@monoes/monomindcli 2.16.1 → 2.16.3

package/.claude/skills/mastermind-new-agent/SKILL.md

@@ -1,13 +1,16 @@
 ---
 name: mastermind-new-agent
-description: Mastermind new-agent — wizard to hire/create a new agent within an org. Configures adapter type, model, role name, reports_to hierarchy, governance policy, skill assignments, and budget. Writes the new role to the org config file.
+description: Mastermind new-agent — wizard to hire/create a new agent within an org. Configures runtime, model, role name, reports_to hierarchy, responsibilities, skill assignments (found with `monomind org skills search`), and budget. Writes a valid Org Runtime v2 role to the org config file and validates it.
 type: domain-skill
 default_mode: confirm
+pick: low
 ---
 
 # Mastermind New Agent
 
-This skill is invoked by `mastermind:new-agent` or directly via `/mastermind:new-agent`.
+This skill is invoked by `mastermind:new-agent` or directly via `/mastermind-new-agent`.
+
+It writes one role in the Org Runtime v2 shape (`RoleSchema` in `packages/@monomind/cli/src/orgrt/types.ts`) — the same shape `mastermind-createorg` produces — and runs `monomind org validate` on the result. There is no `adapter` object, heartbeat or per-role system prompt in v2: the runtime reads `runtime`, `adapter_config.model`, `responsibilities`, `skills` and `skill_pool`.
 
 ---
 
@@ -15,35 +18,33 @@ This skill is invoked by `mastermind:new-agent` or directly via `/mastermind:new
 
 - `brain_context`: BRAIN CONTEXT block (injected by command, or loaded below if standalone)
 - `org_name`: org to add the agent to (required)
-- `action`: create | preview | list-roles | list-adapters
+- `action`: create | preview | list-roles | list-runtimes
 - `agent_id`: unique slug for the new agent (required for create; auto-generated if omitted)
 - `title`: display title for the agent (required for create)
-- `adapter_type`: claude-local | gemini-local | codex-local | cursor | opencode | hermes | http | acpx (default: claude-local)
-- `model`: model identifier (e.g. claude-sonnet-5, claude-opus-5, gemini-2.0-flash)
-- `max_tokens`: max tokens per run (default: 8192)
-- `reports_to`: parent agent id in the hierarchy (null = top-level)
-- `governance`: auto | board | strict (default: inherit from org)
-- `skills`: comma-separated skill names to assign (e.g. mastermind:tasks,mastermind:goals)
-- `budget_tokens`: per-run token budget cap (optional)
-- `heartbeat_enabled`: true | false (default: false)
-- `heartbeat_interval`: heartbeat interval in seconds (default: 900)
-- `system_prompt`: brief system prompt override (optional)
+- `responsibilities`: what the role does, `;`-separated (3–6 specific duties — this is the role's briefing and what `assignee: "auto"` matches tasks against)
+- `runtime`: claude | codex | antigravity | kimicode | opencode | vercel (default: claude — leave unset for the default Claude runtime)
+- `model`: model identifier (default: the latest model for the runtime, see below)
+- `max_tokens`: optional max output tokens per model call (omit to inherit)
+- `reports_to`: parent role id in the hierarchy (required — an org has exactly one root role, and that is the boss)
+- `skills`: comma-separated org-library skill names pinned into the prompt (1–3; found in Step 1.5)
+- `skill_pool`: comma-separated skill names or `tag:<tag>` the role may load mid-run with `org_skill_load`
+- `budget_tokens`: optional per-role token budget (positive integer; omitted when unset — the role then gets its even share of the org budget)
 - `caller`: command | master
 
 ---
 
-## Adapter Types
+## Runtimes and their latest models
+
+Every role gets an explicit `adapter_config.model` — never leave it to a runtime default, which moves when a release changes it. Unless the user asked for a specific model, use the latest one for the role's runtime (`resolveModel()` in `orgrt/session.ts` is the source of truth; if it disagrees with this table, it wins):
 
