Sprint Management Skills
Relevant source files
The following files were used as context for generating this wiki page:
Sprint Management Skills encompass the automated and semi-automated workflows used to plan, initialize, execute, and monitor development cycles within the eng-platform. These skills leverage a combination of specialized AI agents, configuration-driven task definitions (TOML), and external tool integrations like Beads to orchestrate complex software engineering tasks.
The system is designed to provide high-level orchestration, ensuring that tasks are decomposed with proper dependencies, assigned to agents with appropriate capabilities, and tracked through a rigorous lifecycle protocol. This ensures 50-70% faster execution compared to sequential manual workflows.
Sources: internal/agents/project-manager-context.md:1-20, mcp-servers/beads-integration/README.md:1-10
Sprint Planning and Initialization
The initialization phase establishes the foundational structure required for sprint management. This includes setting up standardized folder structures, Git workflows, and the initial sprint configuration file.
Initialization Workflow
The /initialize skill automates the setup of the project environment, creating essential directories such as sprints/, docs/patterns/, and .claude/skills/. It also generates an initial sprints/SPRINT_INITIAL.toml to baseline the project.
Sources: skills/initialize/SKILL.md:10-40, skills/initialize/SKILL.md:460-485
Sprint TOML Configuration
Sprints are defined using TOML files located in the sprints/ directory. These files serve as the "source of truth" for the sprint's goals, versioning, and task lists.
| Field | Type | Description |
|---|---|---|
id | Integer | Unique identifier for the sprint. |
version | String | Target release version (e.g., "v2.1.0"). |
github_issues | Array | List of GitHub issue IDs associated with the sprint. |
tasks | Array of Tables | Detailed task definitions including titles and assignees. |
depends_on | Array | List of task IDs that must be completed first. |
Sources: mcp-servers/beads-integration/README.md:148-165, scripts/orchestrate-sprint.js:46-92
Task Orchestration and Execution
The execution of a sprint is managed by the Project Manager (PM) Agent and the Beads integration, which handle task decomposition and multi-agent coordination.
The Beads Integration Flow
The system uses Beads for subtask decomposition, allowing complex TOML definitions to be broken down into granular, dependency-tracked units.
The orchestration script ensures that GitHub issues are validated and the working directory is clean before subtasks are mapped to specific agents.
Sources: scripts/orchestrate-sprint.js:10-30, mcp-servers/beads-integration/README.md:10-30
Multi-Agent Coordination
Tasks are assigned to specialized agents based on their role (e.g., agent-api, agent-ui, agent-test). The PM Agent builds a Directed Acyclic Graph (DAG) to identify the critical path and parallelizable tasks.
Sprint Task Lifecycle Protocol (Pattern-TRACKING-001)
A mandatory protocol is enforced for every task to ensure the Sprint Panel UI and TOML files accurately reflect current work. This protocol applies to all agents, including Documentation, Test, and Infrastructure agents.
Lifecycle Status Transitions
- Before Starting: Update status to
in_progressusing/sprint-task-lifecycle start {TASK-ID}. - After Completion: Update status to
completedand set thecompleted_date. - If Blocked: Update status to
deferredand document thedeferred_reason.
Sources: internal/agents/documentation-agent-context.md:273-315, internal/agents/infrastructure-agent-context.md:55-100
Automated Validation
A pre-commit hook runs validate-sprint-schema.js to ensure that TOML updates comply with the required schema and that all documents are properly linked to tasks.
Sources: skills/sprint-plan/CHANGELOG.md:15-25, internal/agents/project-manager-context.md:315-320
Monitoring and Metrics
The orchestrate-sprint.js script and Beads MCP tools provide real-time visibility into sprint progress.
Progress Tracking
The beads_sprint_progress tool returns a summary of total, completed, in-progress, and blocked tasks.
| Metric | Calculation | Goal |
|---|---|---|
| Completion Rate | (Done / Total) * 100 | 100% by sprint end |
| Cycle Time | Avg time from in_progress to completed | Minimize per agent type |
| Blocked Time | Total time in deferred status | Identify infrastructure bottlenecks |
Sources: mcp-servers/beads-integration/README.md:65-75, scripts/orchestrate-sprint.js:300-330
Sprint-Aware Naming Conventions
To prevent document collisions across different sprints, a naming convention is enforced (Pattern-DOCS-002): {SPRINT_ID}_{TASK_ID}_{TYPE}.md. For example, 17.1-BUGS_BUG-011_ENHANCED_PROMPT.md ensures that tasks in Sprint 17.1 do not conflict with similar IDs in previous sprints.
Sources: skills/sprint-plan/CHANGELOG.md:10-20
Summary
Sprint Management Skills provide a robust framework for autonomous execution. By combining the Project Manager Agent's dependency analysis with the Beads MCP's task tracking and the Pattern-TRACKING-001 lifecycle protocol, the platform ensures that complex engineering goals are met with high technical accuracy and minimal manual overhead. The transition from manual planning to automated multi-agent execution is the core value proposition of these skills.
Sources: internal/agents/project-manager-context.md:255-270, mcp-servers/beads-integration/README.md:142-145