By taipt1504
Adaptive 5-layer workflow Claude Code plugin for Java Spring backend — 1% pre-flight discovery, triage routing, two-stage review, subagent dispatch, auto-evolution
Grill ambiguity, surface assumptions, extract vocabulary. Run before Plan when requirements vague (standard lane) or always (high-stakes lane). Output feeds Plan + Spec gates.
Tournament-style solution exploration. Generates 2-5 options with adversarial challenges, recommends with trade-offs. Mandatory high-stakes (≥3 options), conditional standard (multi-path).
Incrementally fix Java/Gradle/Maven build and compilation errors for Spring Boot projects. One error at a time for safety.
Execute gate — dispatches one slice-executor subagent per plan slice (parallel where independent). Reads plan + spec + pre-flight 4 artifact. Lane-aware (trivial: direct execute, no dispatch).
Create and validate Flyway database migrations for schema changes. Supports MySQL and PostgreSQL.
> Common protocol ALL agents follow. Referenced via `protocol: _shared-protocol.md` frontmatter.
High-stakes lane pre-plan architectural review. Reads ADRs + CONTEXT.md + brainstorm artifact. Validates chosen solution against existing architecture. Output feeds Plan gate. Optional for standard lane, auto-fire for high-stakes.
Build and compilation error resolution specialist for Java/Gradle/Maven projects. Use PROACTIVELY when build fails or compilation errors occur. Fixes build/compilation errors only with minimal diffs, no architectural edits. When NOT to use: for test failures (use slice-executor), for architectural refactoring (use planner).
Stage 2 of two-stage review. Spec compliance already verified by Stage 1. Reviews security, performance, maintainability, readability, test quality. Severity-tagged output (P0/P1/P2/P3/P4) with MANDATORY rule ID citation per finding.
Decompose chosen solution into vertical slices with dependency graph + risk register. Consumes Align + Brainstorm artifacts. Output feeds Spec gate. Use after Brainstorm gate (high-stakes) or directly after Align (standard with obvious solution).
Surface ambiguity, missing requirements, hidden assumptions BEFORE planning. Fires after triage for standard (vague request) or high-stakes (always). Updates CONTEXT.md vocabulary. Produces agreed-requirements artifact feeding Plan gate.
REST API design patterns for Spring Boot — HTTP methods, status codes, URL conventions, RFC 7807 ProblemDetail errors, pagination, versioning, validation, and OpenAPI documentation. Use when designing REST endpoints, choosing HTTP status codes, implementing error responses, adding pagination to list APIs, versioning APIs, or generating OpenAPI/Swagger specs.
Hexagonal architecture, DDD, CQRS, and event sourcing for Java Spring applications. Use when designing package structure, creating Aggregate Roots or Value Objects, implementing ports and adapters, setting up CQRS command/query separation, wiring domain events, writing ArchUnit tests, or creating solution architecture documents and ADRs.
Foundation skill — auto-loads at SessionStart. Teaches 5-layer adaptive workflow, 1% pre-flight discovery rule, triage routing, skill announcement contract, project detection. All other skills depend on this. Do NOT manually load.
Tournament-style iterative solution exploration. Generates 2-5 candidates via adversarial challenges, compares multi-dimensionally, recommends with explicit trade-offs. Mandatory high-stakes (≥3), conditional standard. Output feeds Plan + ADR for high-stakes.
Matches all tools
Hooks run on every tool call, not just specific ones
Executes bash commands
Hook triggers when Bash tool is used
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Modifies files
Hook triggers on file write and edit operations
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
A Claude Code plugin for Java Spring backend development. v4.1 ships a fully agentic 5-layer adaptive workflow: triage → align → brainstorm → plan → spec → execute → verify → review → learn.
v4.1.0-beta. Agent = Model + Harness. Plugin provides the harness — 1% pre-flight discovery, lane-aware workflow, native Claude auto-memory, MCP optional layer, hook output caps, lifecycle-organized state, template-enforced plans/specs.
REQ → BOOT → PREFLIGHT0 → TRIAGE
├── trivial → EXECUTE (light) → REVIEW (S2) → COMMIT
└── standard / high-stakes →
ALIGN → BRAINSTORM (if needed) →
PLAN → SPEC → EXECUTE (subagent per slice) →
REVIEW (S1 + S2) → LEARN → COMMIT
Three lanes. Six pre-flight variants (one per gate). 1% rule: every gate enumerates ALL skills + rules with ≥1% relevance, justifies every SKIP with concrete evidence.
