opencode-thrum-pro 0.1.9 → 0.1.11

package/assets/commands/thrum-sleep-extended.md

@@ -155,7 +155,7 @@ relocation, or teardown error) is signaled by the RPC returning an explicit
 error message BEFORE the pane dies — that is the only failure signal
 observable from inside this pane. If independent confirmation is needed, it
 must come from a DIFFERENT caller (this pane will be gone) checking
-`thrum agent list` / `thrum team list` afterward.
+`thrum team` afterward.
 
 **Do not fall back to a raw `tmux kill-session`.** A raw kill bypasses
 `agent.sleep`'s relocate-before-kill ordering and daemon bookkeeping

package/assets/commands/thrum-sleep.md

@@ -156,7 +156,7 @@ relocation, or teardown error) is signaled by the RPC returning an explicit
 error message BEFORE the pane dies — that is the only failure signal
 observable from inside this pane. If independent confirmation is needed, it
 must come from a DIFFERENT caller (this pane will be gone) checking
-`thrum agent list` / `thrum team list` afterward.
+`thrum team` afterward.
 
 **Do not fall back to a raw `tmux kill-session`.** A raw kill bypasses
 `agent.sleep`'s relocate-before-kill ordering and daemon bookkeeping

package/assets/skills/choosing-subagent-models/SKILL.md

@@ -127,7 +127,7 @@ An agent definition can also carry a default (Claude example — resolve to
 your own runtime's `mechanical`-tier value if you are not Claude):
 
 ```yaml
-# claude-plugin/agents/message-listener.md
+# claude-plugin/agents/message-listener.md (non-tmux fallback only)
 name: message-listener
 model: sonnet
 effort: low

package/assets/skills/configure-roles/SKILL.md

@@ -45,7 +45,7 @@ Run these commands and collect the output (suppress errors):
 thrum runtime list 2>/dev/null          # Installed runtimes
 git worktree list 2>/dev/null           # Worktrees and branches
 bd stats 2>/dev/null                    # Beads task tracker state
-thrum agent list --context 2>/dev/null  # Registered agents
+thrum team --all --include-inactive --compact 2>/dev/null # Fleet roster, including offline, stale and retired agents
 ls .claude/skills/ 2>/dev/null          # Installed Claude skills
 thrum config show 2>/dev/null           # Thrum configuration
 ```

package/assets/skills/coordinator-assessing-agent-completion/SKILL.md

@@ -52,7 +52,7 @@ orphaned (reap).
 | How long has it been quiet? | transcript mtime — **weak, see below** |
 | What is it waiting for? | its own last message, read as prose |
 
-**Do NOT use `thrum team`, `thrum agent list`, `phase`, or `last_seen_at` for any
+**Do NOT use `thrum team`, `phase`, or `last_seen_at` for any
 of this.** They are stale, and they fail toward
 reporting a live agent as absent.
 

package/assets/skills/coordinator-context-monitoring/SKILL.md

@@ -377,12 +377,13 @@ apply the coordinator-equivalent 75% floor above while working). The >85%
 bracket still fires regardless. These are EXPLICIT DECLARED states, not a
 heuristic guess at coordination intent — a consultant/on-call/done-planning
 researcher must set one of these prefixes ITSELF via
-`thrum agent set-intent "warm-hold: <reason>"` (or `on-call:`/`parked:`).
+`thrum session set-intent "warm-hold: <reason>"` (or `on-call:`/`parked:`).
 
-**A coordinator CANNOT set another agent's intent.** `thrum agent set-intent`
+**A coordinator CANNOT set another agent's intent.** `thrum session set-intent`
 takes TEXT only — it sets the intent for the CURRENT SESSION and accepts no
 agent argument and no `--to`. To get an intent onto another agent, message that
-agent and have it set its own; then verify by reading `thrum agent list --json`
+agent and have it set its own; then verify in the focused fleet view with
+`thrum team @AGENT --offline --include-inactive --json`
 rather than by counting acks.
 
