效率与协作 · SKILL.md

automation-pilot-executions-api

automation-pilot-executions-api

【效率与协作】Monitor and manage SAP Automation Pilot executions via API. Use when triggering commands, checking execution status, retrieving logs, aborting or pausing executions, or troubleshooting failed runs.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

Wwilliamcorrea23 / sap-router-skillMIT

技能说明

GitHub

name: automation-pilot-executions-api description: Monitor and manage SAP Automation Pilot executions via API. Use when triggering commands, checking execution status, retrieving logs, aborting or pausing executions, or troubleshooting failed runs.

SAP Automation Pilot Executions API Management

This skill manages command executions in SAP Automation Pilot - triggering, monitoring, controlling, and troubleshooting executions.

Prerequisites

Set the following required environment variables:

export AUTOPI_HOSTNAME="emea.autopilot.cloud.sap"
export AUTOPI_USERNAME="your-username"
export AUTOPI_PASSWORD="your-password"
export AUTOPI_DEFAULT_CATALOG="mycommands-<<<TENANT_ID>>>"

Ensure curl is available in your environment.

Execution Lifecycle

Statuses

Status	Description	Active or Terminal
`RUNNING`	Execution is currently running. During delays (initial delay, retry backoff, repeat interval) the execution stays RUNNING while individual executors show their delay reason.	Active
`PAUSED`	Manually paused — must be resumed to continue.	Active
`INPUT_REQUIRED`	Waiting for a user choice (Confirm action required).	Active
`FAILED`	Failed due to an error or failed validation. Can be retried or reset.	Active
`FINISHED`	Completed successfully.	Terminal
`ABORTED`	Canceled by the user.	Terminal
`SUSPENDED`	Canceled by Automation Pilot (runtime limit exceeded, tenant execution limit reached, or execution/output size limit exceeded).	Terminal

Terminal statuses (FINISHED, ABORTED, SUSPENDED) mean the execution is complete and no further actions can be applied. Completed executions can be retriggered and deleted.

Active statuses (RUNNING, PAUSED, INPUT_REQUIRED, FAILED) mean the execution is still live and can be acted upon.

Actions

Each action is only valid from specific statuses — applying an action from the wrong status returns HTTP 409.

Action	Valid from	Result
PAUSE	`RUNNING`	→ `PAUSED`
RESUME	`PAUSED`	→ `RUNNING`
RETRY	`FAILED`	→ `RUNNING`. Re-attempts the execution, preserving progress for provided composite commands.
RESET	`FAILED`, `PAUSED`, `INPUT_REQUIRED`	→ `RUNNING` from a chosen previous child command. Not available for direct executions of provided commands. For provided composite commands, starts the whole command from the beginning (unlike RETRY which preserves progress).
ABORT	`PAUSED`, `FAILED`	→ `ABORTED`
COMMENT	Any	Adds a comment to the execution's Action Log (max 10 comments per execution).

EXEC_ID="execution-uuid-here"

# Pause
curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "PAUSE", "reason": "Blocking on manual verification"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

# Resume
curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "RESUME", "reason": "Resuming after verification"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

# Retry
curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "RETRY", "reason": "Retrying after transient failure"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

# Abort
curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "ABORT", "reason": "Superseded by a newer run"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

Triggering Executions

Note: To retrieve logs after execution, include feature:logs in the tags of the trigger request. This tag is opt-in and cannot be added retroactively — an execution triggered without it will return no logs from GET /executions/{id}/logs.

Trigger a Command

Add "feature:dryRun": "" to the tags to execute without making actual changes.