-| Type | Description | Default Model |
-|------|-------------|---------------|
-| `claude-local` | Claude via Anthropic API | claude-sonnet-5 |
-| `gemini-local` | Gemini via Google API | gemini-2.0-flash |
-| `codex-local` | OpenAI Codex | gpt-4o |
-| `cursor` | Cursor IDE agent | cursor-default |
-| `opencode` | OpenCode agent | opencode-default |
-| `hermes` | Hermes local model | hermes-3 |
-| `http` | Custom HTTP adapter | (custom) |
-| `acpx` | ACPX protocol | (custom) |
+| `runtime` | Latest model |
+|-----------|--------------|
+| `claude` (default — omit the field) | `claude-sonnet-5` |
+| `codex` | `gpt-5.6-terra` |
+| `antigravity` | `gemini-3.6-flash-high` |
+| `kimicode` | `kimi-code/k3` |
+| `opencode` | `glm-5.2` |
+| `vercel` | `gpt-5.5` |
 
 ---
 
@@ -62,26 +63,35 @@ orgFile=".monomind/orgs/${org_name}.json"
 
 ---
 
+## Step 1.5 — Find Skills (create / preview)
+
+Skills are org-library skills, not slash commands: names like `mastermind:tasks` are not valid here and fail `org validate`. When `skills` / `skill_pool` were not given, search the library with the role's title and responsibilities and choose from the hits:
+
+```bash
+npx monomind org skills search "${title} ${responsibilities}" --limit 8 --format json \
+  | jq -r '.skills[] | "\(.name)\t\(.description)"'
+npx monomind org skills show <name>   # read one before choosing it
+```
+
+Pin 1–3 that define the role in `skills`; put the ones it needs only for some tasks in `skill_pool`. Leave both empty when nothing fits cleanly — `responsibilities` alone is a complete role.
+
+---
+
 ## Step 2 — Execute Action
 
-### list-adapters
+### list-runtimes
 
 ```bash
-echo "AVAILABLE ADAPTERS"
+echo "ORG RUNTIME v2 RUNTIMES (latest model)"
 echo "────────────────────────────────────────────────────────"
-cat <<'ADAPTERS'
-  claude-local    Claude via Anthropic API   (recommended)
-    Models: claude-sonnet-5, claude-opus-5, claude-fable-5-1, claude-haiku-4-5-20251001
-  gemini-local    Gemini via Google AI
-    Models: gemini-2.0-flash, gemini-1.5-pro
-  codex-local     OpenAI Codex
-    Models: gpt-4o, gpt-4o-mini, o3-mini
-  cursor          Cursor IDE agent
-  opencode        OpenCode agent
-  hermes          Hermes local LLM (Ollama)
-  http            Custom HTTP adapter endpoint
-  acpx            ACPX protocol adapter
-ADAPTERS
+cat <<'RUNTIMES'
+  claude        claude-sonnet-5          (default; also claude-opus-5, claude-fable-5-1, claude-haiku-4-5-20251001)
+  codex         gpt-5.6-terra
+  antigravity   gemini-3.6-flash-high
+  kimicode      kimi-code/k3
+  opencode      glm-5.2
+  vercel        gpt-5.5                  (pair with provider.vendor)
+RUNTIMES
 ```
 
 ### list-roles
@@ -89,135 +99,108 @@ ADAPTERS
 ```bash
 echo "CURRENT ROLES IN ORG: $org_name"
 echo "────────────────────────────────────────────────────────"
-printf "%-22s %-20s %-14s %s\n" "ID" "TITLE" "ADAPTER" "REPORTS TO"
+printf "%-22s %-24s %-12s %-24s %s\n" "ID" "TITLE" "RUNTIME" "MODEL" "REPORTS TO"
 echo "────────────────────────────────────────────────────────"
 jq -r '(.roles // [])[] |
-  [.id, (.title // "-"), (.adapter.type // "claude-local"), (.reports_to // "(root)")] | @tsv' \
-  "$orgFile" | while IFS=$'\t' read -r id title adapter rt; do
-  printf "%-22s %-20s %-14s %s\n" "$id" "$title" "$adapter" "$rt"
+  [.id, (.title // "-"), (.runtime // "claude"), (.adapter_config.model // "(unset)"), (.reports_to // "(root)")] | @tsv' \
+  "$orgFile" | while IFS=$'\t' read -r id title rt model parent; do
+  printf "%-22s %-24s %-12s %-24s %s\n" "$id" "$title" "$rt" "$model" "$parent"
 done
 ```
 
