From flowstate-skills
Use when responding to an inbox item, handing work from one agent to another, escalating from a mention or reply, or acknowledging completed agent-to-agent collaboration - provides reply, escalation, and handoff patterns that keep inbox state coherent
How this skill is triggered — by the user, by Claude, or both
Slash command
/flowstate-skills:flowstate-agent-handoff-responseThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Status:** Active
Status: Active
Purpose: Turn inbox items into clear replies, escalations, task handoffs, or closure acknowledgements
Scope: Mentions, replies, active tasks, conversations, approvals, and description mentions from flowstate inbox
Trigger: An agent is triaging inbox and needs to involve, redirect, or acknowledge another agent
Input: Inbox row, full entity/discussion/conversation/approval context, target agent if escalating
Output: One coherent response path with IDs preserved and next owner clear
Inbox triage should leave the system more legible than it found it. Every response answers:
What did I receive? -> What did I decide? -> Who owns the next action? -> Where is it recorded?
Use this skill after flowstate-agent-inbox surfaces an item.
Never act from the inbox snippet alone.
| Inbox section | Pull before responding |
|---|---|
| Active tasks | collection-get tasks <task_id> |
| Mentions | collection-get discussions <discussion_id> plus target entity |
| Replies | Thread root parentDiscussionId, new discussionId, and target entity |
| Conversations | flowstate agent conversation show <conv_id> |
| Approvals | collection-get approvals <appr_id> plus related entity/document |
| Description mentions | `collection-get <tasks |
Preserve IDs exactly as printed by the inbox.
| Inbox item means | Response |
|---|---|
| You can answer directly | Reply in the same discussion or conversation |
| You own the work | Start/continue the task and update status |
| Another agent owns the work | Assign or reassign a task using flowstate-agent-task-assignment |
| Another agent should weigh in | Mention them on the same entity using flowstate-agent-mentioning |
| A bounded decision is needed | Start an agent conversation |
| A formal gate is needed | Create/update an approval |
| The item is informational | Post a short acknowledgement only if it helps future readers |
Avoid scattering the same decision across multiple places. If you escalate, say where the new action lives.
For mentions and replies, post to the same entity.
New mention response:
discussion-reply {
entityType: "<task|project|milestone>",
entityId: "<entity_id>",
parentId: "<discussion_id_or_parentDiscussionId>",
content: "<decision, answer, or escalation note>",
userName: "<my userName>",
userId: "<my teamMemberId>",
orgId: "<orgId>",
workspaceId: "<workspaceId>",
threadDepth: 1,
isEdited: false,
isDeleted: false
}
For inbox Replies, use parentDiscussionId as parentId, not the new reply's discussionId.
Escalation should create exactly one next-action owner:
| Escalation need | Action |
|---|---|
| Needs execution | Create or reassign a task to target agent |
| Needs comment/review | Mention target agent in the same entity discussion |
| Needs short decision | Start agent-chat conversation |
| Needs formal sign-off | Create approval with target approverId |
Always include:
Then reply to the original thread with the new action ID:
Escalated to @<target-slug> via task_<id> because <reason>. I will monitor the original thread for follow-up.
When handing off a task you own:
metadata.assignedAgent.Handoff does not mean abandoning context. Leave enough evidence for the receiver to continue cleanly.
When a collaboration item is resolved:
| Channel | Closure action |
|---|---|
| Discussion | Reply with result and any linked task/PR/approval ID |
| Task | Update status/completed fields according to flowstate-task-execution |
| Conversation | Close with consensus or decision-made and summarize durable outcome elsewhere if needed |
| Approval | Update approval status and comments |
| Description mention | Reply on the entity with outcome; do not remove description text unless it is stale or wrong |
| Situation | Action |
|---|---|
| Inbox snippet lacks enough context | Pull full records before acting |
| Reply target is unclear | Use the entity discussion root; do not create a separate orphan conversation |
| Escalation creates duplicate work | Link to the existing task/conversation instead of making another |
| Target agent is overloaded | Mention manager or lower priority unless the work is urgent/blocking |
| Approval is pending and you need input | Request revision or comment with blocker; do not leave workflow silently blocked |
| Item | Convention |
|---|---|
| Thread replies | Use root parentDiscussionId |
| Escalation note | Include new action ID in original thread |
| Handoff owner | One next owner at a time |
| Closure | Record outcome where future inbox readers will find it |
| No silent drops | If choosing no action, be able to state why |
flowstate-agent-inboxflowstate-agent-collaboration-routingflowstate-agent-task-assignmentflowstate-agent-mentioningflowstate-agent-conversation-protocolflowstate-approval-workflowCreated: 2026-05-26
npx claudepluginhub epic-digital-im/epic-flowstate-skills --plugin flowstate-skillsGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.