 **Background-child-aware busy detection:** an agent with an OPEN in-progress bead, a recent JSONL tool_use
@@ -608,8 +609,9 @@ coordinator manually invoking the skill, run these guards BEFORE firing a
 restart:
 
 1. **Verify the daemon is reachable**:
-   `thrum team --json | jq '.members | length'` — if 0 or error, daemon is down;
-   skip the sweep, surface to operator.
+   Run `thrum daemon status`; on error, skip the sweep and surface to operator.
+   Inspect the local online roster with `thrum team --json` (`.team.members`);
+   use `--offline --include-inactive` for offline, stale and retired rows.
 2. **Confirm the monitor is alive**: `thrum monitor list` should show
    `context-monitoring` in `running` status with a non-empty `SCHEDULE` column.
    If absent or dead, the scheduled ALERTs aren't firing and the skill must be
@@ -654,7 +656,7 @@ When a restart fires (Step 4 or 5 — tmux-send nudge or snapshot-gated restart)
 1. Surface a brief status note to the coordinator: "Restarted @\<agent_name\>
    at \<ctx\>% — snapshot at
    `.thrum/agents/<agent>/sessions/<timestamp>-restart.md`".
-2. Wait for the agent to come back online (`thrum team` shows them active again,
+2. Wait for the agent to come back online (`thrum team @AGENT --offline --include-inactive` shows them online again,
    or their pane shows the runtime prompt).
 3. Re-send their current dispatch with the full scope + plan refs + AC targets —
    treat them as a fresh implementer who needs the full briefing again.

package/assets/skills/coordinator-dispatching-work/SKILL.md

@@ -41,7 +41,8 @@ identity drift, broken nudges, and silent message loss.
 worktree shows a registered agent, communicate with it via
 `thrum send --to @<agent_name> --body-file msg.md` instead. Sub-agents are for
 research/explore, code review, and message listeners running in the main repo —
-never for implementation work in another agent's worktree.
+use message listeners only when tmux pane delivery is unavailable, and never
+for implementation work in another agent's worktree.
 
 ## Dispatch via `thrum send` after launch — not via tmux send-keys
 

package/assets/skills/coordinator-managing-state-and-lifecycle/SKILL.md

@@ -135,7 +135,7 @@ thrum agent delete <agent>         # SEPARATE reap, after the worktree is gone
 ```
 
 `worktree teardown` does NOT delete the agent: it is preparation for deleting
-one. The row stays `retired` (hidden from default `agent list` / `team` /
+one. The row stays `retired` (hidden from the default team roster /
 overview / UI; visible with `--include-inactive` / `--phase retired`) until
 `thrum agent delete` or the sweeper reaps it. `agent delete` refuses an agent
 whose linked worktree still exists or whose process (pid/tmux) is live, and

package/assets/skills/coordinator-merging-code/SKILL.md

@@ -323,7 +323,7 @@ covered in the runner's preamble; the agent/session half needs the same check.
 thrum tmux status | grep -E '^\s*g[0-9]+'    # gate sessions, agentless ones included
 ```
 
-🔴 **`thrum team` and `thrum agent list` CANNOT SEE an agentless session.** Five of
+🔴 **`thrum team` CANNOT SEE an agentless session.** Five of
 those nine had no registered agent, so a registry-based enumeration returns a
 confident answer that omits most of the population. **Enumerate sessions, not
 agents.**

package/assets/skills/coordinator-rolling-out-to-nodes/SKILL.md

@@ -61,6 +61,10 @@ Build once per artifact on its `artifact.build_host` via `artifact.build_command
 