-### preview
+### preview / create — shared setup
 
 ```bash
-[ -z "$title" ] && { echo "ERROR: --title required for preview."; exit 1; }
+[ -z "$title" ] && { echo "ERROR: --title required."; exit 1; }
+[ -z "$reports_to" ] && { echo "ERROR: --reports-to required (the org already has its root role)."; exit 1; }
 
-adapterType="${adapter_type:-claude-local}"
+runtimeVal="${runtime:-claude}"
 modelId="${model}"
-# Default model per adapter
 if [ -z "$modelId" ]; then
-  case "$adapterType" in
-    claude-local)  modelId="claude-sonnet-5" ;;
-    gemini-local)  modelId="gemini-2.0-flash" ;;
-    codex-local)   modelId="gpt-4o" ;;
-    cursor)        modelId="cursor-default" ;;
-    opencode)      modelId="opencode-default" ;;
-    hermes)        modelId="hermes-3" ;;
-    *)             modelId="(custom)" ;;
+  case "$runtimeVal" in
+    claude)       modelId="claude-sonnet-5" ;;
+    codex)        modelId="gpt-5.6-terra" ;;
+    antigravity)  modelId="gemini-3.6-flash-high" ;;
+    kimicode)     modelId="kimi-code/k3" ;;
+    opencode)     modelId="glm-5.2" ;;
+    vercel)       modelId="gpt-5.5" ;;
+    *) echo "ERROR: unknown runtime '$runtimeVal' (see --action list-runtimes)."; exit 1 ;;
   esac
 fi
 
-# Generate slug if not set
-agentIdPreview="${agent_id}"
-if [ -z "$agentIdPreview" ]; then
-  agentIdPreview=$(echo "$title" | tr '[:upper:]' '[:lower:]' | tr -cs 'a-z0-9' '-' | sed 's/^-//;s/-$//')
+[ -z "$agent_id" ] && agent_id=$(echo "$title" | tr '[:upper:]' '[:lower:]' | tr -cs 'a-z0-9' '-' | sed 's/^-//;s/-$//')
+
+if [ -n "$budget_tokens" ] && ! [[ "$budget_tokens" =~ ^[1-9][0-9]*$ ]]; then
+  echo "ERROR: --budget-tokens must be a positive integer (omit it for the even share)."; exit 1
 fi
+```
+
+### preview
 
-echo "PREVIEW — new agent to be added to org '$org_name'"
+```bash
+echo "PREVIEW — new role to be added to org '$org_name'"
 echo "────────────────────────────────────────────────────────"
-echo "  ID:          $agentIdPreview"
-echo "  Title:       $title"
-echo "  Adapter:     $adapterType / $modelId"
-echo "  Max tokens:  ${max_tokens:-8192}"
-echo "  Reports to:  ${reports_to:-(root/top-level)}"
-echo "  Governance:  ${governance:-inherit}"
-echo "  Skills:      ${skills:-(none)}"
-echo "  Budget:      ${budget_tokens:-(unlimited)}"
-echo "  Heartbeat:   ${heartbeat_enabled:-false} (${heartbeat_interval:-900}s)"
-[ -n "$system_prompt" ] && echo "  Prompt:      $system_prompt"
+echo "  ID:               $agent_id"
+echo "  Title:            $title"
+echo "  Runtime / model:  $runtimeVal / $modelId"
+echo "  Reports to:       $reports_to"
+echo "  Responsibilities: ${responsibilities:-(none — add some: they are the role's briefing)}"
+echo "  Skills:           ${skills:-(none)}"
+echo "  Skill pool:       ${skill_pool:-(none)}"
+echo "  Budget:           ${budget_tokens:-(even share of the org budget)}"
 echo ""
-echo "Run with --action create to add this agent to the org."
+echo "Run with --action create to add this role to the org."
 ```
 