# Basic trigger (opt-in feature:logs tag enables log retrieval)
curl -s -X POST \
  -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{
    "commandId": "my-catalog:MyCommand:1",
    "inputValues": {
      "param1": "value1",
      "param2": "value2"
    },
    "tags": {"feature:logs": ""}
  }' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions"

# With dry run mode (adds feature:dryRun tag)
curl -s -X POST \
  -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{
    "commandId": "my-catalog:MyCommand:1",
    "inputValues": {"param1": "value1"},
    "tags": {"feature:logs": "", "feature:dryRun": ""}
  }' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions"

All inputValues entries must be strings on the wire, regardless of the declared parameter type. The runtime converts each value to its declared type inside the executor:

"inputValues": {
  "appName": "my-app",
  "instances": "3",
  "enabled": "true",
  "regions": "[\"cf-eu10\",\"cf-us10\"]",
  "config": "{\"maxRetries\":5}"
}

Monitoring Executions

List Executions

# List all (up to limit)
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions?limit=100"
# Filter by status
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions?status=FAILED&limit=50"
# Filter by command
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions?commandId=my-catalog:MyCommand:1"
# Filter by time (epoch milliseconds)
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions?startedAfter=1707000000000"

Get Execution Details

EXEC_ID="execution-uuid-here"
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID"

Get Execution Summary

# Get counts by status
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/summary"

Get Input/Output

# Get execution input
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/input"
# Get execution output (only for FINISHED)
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/output"

Controlling Executions

All actions use the same endpoint: POST /api/v1/executions/$EXEC_ID/actions

See the Actions table in the Execution Lifecycle section for valid statuses and expected outcomes.

EXEC_ID="execution-uuid-here"

# Reset — resets to a previous child command (provide the executor path via "value")
curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "RESET", "value": "stepAlias", "reason": "Retry after fixing the input"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

# Comment — add a note to the Action Log (max 10 per execution)
curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "COMMENT", "reason": "Escalated to L2 on-call"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

# List action history
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

Viewing Logs

Get Execution Logs

EXEC_ID="execution-uuid-here"

# Get paginated logs
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/logs?page=0&maxPageSize=20"
# Get logs for specific executor
EXECUTOR_PATH="step1"
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/logs/$EXECUTOR_PATH"

Troubleshooting

Quick Troubleshooting

EXEC_ID="your-execution-id"

# 1. Get status and error
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID"

# 2. Check input used
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/input"

# 3. Check recent logs
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/logs?page=0&maxPageSize=10"

# 4. Check output (if FINISHED)
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/output"

Common Issues

Execution Stuck in RUNNING

Check logs for the current step
Look for external dependencies (HTTP timeouts, etc.)
Consider aborting and retrying

curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/logs?page=0&maxPageSize=20"

curl -s -X POST -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{"type": "ABORT", "reason": "Aborting stuck execution"}' \
  "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/actions"

Execution Failed

Get execution details to see error
Check logs for the failed step
Review input parameters

curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID"
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID/input"

Input Required

The execution is waiting for manual input:

curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID"
# Inspect the response — check the suspendedStep field

No Logs Available

Cause: The execution was triggered without the feature:logs tag, so logging was not enabled. Solution: Always include "feature:logs": "" in the tags object when triggering. Logs cannot be retrieved retroactively for executions that were triggered without this tag. Re-trigger the execution with the tag included.

Filtering & Pagination

Query Parameters for List Executions

Parameter	Type	Description
`status`	array	Filter by status (RUNNING, FAILED, FINISHED, etc.)
`commandId`	array	Filter by command ID
`executionId`	array	Filter by execution ID
`startedAfter`	integer	Unix epoch milliseconds
`startedBefore`	integer	Unix epoch milliseconds
`tag`	array	Key-value pairs
`limit`	integer	Max results (1-1000, default 1000)

Examples

# Last 24 hours, failed only
YESTERDAY=$(($(date +%s) * 1000 - 86400000))
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  "https://$AUTOPI_HOSTNAME/api/v1/executions?status=FAILED&startedAfter=$YESTERDAY&limit=50"
# Multiple statuses
curl -s -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  "https://$AUTOPI_HOSTNAME/api/v1/executions?status=RUNNING&status=PAUSED"

Delete Executions

Only completed executions can be deleted (not RUNNING).

