AgentSkillsCN

telemetry-logger

记录并结构化来自自主操作的遥测数据。当您需要记录决策、结果以及系统状态，以便进行审计追踪、经验学习和操作人员分析时，可选用此功能。该工具可与各类日志后端集成（目前基于文件存储，未来将支持 Aspire 或 QuestionManager 集成）。

SKILL.md
--- frontmatter
name: telemetry-logger
version: specs-v1.0
description: Records and structures telemetry data from autonomous operations. Use when you need to log decisions, results, and system state for audit trails, learning, and operator analysis. Integrates with logging backends (currently file-based, future Aspire/QuestionManager integration).
license: Internal use. RoadTrip project.

Telemetry Logger Skill

Overview

Logs structured telemetry from autonomous workflows. Phase 1: File-based. Phase 2: Integration with QuestionManager/Aspire.

What It Logs

  • •Decision points: What was decided and why
  • •Confidence scores: How confident was each decision (0-1)
  • •Execution results: What succeeded, what failed, why
  • •Timing: When each step occurred
  • •Context: Files affected, branches, operators, etc.

Input Format

json
{
  "timestamp": "ISO8601",
  "orchestrator": "name",
  "decision": "APPROVED|BLOCKED|ERROR",
  "steps": [ { "name": "str", "status": "PASS|FAIL", "confidence": 0.0-1.0 } ],
  "context": { "files": [], "branch": "str" },
  "result": { "success": bool, "error": "str" }
}

Output

json
{
  "status": "RECORDED",
  "log_id": "uuid",
  "location": "path/to/log/file"
}

Configuration

Reads: config/telemetry-config.yaml

yaml
log_dir: logs/
log_level: DEBUG|INFO|WARN|ERROR
retention_days: 30
format: JSON  # Phase 1: JSON files; Phase 2: Aspire integration

Phase 2 Integration (Placeholder)

  • •Send logs to QuestionManager for analysis
  • •Real-time dashboard of skill decisions
  • •Auto-detect patterns for learning

Status: Phase 1 stub. Core logic: append structured JSON to log file.