Writing
PromptBeginner5 minmarkdown
Agentic Bill Payments MCP Server
generic skill
1
Tools prefixed with `A2A_` are autonomous remote agents.
Sign in to like and favorite skills
generic skill
I want you to act as an English translator, spelling corrector and improver. I will speak to you in any language and you will detect the language, translate it and answer in the corrected and improved...
I want you to act as a text based excel. you'll only reply me the text-based 10 rows excel sheet with row numbers and cell letters as columns (A to L). First column header should be empty to reference...
Tools prefixed with
A2A_ are autonomous remote agents.
You interact with them via structured JSON messages. Each tool description specifies the agent's capabilities.
Every response has a
kind property that determines how to proceed:
kind: "message" → Complete Responsecontents arraykind: "task" → Ongoing WorkflowtaskId, contextId, and statusstatus.state to determine next actionkind: Always check first ("message" or "task")taskId: Task identifier assigned by remote agent (in task responses only)contextId: Session identifier assigned by remote agent on first responsestatus.state: Current task state (in task responses only)contents: Array of content objects (in message responses)artifacts: Final outputs (only in completed tasks)referenceTaskIds: Optional array linking to prior task dependencies| State | Action |
|---|---|
| Reuse + , provide requested input in the subsequent message |
| Extract , use results. Optionally reference in future requests |
/ | Stop processing. The agent didn't response in a timely manner |
| Check error details. Start new request (omit ) if retrying |
/ | Stop processing. Start new request if needed |
input-required, querying status)taskIds for concurrent requests to same agent (they may share contextId)Request 1: {contents: [...]} Response 1: {kind: "task", taskId: "T1", contextId: "C1", status: {state: "input-required"}} Request 2: {taskId: "T1", contextId: "C1", contents: [...]} Response 2: {kind: "task", taskId: "T1", contextId: "C1", status: {state: "completed"}, artifacts: [...]} Request 3: {contextId: "C1", referenceTaskIds: ["T1"], contents: [...]} Response 3: {kind: "message", contextId: "C1", contents: [...]}
Check kind
"message": Extract contents and proceed with your work"task": Continue to step 2Check status.state
input-required: Build continuation message with same taskId + contextIdcompleted: Extract artifacts, taskId can be used for potential future referencefailed/cancelled/rejected: Handle error, start new request if neededsubmitted/working: Stop processing❌ Not checking
kind before processing response
❌ Generating your own taskId or contextId values
❌ Including taskId when starting a new request
❌ Omitting taskId when responding to input-required status
❌ Reusing taskId from terminal states (completed/failed/cancelled/rejected)
❌ Mixing up taskIds when managing multiple concurrent tasks to same agent
❌ Including contextId in your very first message to an agent
❌ Treating message responses as requiring follow-up
taskId, include contextId (except first message)taskId and contextIdreferenceTaskIds with new request