 ### create
 
 ```bash
-[ -z "$title" ] && { echo "ERROR: --title required."; exit 1; }
-
-adapterType="${adapter_type:-claude-local}"
-modelId="${model}"
-if [ -z "$modelId" ]; then
-  case "$adapterType" in
-    claude-local)  modelId="claude-sonnet-5" ;;
-    gemini-local)  modelId="gemini-2.0-flash" ;;
-    codex-local)   modelId="gpt-4o" ;;
-    cursor)        modelId="cursor-default" ;;
-    opencode)      modelId="opencode-default" ;;
-    hermes)        modelId="hermes-3" ;;
-    *)             modelId="(custom)" ;;
-  esac
-fi
-
-# Generate id from title if not provided
-if [ -z "$agent_id" ]; then
-  agent_id=$(echo "$title" | tr '[:upper:]' '[:lower:]' | tr -cs 'a-z0-9' '-' | sed 's/^-//;s/-$//')
-fi
-
-# Check for duplicate id
 duplicate=$(jq -r --arg id "$agent_id" '[(.roles // [])[] | select(.id == $id)] | length' "$orgFile")
-[ "$duplicate" -gt 0 ] && { echo "ERROR: Agent id '$agent_id' already exists in org '$org_name'. Use --agent-id to specify a unique id."; exit 1; }
+[ "$duplicate" -gt 0 ] && { echo "ERROR: Role id '$agent_id' already exists in org '$org_name'. Use --agent-id to specify a unique id."; exit 1; }
 
-# Validate reports_to if set
-if [ -n "$reports_to" ]; then
-  parentExists=$(jq -r --arg pid "$reports_to" '[(.roles // [])[] | select(.id == $pid)] | length' "$orgFile")
-  [ "$parentExists" -eq 0 ] && { echo "ERROR: Parent agent '$reports_to' not found in org '${org_name}'. Check the agent ID with /mastermind:org-chart --org ${org_name}."; exit 1; }
-fi
+parentExists=$(jq -r --arg pid "$reports_to" '[(.roles // [])[] | select(.id == $pid)] | length' "$orgFile")
+[ "$parentExists" -eq 0 ] && { echo "ERROR: Parent role '$reports_to' not found in org '${org_name}'. Check ids with --action list-roles."; exit 1; }
 
-# Build skills array
-skillsArray="[]"
-if [ -n "$skills" ]; then
-  skillsArray=$(echo "$skills" | tr ',' '\n' | jq -Rsc 'split("\n") | map(select(. != ""))')
-fi
-
-ts=$(date -u +%Y-%m-%dT%H:%M:%SZ)
+list() { [ -n "$1" ] && echo "$1" | tr "$2" '\n' | sed 's/^ *//;s/ *$//' | jq -Rsc 'split("\n") | map(select(. != ""))' || echo '[]'; }
 
 tmp="${orgFile}.tmp"
 jq --arg id "$agent_id" \
    --arg title "$title" \
-   --arg adapter "$adapterType" \
+   --arg rt "$reports_to" \
+   --arg runtime "$runtimeVal" \
    --arg model "$modelId" \
-   --argjson maxTok "${max_tokens:-8192}" \
-   --arg rt "${reports_to:-}" \
-   --arg gov "${governance:-}" \
-   --argjson skills "$skillsArray" \
-   --argjson budget "${budget_tokens:-null}" \
-   --argjson hbEnabled "${heartbeat_enabled:-false}" \
-   --argjson hbInterval "${heartbeat_interval:-900}" \
-   --arg prompt "${system_prompt:-}" \
-   --arg ts "$ts" \
-  '.roles += [{
-    "id": $id,
-    "title": $title,
-    "adapter": {"type": $adapter, "model": $model, "max_tokens": $maxTok},
-    "reports_to": (if $rt != "" then $rt else null end),
-    "governance": (if $gov != "" then $gov else null end),
-    "skills": $skills,
-    "budget_tokens": $budget,
-    "heartbeat": {"enabled": $hbEnabled, "interval": $hbInterval},
-    "system_prompt": (if $prompt != "" then $prompt else null end),
-    "created_at": $ts
-  }]' \
-  "$orgFile" > "$tmp" && mv "$tmp" "$orgFile"
-
-echo "Agent created: $agent_id"
+   --arg maxTok "${max_tokens:-}" \
+   --arg budget "${budget_tokens:-}" \
+   --argjson resp "$(list "$responsibilities" ';')" \
+   --argjson skills "$(list "$skills" ',')" \
+   --argjson pool "$(list "$skill_pool" ',')" \
+  '.roles += [
+    {id: $id, title: $title, type: "specialist", reports_to: $rt, responsibilities: $resp,
+     adapter_config: ({model: $model} + (if $maxTok != "" then {max_tokens: ($maxTok | tonumber)} else {} end))}
+    + (if $runtime != "claude" then {runtime: $runtime} else {} end)
+    + (if ($skills | length) > 0 then {skills: $skills} else {} end)
+    + (if ($pool | length) > 0 then {skill_pool: $pool} else {} end)
+    + (if $budget != "" then {budget_tokens: ($budget | tonumber)} else {} end)
+  ]' \
+  "$orgFile" > "$tmp" || { rm -f "$tmp"; echo "ERROR: could not build the role."; exit 1; }
+
+# Validate BEFORE replacing the config: the same parse org run/serve use,
+# plus unknown-skill checks. A refused role leaves the org file untouched.
+cp "$orgFile" "${orgFile}.bak" && mv "$tmp" "$orgFile"
+if ! npx -y monomind@latest org validate "$org_name"; then
+  mv "${orgFile}.bak" "$orgFile"
+  echo "ERROR: the new role did not validate (see above) — org config restored."; exit 1
+fi
+rm -f "${orgFile}.bak"
+
+echo "Role created: $agent_id"
 echo "  Title:    $title"
-echo "  Adapter:  $adapterType / $modelId"
-echo "  Reports:  ${reports_to:-(root)}"
-echo "  Skills:   ${skills:-(none)}"
+echo "  Runtime:  $runtimeVal / $modelId"
+echo "  Reports:  $reports_to"
+echo "  Skills:   ${skills:-(none)}  pool: ${skill_pool:-(none)}"
 echo ""
-echo "Org '${org_name}' now has $(jq '.roles | length' "$orgFile") agent(s)."
-echo "View: /mastermind:agent-detail --org $org_name --agent-id $agent_id"
+echo "Org '${org_name}' now has $(jq '.roles | length' "$orgFile") role(s)."
+echo "View: /mastermind-agent-detail --org $org_name --agent-id $agent_id"
 ```
 