 ## 5. The staged order — canary, waves, prepare-and-hold, promotion gate
 
+> 🔴 **If any node fails during the rollout — canary or any wave — STOP THE ROLLOUT.** Do not roll, restart, or recommend
+> restarting any further node (including the build host / primary) until the failure is root-caused and fixed or the
+> owner explicitly decides to continue. A node that went down and was recovered by hand is a FAILURE, not a green.
+
 - **Canary first, alone.** One node (adapter `topology.canary`) rolls first. Its health must be observed green before any wave proceeds.
 - **Waves with coordinated dependency/order/rollback.** Subsequent waves follow `topology.waves`. When related daemons share ordering (e.g., connected-repo daemons that must restart together), respect the adapter-defined dependency and restart order; a failure in one dependent daemon triggers coordinated rollback per adapter `rollback` / `backup.restore_command`.
 - **Prepare-and-hold + release BY NAME.** If the mechanism supports it, prepare all nodes in a wave (acquire + verify artifact, take BEFORE snapshot, but do not restart) then release by explicit node name, not by broadcast.

package/assets/skills/coordinator-running-brainstorm-cycles/SKILL.md

@@ -198,7 +198,7 @@ For each `*-brainstorm` worktree that looks relevant, inspect:
 | Idle + domain-relevant researcher found | **REUSE** — send the Phase 2 briefing to the existing agent in its current worktree. Skip the rest of Phase 1. |
 | No idle relevant researcher             | Fall through → proceed with "Pick the base branch" below.                                                      |
 
-When reusing, have the agent itself run `thrum agent set-intent "<new one-line
+When reusing, have the agent itself run `thrum session set-intent "<new one-line
 topic>"` so `thrum team` output stays accurate (there is no command to set
 ANOTHER agent's intent — `set-intent` acts on the caller), then send the Phase 2
 briefing as normal. The existing

package/assets/skills/project-setup/SKILL.md

@@ -716,8 +716,8 @@ thrum daemon status
 cat .thrum/redirect
 cat .beads/redirect
 
-# Agent registered
-thrum agent list --context
+# Verify this agent across the fleet, including offline, stale and retired rows
+thrum team @AGENT --offline --include-inactive
 ```
 
 ### Step 5: Update CLAUDE.md
@@ -823,7 +823,7 @@ worktree-related values come from the Phase 3 assignments:
 | `{{AGENT_NAME}}`       | **From Phase 3 agent registration**                                                                                                                                                                                         |
 | `{{PLAN_FILE}}`        | **Absolute path** to the plan file (primary input)                                                                                                                                                                          |
 | `{{ANTI_PATTERNS}}`    | Generated in Step 1.5; refer to the `project-philosophy` skill for the anti-pattern format spec.                                                                                                                            |
-| `{{SUPERVISOR_NAME}}`  | From `thrum team` — first agent with role=orchestrator; if none, first with role=coordinator; if none, ask user                                                                                                             |
+| `{{SUPERVISOR_NAME}}`  | From the local online roster (`thrum team`) — first agent with role=orchestrator; if none, first with role=coordinator; if none, ask user                                                                                                             |
 | `{{CROSS_EPIC_DEPS}}`  | From Phase 2 cross-epic dependency map. If no cross-epic deps, replace with "No cross-epic dependencies."                                                                                                                   |
 
 **IMPORTANT — Absolute paths for gitignored files:** `{{DESIGN_DOC}}`,

package/assets/skills/project-setup/resources/implementation-agent.md

@@ -151,9 +151,8 @@ names. Always send to agent names (e.g., `--to @coord_main`), not role names
 (e.g., `--to @coordinator`). Sending to a role fans out to ALL agents with that
 role.
 
-**Use the message listener** — Spawn a background listener to get async
-notifications. Re-arm it every time it returns (both MESSAGES_RECEIVED and
-NO_MESSAGES_TIMEOUT).
+Messages arrive in your pane. Check `thrum inbox --unread` at natural
+breakpoints. Never spawn a background listener.
 
 When your work is complete (Phase 4), send a completion message prefixed with
 the appropriate **status token** (see Status Vocabulary at the start of Phase
@@ -568,7 +567,7 @@ Heads up: I'm modifying internal/daemon/rpc.go which may overlap with your work
 EOF
 
 # Update your intent when switching tasks
-thrum agent set-intent "Working on {{TASK_ID}}: <description>"
+thrum session set-intent "Working on {{TASK_ID}}: <description>"
 ```
 