Full operational reference: WORKING_WORKFLOW.md. End-to-end walkthroughs: examples/. Migration from v1 (5-phase rigid): MIGRATION_v2.md.
# 1. Install (one-time per machine)
/plugin marketplace add taipt1504/agent-skills
/plugin install devco-agent-skills
# 2. Setup project (one-time per project)
/dc-setup
# 3. Commit + share
git add .claude/ && git commit -m "chore: add Claude Code project context"
# 4. Start coding
/triage "describe your task" # auto-routes to trivial / standard / high-stakes lane
Teammates clone → claude → auto-prompted to install. Zero setup.
| Category | Count | Lazy-loaded? |
|---|---|---|
| Agents | 8 | yes (Task tool) |
| Skills | 28 | yes (Skill tool — only description in catalog) |
| Commands | 16 | yes (invoke time) |
| Rules | 18 (common 8 + java 7 + summer 3) | partial — summer/ conditional on io.f8a.summer detection |
| Hooks | 20 | always — registered via hooks/hooks.json |
| Templates | 7 | manual reference |
Plugin bundle: ~628 KB / ~157k tokens (12% reduction from v3.3 baseline).
| Layer | Gate | Output |
|---|---|---|
| 0 — Boot | session-init.sh | Detect project, load CONTEXT.md + ADRs, pre-flight 0 |
| 1 — Triage | /triage | Lane: trivial / standard / high-stakes |
| 2 — Align (if vague / high-stakes) | /align | Requirements + assumptions + vocabulary → CONTEXT.md |
| 2 — Brainstorm (if multi-path / mandatory high-stakes) | /brainstorm | ≥3 options + comparison matrix + recommendation + ADR |
| 2 — Plan | /plan | Slices + dependency graph + risk register |
| 2 — Spec | /spec | Per-slice behavioral contracts (template-enforced) |
| 3 — Execute | /build | Subagent dispatch per slice (worktree for high-stakes) |
| 3 — Verify | /verify | Compile + tests + coverage + (high-stakes) security scan |
| 4 — Review S1 | /dc-review Stage 1 | Spec compliance binary verdict |
| 4 — Review S2 | /dc-review Stage 2 | Severity-tagged quality findings |
| 5 — Learn | /meta evolve | Pattern observations → Claude native auto-memory |
5 layers, native-first:
LAYER 1 — Auto-loaded by Claude Code
~/.claude/CLAUDE.md user-level
./CLAUDE.md or ./.claude/CLAUDE.md project-level
./CLAUDE.local.md per-user project
~/.claude/projects/<project>/memory/MEMORY.md Claude native auto-memory (≤200 lines/25KB)
LAYER 2 — Loaded on-demand by Claude Code
./.claude/rules/*.md with paths: frontmatter path-conditional
~/.claude/projects/<project>/memory/<topic>.md Claude native topic files
LAYER 3 — Plugin / agents / hooks (NOT auto-loaded)
Plugin SKILL.md bodies Skill tool
Plugin agent .md Task tool dispatch
Plugin rules/{common,java,summer}/*.md agent Read in pre-flight
./.claude/memory/active/<feature-id>/ per-feature state
Hook additionalContext capped via DEVCO_*_MAX_CHARS
LAYER 4 — MCP memory server (optional)
mcp__memory__search_nodes / read_graph cross-feature search
mcp__memory__create_entities persist decisions
Plugin probes availability; degrades to filesystem-only if absent
LAYER 5 — Git-tracked durable docs
./CONTEXT.md, ./docs/adr/*.md
./docs/plans/**, ./docs/specs/**
.claude/memory/)npx claudepluginhub taipt1504/agent-skills --plugin devco-agent-skills7-phase agentic workflow for Java/Spring Boot backends (over a pre-indexed codebase): discover → brainstorm → spec → plan → implement → review → learn, with complexity-tier fast lanes.
A collection of agents, skills, hooks, commands, and rules for AI Agentic development workflow
Best skills for Spring Boot development powered by Amplicode Team
Comprehensive Java development toolkit with Spring Boot, testing, LangChain4J, and AWS integration
Spring Boot scaffold skill and Spring expert agent for Java 8+ projects
Autonomous multi-agent development framework with spec-driven sprints and convergent iteration
Engineering discipline layer for Claude Code — 5 workflows, 69 commands, 21 rules, 29 skills, 9 agents organized in 12 packs