SuperAPI AI Directory

sprint-team

Agents & models

View repository

Updated 10-16 01:34

About this skill

Sprint Team — Agent Role Definitions

This skill defines the team roles for a sprint-based improvement loop. Each role is a standalone markdown file in references/ that can be loaded as a prompt template by the orchestrator or by agents directly.

Roles

Role File Phase Purpose
Researcher references/researcher.md Research Scans codebase for bugs, tech debt, coverage gaps
Architect references/architect.md Plan Produces implementation plan with work units
Security Reviewer references/security-reviewer.md Review Gate Reviews plan for security implications
Quality Reviewer references/quality-reviewer.md Review Gate Reviews plan for test coverage and correctness
Scope Reviewer references/scope-reviewer.md Review Gate Reviews plan for feasibility and blast radius
Coder references/coder.md Execute Implements work units with TDD
Validator references/validator.md Validate Independently verifies coder's work (cross-model)
Integrator references/integrator.md Integrate Merges sub-PRs, runs full suite, fixes breakage
Retrospector references/retrospector.md Retrospective Captures learnings for future runs

Sprint Pipeline

Research -> Plan -> Review Gate -> Decompose -> Execute -> Validate -> Integrate -> PR -> Retrospective

Template Variables

Role files use {{variable}} placeholders. The orchestrator substitutes these at spawn time:

Variable Description
{{project_dir}} Absolute path to the project
{{base_branch}} The improvement branch name
{{src_files}} Newline-separated list of source files to review
{{knowledge_entries}} Prior knowledge base entries (may be empty)
{{coverage_minimum}} Coverage floor percentage (e.g. 90)
{{coverage_command}} Full enforcement command
{{coverage_baseline}} Coverage percentage at sprint start
{{work_unit_json}} JSON blob describing one work unit
{{coder_backend}} Backend the coder used (for cross-model validation)
{{focus_areas}} Reviewer-specific focus areas

Usage

Any agent can read a role file and adopt that persona:

# As an orchestrator spawning a researcher
orca spawn "$(cat skills/sprint-team/references/researcher.md)" \
  -b cc -d ~/proj --base-branch improve/sprint-xxx --orchestrator cc

The orchestrator loads these files and substitutes the template variables before spawning workers.