 ---
@@ -230,7 +213,7 @@ status: complete
 action: <action>
 org: <org_name>
 agent_id: <agent_id>
-adapter_type: <type>
+runtime: <runtime>
 model: <model>
 ```
 

package/.claude/skills/mastermind-org-settings/SKILL.md

@@ -3,16 +3,17 @@ name: mastermind-org-settings
 description: Mastermind org-settings — edit Org Runtime v2 configuration (name, goal, schedule, budget_tokens, memory_namespace, max_turns_per_message), export org as portable JSON, and import an org from a previously exported file.
 type: domain-skill
 default_mode: confirm
+pick: low
 ---
 
 # Mastermind Org Settings
 
-This skill is invoked by `mastermind:org-settings` or directly via `/mastermind:org-settings`.
+This skill is invoked by `mastermind:org-settings` or directly via `/mastermind-org-settings`.
 
 It edits `.monomind/orgs/<org_name>.json` in place, and only ever touches fields that `OrgDefSchema` (`packages/@monomind/cli/src/orgrt/types.ts`) actually defines — every field listed below is read by the daemon (`org.ts`/`daemon.ts`/`session.ts`) at `monomind org run`/`serve` time. Its primary edit flow targets v2 fields only: `name`, `goal`, `schedule`, and `run_config` (`budget_tokens`, `memory_namespace`, `max_turns_per_message`). Role edits are not yet supported by this skill's `edit` action — `roles` is shown read-only via `show`.
 
 <!-- LEGACY-ORG-V1: remove this note when v1 orgs are gone -->
-There is no `topology`, `governance`, `alert_threshold`, or `ceo_adapter` in Org Runtime v2 — those were v1 board/prompt-orchestration fields with no runtime effect and have been removed from this skill's edit surface. Use `/mastermind:org-settings` only for v2-shaped orgs; v1 orgs must go through `monomind org migrate` first.
+There is no `topology`, `governance`, `alert_threshold`, or `ceo_adapter` in Org Runtime v2 — those were v1 board/prompt-orchestration fields with no runtime effect and have been removed from this skill's edit surface. Use `/mastermind-org-settings` only for v2-shaped orgs; v1 orgs must go through `monomind org migrate` first.
 
 ---
 
@@ -64,7 +65,7 @@ jq -r '
   "  Roles:             \(.roles | length) agents"
 ' "$orgFile"
 echo ""
-echo "  Run /mastermind:org-settings --org ${org_name} --action edit --field <field> --value <value>"
+echo "  Run /mastermind-org-settings --org ${org_name} --action edit --field <field> --value <value>"
 echo "  Fields: name | goal | schedule | budget_tokens | memory_namespace | max_turns_per_message"
 ```
 
