AgentSkillsCN

types

Splitrail 核心数据类型的参考文档。在处理 ConversationMessage、Stats、DailyStats 或其他类型定义时使用。

SKILL.md
--- frontmatter
name: types
description: Reference for Splitrail's core data types. Use when working with ConversationMessage, Stats, DailyStats, or other type definitions.

Key Types

Read src/types.rs for full definitions.

Core Types

  • •

    ConversationMessage - Normalized message format across all analyzers. Contains application source, timestamp, hashes for deduplication, model info, token/cost stats, and role.

  • •

    Stats - Comprehensive usage metrics for a single message including token counts, costs, file operations, todo tracking, and composition stats by file type.

  • •

    DailyStats - Pre-aggregated stats per date with message counts, conversation counts, model breakdown, and embedded Stats.

  • •

    Application - Enum identifying which AI coding tool a message came from.

  • •

    MessageRole - User or Assistant.

Hashing Strategy

  • •local_hash: Deduplication within a single analyzer
  • •global_hash: Deduplication on upload to Splitrail Cloud

Aggregation

Use crate::utils::aggregate_by_date() to group messages into daily stats. See src/utils.rs.