CRITICAL - Before starting, MUST first read
../te-shared/SKILL.md: authentication, host configuration, global parameters, and write-operation confirmation rules are all in there. CRITICAL - Before executing any+<tool_name>command, MUST first read the corresponding reference document./references/<tool-name>.mdFor example: before executing+get_resource_url, you must first read./references/get-resource-url.md
When to Use
When the requirement belongs to the unified execution constraints across te-analysis / te-meta / te-audience, you must use te-common first:
- Post-write actions, such as resource link completion
- Unified output conventions and fallback policies
- Future cross-module common constraints
Execution Sequence (AI Agent Must Follow)
[Start] → [Read command reference document] → [Execute write operation] → [Succeeded?]
↓ Yes
[Pause immediately, review the "write-operation post-link completion" section of this document]
↓
[Call +get_resource_url]
↓
[Output complete result + link]
↓
[End]
Key point: after a write operation succeeds → immediately go back to te-common, do not skip it.
Current Mandatory Constraints (MUST)
A. PROJECT_ID_GATE (preflight gate, no guessing)
- For any command about to be executed, you must first confirm whether that command requires
project_id(per the correspondingreferences/<tool-name>.mddocument). - Regardless of whether the user provided a project name or project_id, you must first call
+list_projectsto retrieve the project list for verification. - Verification rules:
- If the user provided a number, you must find the matching project_id in the list
- If the user provided a name, you must find the matching project name in the list and obtain the corresponding project_id
- If no match is found, you must tell the user and show the available project list so they can choose again
- Before verification passes and the correct
project_idis confirmed, do not call the target command.
Recommended interaction flow:
Scenario 1: The user did not provide project information
This step requires project_id.
I will first list the available projects for you to choose from:
[Call +list_projects to display the project list]
Please choose one from the projects above (for example, reply with the project ID or the project name).
I will not execute the target command before you confirm.
Scenario 2: The user provided a project name or ID
The project you mentioned is "xxx".
I will first verify whether this project exists:
[Call +list_projects to get the project list]
Verification result:
- ✅ Project "xxx" exists, project_id = 123
- Or ❌ Project "xxx" was not found, available projects are listed below: [display list]
Please confirm whether to use this project (project_id = 123) for the next operation.
B. Write-operation post-link completion (mandatory closed loop)
Trigger condition: When any of the following commands succeeds, MUST enter the link completion flow:
| resource_type | Create command | Update command |
|---|---|---|
dashboard |
+create_dashboard |
+update_dashboard |
report |
+create_report |
- |
metric |
+create_metric |
+update_metric |
alert |
+create_alert |
+update_alert |
tag |
+create_tag |
+update_tag |
cluster |
+create_cluster, +create_result_cluster |
+update_cluster |
virtual_event |
+create_virtual_event |
- |
event_virtual_prop |
+create_virtual_property (--table_type event) |
- |
user_virtual_prop |
+create_virtual_property (--table_type user) |
- |
super_event |
+batch_create_metadata (creating events) |
+batch_edit_metadata --type event |
super_prop_event |
+batch_create_metadata (creating event_properties) |
+batch_edit_metadata --type event_property |
super_prop_user |
+batch_create_metadata (creating user_properties) |
+batch_edit_metadata --type user_property |
Closed loop execution (must be completed):
- ✅ Write operation succeeded → immediately check whether the return value contains
resource_id - ✅ Has
resource_id→ call+get_resource_url(using the verifiedproject_id) - ✅ Link generated successfully → output the complete response including the clickable link
- ⚠️ Link generation failed → output the main result + explicitly state "link completion failed: [reason]"
Multi-resource result handling (for batch metadata only):
+batch_create_metadata/+batch_edit_metadatamay return multiple resources at once (events/properties).- For each resource from which
resource_idcan be extracted, you must call+get_resource_urlseparately. - The final output must provide links for each resource one by one; if some fail, mark the failed items and the reasons, while the successful items are returned normally.
Prohibited behaviors:
- ❌ End the response immediately after a write operation succeeds (you must complete link completion or explain the failure)
- ❌ Treat a write operation as failed because link completion failed (the write result is independently valid)
Closed-loop check template (must explicitly confirm each step):
[Write operation result]
→ Check resource_id: [yes/no]
→ Call get_resource_url: [called / skipped (no ID)]
→ Link completion status: [successful link / failure reason / skipped]
→ Closed loop complete ✅
If the ID cannot be extracted or link generation fails, you must explain the failure reason, but you cannot roll back or judge the main write operation as failed.
Execution Template
- Execute the main write-operation command.
- Extract
project_id,resource_type,resource_idfrom the input or return value. - Call
ae-cli analysis_common +get_resource_url --project_id <id> --resource_type <type> --resource_id <id>. - Output the main result +
markdown_link(clickable) /raw_url.
Unified Command Entry
+list_projects(documentation)+get_resource_url(documentation)