@@ -190,7 +191,7 @@ jq -n \
 
 echo "Exported: $outPath"
 echo "$(wc -c < "$outPath") bytes — share this file to recreate the org on another machine."
-echo "Import with: /mastermind:org-settings --action import --import-path $outPath"
+echo "Import with: /mastermind-org-settings --action import --import-path $outPath"
 ```
 
 ### import
@@ -241,7 +242,7 @@ echo "Agents: $(jq '.config.roles | length' "$import_path")"
 npx -y monomind@latest org validate "$targetOrg" \
   || { echo "ERROR: imported org failed validation — fix .monomind/orgs/${targetOrg}.json before running it."; exit 1; }
 
-echo "Run /mastermind:env --org ${targetOrg} --action validate to check provider keys."
+echo "Run /mastermind-env --org ${targetOrg} --action validate to check provider keys."
 ```
 
 ---

package/.claude/skills/mastermind-orgs/SKILL.md

@@ -3,6 +3,7 @@ name: mastermind-orgs
 description: Mastermind orgs — list all saved orgs with their status, schedule, and last/next run times. Shows stopped/active/paused state for scheduled orgs.
 type: domain-skill
 default_mode: auto
+pick: low
 ---
 
 # Mastermind Orgs

package/.claude/skills/mastermind-orgstatus/SKILL.md

@@ -3,6 +3,7 @@ name: mastermind-orgstatus
 description: Mastermind orgstatus — show detailed status for a single org including lifecycle state, schedule, last/next run, recent activity, and roles. For scheduled orgs shows loop health and time until next iteration.
 type: domain-skill
 default_mode: auto
+pick: low
 ---
 
 # Mastermind Org Status
@@ -223,7 +224,7 @@ if [ "$is_v2" = "yes" ]; then
   echo "  Logs:         monomind org logs $name --follow"
   echo "  Report:       monomind org report $name   (add --all for run history)"
   echo "  Validate:     monomind org validate $name"
-  echo "  Settings:     /mastermind:org-settings --org $name"
+  echo "  Settings:     /mastermind-org-settings --org $name"
 # LEGACY-ORG-V1: remove the next two branches when v1 orgs are gone
 elif [ "$has_schedule" = "yes" ]; then
   case "$status" in

package/.claude/skills/mastermind-plan-to-tasks/SKILL.md

@@ -7,7 +7,7 @@ default_mode: confirm
 
 # Mastermind Plan to Tasks
 
-This skill is invoked by `mastermind:plan-to-tasks` or directly via `/mastermind:plan-to-tasks`.
+This skill is invoked by `mastermind:plan-to-tasks` or directly via `/mastermind-plan-to-tasks`.
 
 ---
 

package/.claude/skills/mastermind-plugin-manager/SKILL.md

@@ -3,11 +3,12 @@ name: mastermind-plugin-manager
 description: Mastermind plugin-manager — install plugins from npm, uninstall with two-step confirmation, enable/disable installed plugins. Extends mastermind:plugins (listing) and mastermind:plugin-settings (per-plugin config) with the install/uninstall lifecycle. Mirrors PluginManager.tsx.
 type: domain-skill
 default_mode: confirm
+pick: low
 ---
 
 # Mastermind Plugin Manager
 
-This skill is invoked by `mastermind:plugin-manager` or directly via `/mastermind:plugin-manager`.
+This skill is invoked by `mastermind:plugin-manager` or directly via `/mastermind-plugin-manager`.
 
 ---
 
@@ -86,7 +87,7 @@ fi
 echo ""
 echo "  Install:    --action install --package-name <npm-package>"
 echo "  Uninstall:  --action uninstall --plugin-id <id>"
-echo "  Settings:   /mastermind:plugin-settings --plugin-id <id>"
+echo "  Settings:   /mastermind-plugin-settings --plugin-id <id>"
 ```
 
 ### install
@@ -125,7 +126,7 @@ if npm install "$package_name" 2>&1 | tail -5; then
 
   echo ""
   echo "Plugin installed: $pluginId @ $pkgVer"
-  echo "  Configure: /mastermind:plugin-settings --plugin-id $pluginId"
+  echo "  Configure: /mastermind-plugin-settings --plugin-id $pluginId"
 else
   echo "  ERROR: npm install failed. Check the package name and network connectivity."
   exit 1

package/.claude/skills/mastermind-plugin-settings/SKILL.md

@@ -3,11 +3,12 @@ name: mastermind-plugin-settings
 description: Mastermind plugin-settings — inspect and configure a single installed plugin. View configuration fields, manage file system access grants (read/write paths), check runtime health/status, and update plugin settings. Mirrors Paperclip's PluginSettings page with configuration and status tabs.
 type: domain-skill
 default_mode: confirm
+pick: low
 ---
 
 # Mastermind Plugin Settings
 
-This skill is invoked by `mastermind:plugin-settings` or directly via `/mastermind:plugin-settings`.
+This skill is invoked by `mastermind:plugin-settings` or directly via `/mastermind-plugin-settings`.
 
 ---
 
@@ -63,10 +64,10 @@ If `caller` is not "command", load brain context following mastermind-protocol/S
 
