Agent Skills
› NeverSight/learn-skills.dev
› ac-task-executor
ac-task-executor
GitHub用于自主编码中执行功能实现任务，管理任务队列并协调规划、实施与验证阶段。支持单功能、队列及批量执行模式，涵盖从选择到提交的全生命周期，具备错误处理与状态追踪能力。
Trigger Scenarios
运行功能实现
执行构建任务
处理功能队列
编排任务完成
Install
npx skills add NeverSight/learn-skills.dev --skill ac-task-executor -g -y
SKILL.md
Frontmatter
{
"name": "ac-task-executor",
"description": "Execute implementation tasks in autonomous coding. Use when running feature implementations, executing build tasks, processing feature queue, or orchestrating task completion."
}
AC Task Executor
Execute implementation tasks for autonomous coding features.
Purpose
Orchestrates the execution of feature implementations, managing the task queue and coordinating between planning, implementation, and verification phases.
Quick Start
from scripts.task_executor import TaskExecutor
executor = TaskExecutor(project_dir)
result = await executor.execute_next_feature()
Execution Modes
Single Feature
Execute one feature at a time:
result = await executor.execute_feature("auth-001")
Queue Mode
Process feature queue continuously:
await executor.execute_queue(max_features=10)
Batch Mode
Execute multiple features in a batch:
results = await executor.execute_batch(["auth-001", "auth-002"])
Task Lifecycle
1. SELECT → Choose next feature from queue
2. PREPARE → Load context, generate tests (RED)
3. IMPLEMENT → Write code to pass tests (GREEN)
4. REFACTOR → Clean up while tests pass
5. VALIDATE → Run criteria validation
6. COMMIT → Git commit changes
7. UPDATE → Mark feature as passes: true
8. NEXT → Continue to next feature
Execution Result
{
"feature_id": "auth-001",
"status": "completed",
"phases": {
"prepare": {"success": true, "duration_ms": 1500},
"implement": {"success": true, "duration_ms": 45000},
"validate": {"success": true, "duration_ms": 5000},
"commit": {"success": true, "commit_hash": "abc123"}
},
"tests": {
"total": 5,
"passed": 5,
"failed": 0
},
"metrics": {
"tokens_used": 15000,
"estimated_cost": 0.23
}
}
Error Handling
- Validation Failure: Retry implementation up to 3 times
- Test Failure: Analyze failures, adjust approach
- Timeout: Save state, create handoff for continuation
- Critical Error: Rollback to last checkpoint
Integration
- Uses:
ac-state-tracker,ac-test-generator,ac-criteria-validator - Triggers:
ac-commit-manager,ac-qa-reviewer - Reports to:
ac-session-manager
API Reference
See scripts/task_executor.py for full implementation.
Version History
- e0220ca Current 2026-07-05 23:06