 ---

package/assets/skills/thrum/SKILL.md

@@ -124,7 +124,7 @@ thrum <cmd> --help                       Detailed command usage
 | ---------------------------------------- | -------------------------- | ----------------------------- |
 | Cross-worktree messaging                 | Same-session task tracking | Single-agent, no coordination |
 | Persistent messages (survive compaction) | Ephemeral task lists       | Temporary scratch notes       |
-| Background listener pattern              | Inline progress tracking   | Simple linear execution       |
+| Non-tmux listener fallback               | Inline progress tracking   | Simple linear execution       |
 | Multi-machine sync via git               | Local to conversation      | No persistence needed         |
 
 **Decision test:** "Do messages need to survive session restart or reach agents
@@ -139,13 +139,14 @@ for cross-session/cross-worktree coordination messages.
 
 When running in a tmux-managed session, messages are delivered directly to your
 pane via daemon nudge — zero token cost, no background sub-agent needed. See
-[TMUX_SESSIONS.md](resources/TMUX_SESSIONS.md).
+[TMUX_SESSIONS.md](resources/TMUX_SESSIONS.md). Check `thrum inbox --unread` at
+natural breakpoints; never spawn a background listener.
 
 ### Listener Pattern (Fallback)
 
-When tmux is not available, use the background message listener. It calls
-`thrum wait` (blocking) and returns when messages arrive. Re-arm after
-processing. See [LISTENER_PATTERN.md](resources/LISTENER_PATTERN.md).
+Use the background message listener only when tmux pane delivery is unavailable.
+It calls `thrum wait` (blocking) and returns when messages arrive. See
+[LISTENER_PATTERN.md](resources/LISTENER_PATTERN.md).
 
 ### Context Management
 

package/assets/skills/thrum/resources/ANTI_PATTERNS.md

@@ -11,18 +11,14 @@ between agents.
 ## 2. Polling Instead of Waiting
 
 **Wrong:** Looping `thrum inbox` every 10 seconds. **Right:** Use `thrum wait`
-which blocks efficiently until a message arrives. Or use the message-listener
-sub-agent pattern for background monitoring.
+which blocks efficiently until a message arrives. Use the listener fallback
+only when tmux pane delivery is unavailable.
 
-## 3. Forgetting to Re-arm the Listener
+## 3. Spawning Listeners in Tmux Sessions
 
-**Wrong:** Processing messages from listener, then continuing work without
-re-arming. **Right:** After processing listener results, always spawn a new
-message-listener:
-
-```text
-Task(subagent_type="message-listener", run_in_background=true, prompt="...")
-```
+In tmux-managed sessions, messages arrive in the pane. Check
+`thrum inbox --unread` at natural breakpoints and never spawn a background
+listener.
 
 ## 4. Sending Without an Explicit Recipient Flag
 

package/assets/skills/thrum/resources/CLI_REFERENCE.md

@@ -22,7 +22,7 @@ These flags apply to every command:
 Many commands append contextual hints after their normal output. Two kinds:
 
 - **detection** — condition-triggered guards (e.g. identity mismatch, empty
-  agent list, peer version drift). Fire whenever the condition is present.
+  team roster, peer version drift). Fire whenever the condition is present.
 - **reinforcement** — periodic memory-aids, band-throttled by context
   percentage. Fire occasionally, not on every command.
 