 ```bash
 registryFile=".monomind/plugins/registry.json"
-[ ! -f "$registryFile" ] && { echo "ERROR: No plugin registry found. Install plugins via /mastermind:plugins."; exit 1; }
+[ ! -f "$registryFile" ] && { echo "ERROR: No plugin registry found. Install plugins via /mastermind-plugins."; exit 1; }
 
 pluginDef=$(jq -r --arg id "$plugin_id" '(.plugins // [])[] | select(.id == $id)' "$registryFile")
-[ -z "$pluginDef" ] && { echo "ERROR: Plugin '$plugin_id' not found. List plugins via /mastermind:plugins --action list."; exit 1; }
+[ -z "$pluginDef" ] && { echo "ERROR: Plugin '$plugin_id' not found. List plugins via /mastermind-plugins --action list."; exit 1; }
 
 # Load org-level overrides if org_name specified
 orgPluginsFile=""
@@ -172,7 +173,7 @@ if [ -n "$lastErr" ]; then
 fi
 
 # Suggest reload
-[ "$healthStatus" = "error" ] && echo "" && echo "  Run: /mastermind:plugins --action reload --plugin-id $plugin_id"
+[ "$healthStatus" = "error" ] && echo "" && echo "  Run: /mastermind-plugins --action reload --plugin-id $plugin_id"
 ```
 
 ### grants

package/.claude/skills/mastermind-plugins/SKILL.md

@@ -3,11 +3,12 @@ name: mastermind-plugins
 description: Mastermind plugins — install, enable, disable, uninstall, and inspect plugins for an org. Plugins extend agent capabilities with workers, events, and custom tools. Supports npm packages and local paths.
 type: domain-skill
 default_mode: confirm
+pick: low
 ---
 
 # Mastermind Plugins
 
-This skill is invoked by `mastermind:plugins` or directly via `/mastermind:plugins`.
+This skill is invoked by `mastermind:plugins` or directly via `/mastermind-plugins`.
 
 ---
 
@@ -94,7 +95,7 @@ cat <<'EXAMPLES'
   @monomind/plugin-webhook     — Generic inbound/outbound webhooks
   @monomind/plugin-memory-ext  — Extended memory backend
 
-Install: /mastermind:plugins --action install --package-name @monomind/plugin-<name>
+Install: /mastermind-plugins --action install --package-name @monomind/plugin-<name>
 EXAMPLES
 ```
 

package/.claude/skills/mastermind-profile/SKILL.md

@@ -3,11 +3,12 @@ name: mastermind-profile
 description: Mastermind profile — view and edit the current operator profile (display name, preferences) and inspect any user's activity statistics, completion rate, and token usage. Merges ProfileSettings.tsx and UserProfile.tsx.
 type: domain-skill
 default_mode: confirm
+pick: low
 ---
 
 # Mastermind Profile
 
-This skill is invoked by `mastermind:profile` or directly via `/mastermind:profile`.
+This skill is invoked by `mastermind:profile` or directly via `/mastermind-profile`.
 
 ---
 

package/.claude/skills/mastermind-protocol/SKILL.md

@@ -14,11 +14,11 @@ This file is a reference loaded by mastermind domain skills and master. It is NE
 
 **Every Task/Agent spawn in mastermind and monomind MUST include the AGENT DELEGATION CAPABILITY block from `mastermind-delegation/SKILL.md`.**
 
-This makes delegation recursive: every spawned agent can itself spawn sub-agents, which can spawn their own sub-agents. The capability block tells each agent which agent categories exist and how to delegate.
+This makes delegation recursive: every spawned agent can itself spawn sub-agents, which can spawn their own sub-agents. The capability block tells each agent how to pick a real agent (the `[PICK]` line, `mcp__monomind__pick`, or local `monomind pick`) and how to delegate.
 
