as `coo:756.k7xm`. A display id already names its mission, so `missionId` is **optional** whenever one is supplied — `overlord_load_mission_context`, `overlord_attach_session`, `overlord_update_session`, `overlord_deliver_session`, `overlord_add_artifact`, `overlord_add_objectives`, and `overlord_update_artifact` all resolve the mission from it. An objective **UUID** names no mission and still needs `missionId`. On `overlord_load_mission_context` and `overlord_attach_session` the objective is also a *pin*: it selects which objective to read or execute, which is the only way to address a mission that is running objectives in parallel.
+- Every mission-scoped tool accepts `objectiveId` (hosted MCP) / `objective_id` (local shim) alongside `missionId`, taking an objective UUID or a display id such as `coo:756.k7xm`. A display id already names its mission, so `missionId` is **optional** whenever one is supplied — `overlord_load_mission_context`, `overlord_attach_session`, `overlord_update_session`, `overlord_deliver_session`, `overlord_add_artifact`, `overlord_add_objectives`, and `overlord_update_artifact` all resolve the mission from it. An objective **UUID** names no mission and still needs `missionId`. On `overlord_load_mission_context` and `overlord_attach_session` the objective is also a _pin_: it selects which objective to read or execute, which is the only way to address a mission that is running objectives in parallel.
 - API requires `agentIdentifier` and `connectionMethod` on attach/connect/prompt. Connector adapters set the appropriate agent identifier; the CLI uses `cli` as the connection method, and the MCP tool defaults to `mcp`. Override with `--agent` / `--method` when calling from a different runtime.
 - Hosted Overlord MCP (`/mcp`) uses camelCase JSON keys (`missionId`, `sessionKey`, `deviceFingerprint`) matching `POST /api/protocol/*` bodies. The local shim uses snake_case keys mapped to CLI flags (`mission_id`, `session_key`, `device_fingerprint`). Some local-only operations (`connect`, `revert`, `prompt`, runner queue tools) intentionally remain out of hosted MCP for now.
 - `discover_project` accepts an explicit project id (`projectId` hosted, `project_id` local shim) or a working directory (`workingDirectory` hosted, `working_directory` local shim). When resolving by directory and a device fingerprint is available, include `deviceFingerprint` / `device_fingerprint` so resource-directory matching prefers the current device.
@@ -10,6 +10,6 @@
 - `overlord_record_work` (hosted MCP) and `ovld protocol record-work` (CLI) record already-completed chat work as a single review-column mission with a completed objective, file-change rationales, and the standard Gemini delivery summary — no `attach`/`deliver`. Hosted MCP requires `projectId` (it never chooses a project implicitly); the CLI resolves the project from the working directory or `--project-id`. The submission format is in [record-work.md](record-work.md). The route is `POST /api/protocol/record-work`.
 - `overlord_add_artifact` (hosted MCP) and `ovld protocol add-artifact` (CLI) create a mission artifact during a turn without delivering. Required: `type` / `--type`, `label` / `--label`, mission scope (`missionId` / `--mission-id`, or an objective display id as `objectiveId` / `--objective-id`), plus at least one of `contentText` / `--content-text`/`--content-text-file` or `externalUrl` / `--external-url`. Optional `sessionKey` / `--session-key` stamps session/objective provenance; `objectiveId` / `--objective-id` does the same when there is no live session, and the session key wins if both are given. Same service as REST `POST /api/missions/:id/artifacts`. The route is `POST /api/protocol/add-artifact`.
 - `overlord_update_artifact` (hosted MCP) and `ovld protocol update-artifact` (CLI) revise an existing mission artifact's label, Markdown content, and/or HTTP(S) URL in place. Required: `artifactId` / `--artifact-id`, `expectedRevision` / `--expected-revision`, and mission scope (`missionId` / `--mission-id`, or an objective display id as `objectiveId` / `--objective-id`). Provide at least one editable field. No session key is required. Same service as REST `PATCH /api/missions/:id/artifacts/:artifactId`. The route is `POST /api/protocol/update-artifact`.
-- `overlord_queue_objective` accepts `objectiveId`, optional `queue` / `after`, and `remove`; it forwards through Protocol to authoritative Run Queue membership and never directly launches work. `overlord_create_mission` / `overlord_add_objectives` optional `autoAdvance`, `overlord_update_objective.autoAdvance`, and `ovld protocol update-objective --auto-advance|--no-auto-advance` map to that same membership. `autoAdvance` is deprecated and derived from `queueEntry`; legacy storage writes retire next release and column removal requires a later contract bump. Instruction text remains editable only in `draft` or `future` state. The routes are `POST /api/protocol/queue-objective`, `dequeue-objective`, and `update-objective`.
+- Run Queue MCP stays a thin Protocol surface: `overlord_list_run_queues` reads compact/full state; `overlord_reorder_run_queue` atomically replaces one queue's complete entry order; and `overlord_queue_objective` accepts `objectiveId`, optional `queue` / `after` / `front` / `position`, and `remove`. `overlord_manage_run_queue` creates, updates, deletes, and reorders queue definitions. Entry operations use `execution_request:create`; definition operations use `project:update`, so they require a full-scope token. `overlord_create_mission` / `overlord_add_objectives` optional `autoAdvance`, `overlord_update_objective.autoAdvance`, and `ovld protocol update-objective --auto-advance|--no-auto-advance` map to that same membership. `autoAdvance` is deprecated and derived from `queueEntry`; legacy storage writes retire next release and column removal requires a later contract bump. Instruction text remains editable only in `draft` or `future` state. The routes are `POST /api/protocol/run-queue`, `reorder-run-queue`, `queue-objective`, `dequeue-objective`, `create-run-queue`, `update-run-queue`, `delete-run-queue`, `reorder-project-run-queues`, and `update-objective`.
 - Objective attachment tools follow the `<verb>_<noun>` MCP naming: `list_attachments`, `prepare_attachment_upload`, `finalize_attachment_upload`, `get_attachment_download_url`, `upload_attachment_file`. CLI commands use `attachment-*`. `mission_id` / `missionId` is optional for attachment calls when the objective or attachment id lets the server derive mission scope.
 - "Artifacts" may be published mid-turn with `add-artifact` / `overlord_add_artifact`, revised with `update-artifact` / `overlord_update_artifact`, or attached at `deliver` time (next_steps, test_results, migration, decision, note, url) — not user-uploaded files. Files attached by users live on objectives via the attachment tools above.