history-manager
This skill contains shell command directives (!`command`) that may execute system commands. Review carefully before installing.
History Manager
Provides iteration and phase state management for the full-stack workflow. Other skills invoke this skill to read/write docs/history.json instead of managing state directly.
Available Commands
All commands use the bundled cross-platform Python 3 script (macOS + Windows):
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py <command> [args]
init
Initialize docs/history.json if it does not exist.
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py init
status
Print the current iteration state, showing which phases are completed and which is next.
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py status
new-iter
Create a new iteration (e.g. iter-001), create its docs/iter-xxx/ directory, and set it as current.
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py new-iter
set-phase <phase> <phase_id>
Record a phase as completed: set its id in the current iteration and advance the state to the next phase.
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py set-phase requirements req-001
Valid phase names: requirements, technology_selection, technical_review, program_design, task_breakdown, unit_testing, code_development, self_verification.
get-phase <phase>
Get the current phase id for a given phase (returns null if not yet started).
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py get-phase requirements
check-file <phase> <phase_id>
Check whether the document file for a given phase exists in the current iteration directory.
python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py check-file requirements req-001
Returns exists:<path> or missing:<path>.
How Other Skills Should Use This
- Before starting a phase: Invoke
history-managerto runget-phase <phase_name>andcheck-fileto determine if the phase was already completed. Skip if both the id exists and the file exists. - After completing a phase: Invoke
history-managerto runset-phase <phase_name> <phase_id>to record completion and advance the state machine. - dev-workflow orchestrator: Invokes
history-managerforinit,new-iter, andstatusat startup, then delegates phase state updates to each phase skill (which in turn invokeshistory-manager).
Current Status
!python ${CLAUDE_SKILL_DIR}/scripts/history_manager.py status
More from rainlib/full-stack-skill
dev-workflow
全栈开发工作流编排器,按 8 阶段顺序执行:需求分析 → 技术选型 → 技术评审 → 程序设计 → 任务拆分 → 单元测试 → 代码开发 → 自我验证。
2unit-testing
根据程序设计与任务拆分产出单元测试用例与预期结果,为代码开发提供验收标准。全栈工作流第 6 阶段。
1technical-review
评审需求与技术选型的架构可行性,定义 API 契约、数据模型、识别风险项。在技术选型之后、程序设计之前使用。
1program-design
根据需求、技术选型与技术评审设计程序结构、流程与工程目录。全栈工作流第 4 阶段。
1task-breakdown
将程序设计拆分为可管理的开发任务单元,定义优先级、依赖和开发顺序。在程序设计之后、单元测试之前使用。
1code-development
根据单元测试用例、任务拆分与程序设计实现代码,并通过测试验证。全栈工作流第 7 阶段。
1