 **Placement:** In the Task/Agent `description` or `prompt` field, insert the full `== AGENT DELEGATION CAPABILITY == ... =================================` block immediately after `BRAIN CONTEXT:`.
 
-**Why this matters:** Agents read their prompts cold. Without the delegation block, a spawned Security Engineer won't know it can delegate codebase exploration to a Code Reviewer, or that it can spawn a backend-dev to fix the issues it finds. With the block, every agent in the chain can self-organize and pull in the right specialist.
+**Why this matters:** Agents read their prompts cold. Without the delegation block, a spawned Security Engineer won't know it can delegate codebase exploration to a reviewer, or that it can spawn a Backend Architect to fix the issues it finds. With the block, every agent in the chain can self-organize and pull in the right specialist.
 
 **Reference:** Full block text is in `mastermind-delegation/SKILL.md`. Copy it verbatim — do not paraphrase.
 
@@ -289,7 +289,7 @@ SUCCESS CRITERIA:
 - [ ] [Concrete checkable item 1]
 - [ ] [Concrete checkable item 2]
 
-AGENT: [agent slug, e.g. backend-dev | frontend-dev]
+AGENT: [agent name from the [PICK] line or mcp__monomind__pick, e.g. Backend Architect | Frontend Developer]
 SWARM: [topology agent-count consensus, e.g. \"hierarchical 4 raft\"]
 
 DEPENDENCIES: [task IDs, or \"none\"]

package/.claude/skills/mastermind-release/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: mastermind-release
-description: Mastermind release domain — versioning, changelog, deployment coordination. Spawns a Release Manager who coordinates testing and devops agents for a safe, traceable release pipeline.
+description: Mastermind release domain — versioning, changelog, deployment coordination. Spawns a Release Manager who coordinates testing and deployment agents for a safe, traceable release pipeline.
 type: domain-skill
 default_mode: auto
 ---
@@ -105,18 +105,23 @@ SCOPE: [packages, environments, services in scope]
 CONSTRAINTS: [breaking change rules, downtime windows, rollback triggers, compliance gates]
 SUCCESS CRITERIA:
 - [ ] [checkable item — e.g. \"all tests green before deploy\"]
-AGENT: [release-manager | tester | DevOps Automator | cicd-engineer]
+AGENT: [specialist picked in STEP 3]
 SWARM: hierarchical 5 raft
 DEPENDENCIES: [prior stage task ID — release pipeline is sequential]
 OUTPUT FORMAT: unified output schema"
 ```
 
 STEP 3 — EXECUTE
-Spawn Task agents in release order (hierarchical raft — coordinator maintains authoritative release state):
-- Pre-release prep: subagent_type "release-manager"
-- Testing gate: subagent_type "tester"
-- Infrastructure and deploy: subagent_type "DevOps Automator"
-- CI/CD pipeline: subagent_type "cicd-engineer"
+Spawn Task agents in release order (hierarchical raft — coordinator maintains authoritative release state).
+Pick one specialist per release stage from the shared agent index (the Jev decision model when configured, keyword ranking otherwise); it only returns agents that exist. In order: the prompt's `[PICK]` line when it fits this release stage; else `mcp__monomind__pick({ task: "<release stage>: <scope>", kind: "agents", top: 1 })` → `agents.ranked[0].name`; else the local CLI through the version-checked `mmpick` helper from `mastermind-agent-select/SKILL.md` (never npx):
+```bash
+mmpick -t "<release stage>: <scope>" --top 1 | jq -r '.agents.ranked[0].name // empty'
+```
+Use the returned name as that release stage's subagent_type. If nothing is returned, use the default below.
+- Pre-release prep: "release-manager"
+- Testing gate: "tester"
+- Infrastructure and deploy: "DevOps Automator"
+- CI/CD pipeline: "workflow-automation"
 
 Also run /mastermind:do --board <board_id> to track execution.
 
@@ -161,7 +166,7 @@ For simple tasks (single agent, single step):
 
 | Task Type | Agent | Swarm |
 |---|---|---|
-| Full end-to-end release | coordinator + tester + devops | hierarchical 5 raft specialized |
+| Full end-to-end release | coordinator + tester + DevOps Automator | hierarchical 5 raft specialized |
 | Multi-package release | coordinator + release-manager | hierarchical 5 raft specialized |
 | Deploy only | DevOps Automator | hierarchical 3 raft specialized |
 | Test gate | tester | star 4 raft parallel |