Agent skill
subagent-driven-development
(ePost) Use when a plan has 3+ independent tasks, phases have non-overlapping file ownership, or user says "parallel implementation", "execute all phases" — provides per-task subagent dispatch with two-stage review
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/subagent-driven-development-klara-copilot-epost-agent-kit
Metadata
Additional technical details for this skill
- keywords
-
parallel subagent dispatch implementation-loop two-stage-review
- platforms
-
[ "all" ] - connections
-
{ "enhances": [ "plan" ] } - agent affinity
-
[ "epost-project-manager", "epost-fullstack-developer", "epost-planner" ]
SKILL.md
Subagent-Driven Development
When to Use
Activate when a plan has 3+ independent tasks that can be implemented and reviewed separately. Do NOT use for single-task or tightly-coupled work.
Per-Task Loop
For each task in the plan:
- Dispatch Implementer — Fresh subagent with task context (see
references/implementer-prompt.md) - Spec Review — Fresh subagent verifies code matches spec (see
references/spec-reviewer-prompt.md) - Quality Review — Fresh subagent reviews code quality (see
references/code-quality-reviewer-prompt.md) - Fix Loop — If either review fails: fix → re-review (max 3 iterations)
- Mark Done — Update plan TODO, move to next task
Two-Stage Review
Every task gets TWO independent reviews, in order:
Stage 1: Spec Compliance
- Does the code do what the spec says?
- Line-by-line comparison against requirements
- Output: ✅ matches / ❌ deviates with file:line
Stage 2: Code Quality
- Is the code well-written?
- Security, performance, maintainability
- Only runs AFTER spec compliance passes
Rules
- Never start work on main/master without user consent
- Never skip either review stage
- Never dispatch parallel implementation subagents (serial only — one task at a time)
- Each subagent gets a FRESH context (no stale state)
- Max 3 fix-review iterations per task; escalate to user after 3
Task Dispatch Order
- Foundation tasks first (shared utilities, types, config)
- Independent features in plan order
- Integration tasks last (wire components together)
Prompt Templates
references/implementer-prompt.md— Task context and self-review checklistreferences/spec-reviewer-prompt.md— Skeptical spec comparisonreferences/code-quality-reviewer-prompt.md— Quality review after spec passes
Related Skills
verification-before-completion— Verify each task before marking donecode-review— Quality review patternsplan— Plan structure that feeds task dispatch
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?