@@ -280,44 +280,42 @@ Flags:
 --re-register      Re-register same agent
 ```
 
-### agent list
+### team roster
 
 ```bash
-thrum agent list                               # List all registered agents
-thrum agent list --role reviewer               # Filter by role
-thrum agent list --module auth                 # Filter by module
-thrum agent list --context                     # Show work context (branch, commits, intent)
-thrum agent list --json
+thrum team                                      # Local active roster
+thrum team --role reviewer                     # Filter by role
+thrum team --module auth                       # Filter by module
+thrum team --compact                            # One line per agent with intent and branch/commits
+thrum team --phase stale                       # Exact stale triage queue
+thrum team --include-inactive                  # Add stale and retired rows
+thrum team --offline                           # Add ordinary offline rows
+thrum team --all                               # Preserve the whole-fleet roster
+thrum team --json
 ```
 
-Flags:
-
-```text
---context         Show work context (branch, commits, intent)
---role string     Filter by role
---module string   Filter by module
-```
+`--include-inactive` includes stale and retired rows, including offline stale/retired rows, while omitting ordinary offline active rows. `--offline` adds ordinary offline active rows. Use both for their union. `--phase` remains exact. Stale rows show last-seen and measured worktree, identity-file, and tmux-session presence; unavailable measurements render as `unknown`. `Reap candidate` is display-only and appears only when worktree and tmux-session presence are both explicitly absent.
 
-### agent set-intent / session set-intent
+### session set-intent
 
 ```bash
-thrum agent set-intent "Fixing memory leak in connection pool"
+thrum session set-intent "Fixing memory leak in connection pool"
 thrum session set-intent "Refactoring login flow"
-thrum agent set-intent ""                      # Clear intent
+thrum session set-intent ""                      # Clear intent
 ```
 
-### agent set-task / session set-task
+### session set-task
 
 ```bash
-thrum agent set-task beads:thrum-xyz
+thrum session set-task beads:thrum-xyz
 thrum session set-task "JIRA-1234"
-thrum agent set-task ""                        # Clear task
+thrum session set-task ""                        # Clear task
 ```
 
-### agent heartbeat / session heartbeat
+### session heartbeat
 
 ```bash
-thrum agent heartbeat
+thrum session heartbeat
 thrum session heartbeat
 thrum session heartbeat --add-scope module:auth
 thrum session heartbeat --remove-ref pr:42
@@ -333,17 +331,17 @@ Flags:
 --remove-ref strings     Remove ref (repeatable, format: type:value)
 ```
 
-### agent start / session start
+### session start
 
 ```bash
-thrum agent start
+thrum session start
 thrum session start
 ```
 
-### agent end / session end
+### session end
 
 ```bash
-thrum agent end
+thrum session end
 thrum session end
 thrum session end --reason crash
 thrum session end --session-id <id>

package/assets/skills/thrum/resources/LISTENER_PATTERN.md

@@ -5,6 +5,10 @@
 > [TMUX_SESSIONS.md](TMUX_SESSIONS.md). Use the listener pattern below when tmux
 > is not available or not practical for your setup.
 >
+> In tmux-managed sessions, messages arrive in the pane. Check
+> `thrum inbox --unread` at natural breakpoints and never spawn a background
+> listener.
+>
 > **Deprecated for newly initialized agents:** the hook injection and
 > scheduled backstop that `thrum init` installs deliver message nudges without a
 > listener. The pattern below still works and is supported, but it holds tokens in

package/assets/skills/thrum/resources/TMUX_SESSIONS.md

@@ -2,7 +2,8 @@
 
 Tmux-managed sessions are the recommended way to run Thrum agents. The daemon
 detects the tmux pane and delivers message notifications directly — zero token
-cost, no background listener needed.
+cost, no background listener needed. Check `thrum inbox --unread` at natural
+breakpoints; never spawn a background listener.
 
 ## Why Tmux Sessions
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "opencode-thrum-pro",
-  "version": "0.1.9",
+  "version": "0.1.11",
   "description": "Thrum multi-agent coordination plugin for Open Code",
   "type": "module",
   "main": "./dist/index.js",