EXEC_ID="execution-uuid-here"
curl -s -X DELETE -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" "https://$AUTOPI_HOSTNAME/api/v1/executions/$EXEC_ID"

Event Triggers

Trigger via Generic Event

curl -s -X POST \
  -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{
    "commandReference": "my-catalog:MyCommand:1",
    "event": {
      "type": "alert",
      "message": "High CPU usage detected"
    }
  }' \
  "https://$AUTOPI_HOSTNAME/api/v1/triggers/generic-event"

Trigger via ANS Event

curl -s -X POST \
  -u "$AUTOPI_USERNAME:$AUTOPI_PASSWORD" \
  -H "Content-Type: application/json" \
  -d '{
    "commandReference": "my-catalog:MyCommand:1",
    "event": {...ANS event payload...}
  }' \
  "https://$AUTOPI_HOSTNAME/api/v1/triggers/ans-event"

Error Handling

HTTP Status Codes

Code	Description
`200`	Success
`202`	Accepted (trigger/action)
`204`	No Content (delete)
`400`	Bad Request
`401`	Unauthorized
`403`	Forbidden
`404`	Not Found
`409`	Conflict (invalid state)
`413`	Payload Too Large
`429`	Rate Limited

Permissions Required

Operation	Permission
List/Get executions	`Read`
Delete executions	`Write`
Trigger/Actions	`Execute`

Examples

Example 1: Trigger and monitor a command

User: "Run my health check command"

POST to /api/v1/executions with commandId, input values, and "tags": {"feature:logs": ""}
Capture the execution ID from the response
GET /api/v1/executions/$EXEC_ID and poll until status is no longer RUNNING
Report the final status and output

Example 2: Troubleshoot a failed execution

User: "My execution failed, what went wrong?"

GET /api/v1/executions/$EXEC_ID — extract status, error, progressMessage
GET /api/v1/executions/$EXEC_ID/input — verify the parameters that were used
GET /api/v1/executions/$EXEC_ID/logs?page=0&maxPageSize=20 — find the failed step
Report the root cause from the error message and log output

Example 3: Find and abort stuck executions

User: "Are there any stuck executions? Abort them."

GET /api/v1/executions?status=RUNNING to list all currently running executions
For each one, check startedAt to identify those running unexpectedly long
For each stuck execution, POST {"type": "ABORT", "reason": "Stuck execution"} to /api/v1/executions/$EXEC_ID/actions
Confirm each abort returned HTTP 202

MAKE IT PART OF YOUR WORKFLOW

安装技能

TERMINAL

npx skills add williamcorrea23/sap-router-skill --skill automation-pilot-executions-api

复制命令到本地终端执行；安装前请先审查源代码和许可证。

ZIP · 完整技能目录，含资源与许可证

SKILL.md · 仅技能说明原文

仓库 Star: 未同步
许可证: MIT
版本 / 分支: HEAD · a9c8529b
仓库更新时间: 未同步

安装前，请检查技能脚本与权限。

源代码

automation-pilot-executions-api

技能说明

name: automation-pilot-executions-api description: Monitor and manage SAP Automation Pilot executions via API. Use when triggering commands, checking execution status, retrieving logs, aborting or pausing executions, or troubleshooting failed runs.

SAP Automation Pilot Executions API Management

Prerequisites

Execution Lifecycle

Statuses

Actions

Triggering Executions

Trigger a Command

Monitoring Executions

List Executions

Get Execution Details

Get Execution Summary

Get Input/Output

Controlling Executions

Viewing Logs

Get Execution Logs

Troubleshooting

Quick Troubleshooting

Common Issues

Execution Stuck in RUNNING

Execution Failed

Input Required

No Logs Available

Filtering & Pagination

Query Parameters for List Executions

Examples

Delete Executions

Event Triggers

Trigger via Generic Event

Trigger via ANS Event

Error Handling

HTTP Status Codes

Permissions Required

Examples

Example 1: Trigger and monitor a command

Example 2: Troubleshoot a failed execution

Example 3: Find and abort stuck executions

发现更多相关技能

计划执行与检查节点

女娲：花叔原版

仓颉：方法论版 v2.5.0