hmj1026/dhpk

hmj1026/dhpk

Agent

Reference index for the agents shipped by the dhpk plugin.

hmj1026/dhpk

Agent

Output-quality evaluator. Scores a completed agent run on a 5-axis rubric (accuracy, completeness, clarity, actionability, conciseness) with grep-verified evidence and a deliver / fix / redo verdict. Use when the user wants an objective quality assessment of a finished task. Scores run OUTPUT — distinct from…

hmj1026/dhpk

Agent

One-release operational compatibility forwarder to agy-worker.

hmj1026/dhpk

Agent

CLI-backed mechanical implementer — the agy variant of fast-worker. Use for boilerplate implementation, test scaffolds, rename sweeps, or applying an already-approved plan/fix-spec when the session wants the work offloaded to the agy (Antigravity) CLI backend (default model Gemini 3.6 Flash (High)) as a cheap…

hmj1026/dhpk

Agent

DDD architecture specialist (framework-agnostic). Use for cross-module design decisions, DDD layer placement (Interface → Domain → Infrastructure, or your stack equivalent), refactoring strategy, and technical-debt analysis. Loads stack-specific layering examples on demand when a matching module is active.

hmj1026/dhpk

Agent

Expert code review specialist. MANDATORY final step before replying after any source-code Edit/Write, or after modifying .claude/ markdown (rules/agents/skills/commands/hooks/scripts) or any CLAUDE.md file. Reviews quality, security, and maintainability. Do NOT skip when: user approved a plan, change seems small…

hmj1026/dhpk

Agent

A forwarding agent that sends a specified task to another Codex model through the command line and returns that model’s response. It is intended for a separate implementation pass or independent review.

hmj1026/dhpk

Agent

One-release operational compatibility forwarder to codex-reasoner.

hmj1026/dhpk

Agent

One-release operational compatibility forwarder to codex-worker.

hmj1026/dhpk

Agent

CLI-backed read-only deep-reasoning worker — the codex variant of deep-reasoner. Use for root-cause analysis, algorithm design, complex multi-file debugging, and design synthesis during the implement phase when the --reasoner=codex backend is selected (default gpt-5.6-sol @ high) instead of the in-process opus…

hmj1026/dhpk

Agent

Internal shared-runner Codex review role; capability-gated and not a native Codex dispatch target.

hmj1026/dhpk

Agent

CLI-backed mechanical implementer — the codex variant of fast-worker. Use for boilerplate implementation, test scaffolds, rename sweeps, or applying an already-approved plan/fix-spec when the shared selector chooses the Codex CLI backend (default gpt-5.6-luna @ xhigh) instead of the in-process sonnet worker.…

hmj1026/dhpk

Agent

Database review specialist (relational + object stores, framework-agnostic). MANDATORY final step after writing migrations, SQL queries, Repository methods, or schema changes. Checks prepared statements, index efficiency, N+1 issues, transaction correctness. Do NOT skip when: the change seems small, manual…

hmj1026/dhpk

Agent

Read-only deep-reasoning worker. Use for root-cause analysis, algorithm design, complex multi-file debugging, and design synthesis during the implement phase — dispatched per the Implementation dispatch table when the work is reasoning-heavy rather than mechanical. Returns a conclusion contract (conclusion + file:line…

hmj1026/dhpk

Agent

Lightweight policy / harness doc reviewer. MANDATORY final step after editing .claude/{agents,rules,commands,skills,manifests}//.{md,json,yml,yaml}, doc-only files under .claude/{hooks,scripts}/ (non-.sh), top-level CLAUDE.md / AGENTS.md / README.md, and any docs//.md or openspec//.md. Covers two concerns: (1)…

hmj1026/dhpk

Agent

Use after a structural, route, setup, distribution, or configuration change can make user or agent docs stale. Research the live checkout and update documentation with evidence. Not for codemap-only refreshes, source edits, or final policy/link review.

hmj1026/dhpk

Agent

When the user asks how to use a library, framework, or API or needs up-to-date code examples, use Context7 MCP to fetch current documentation and return answers with examples. MUST BE USED when user asks "how to use X library/API" or requests current/up-to-date docs.

hmj1026/dhpk

Agent

End-to-end test specialist. Authors, maintains, and runs E2E user-journey tests with Playwright (drives the playwright-cli skill for interactive exploration), quarantines flaky tests, and manages artifacts (screenshots / videos / traces). Use PROACTIVELY when the user asks to write, run, or stabilize E2E tests for…

hmj1026/dhpk

Agent

Write-capable mechanical implementer. Use for boilerplate implementation, test scaffolds, rename sweeps, or applying an already-approved plan/fix-spec — dispatched per the Implementation dispatch table when the task has a clear, precise spec. Accepts a task spec (target files + exact change intent + verification…

hmj1026/dhpk

Agent

Frontend reviewer for legacy or progressive JS/TS bundles. Use PROACTIVELY after editing any frontend-tier .{js,ts,jsx,tsx,vue,svelte} file, or after touching blocks inside server-side template files (PHP / ERB / Twig / Razor — the hook can't detect view-embedded script edits, so AI judgment backstops here). Audits…

hmj1026/dhpk

Agent

Deterministic harness trim/dedupe/validate driven by the harness-revise skill and G1-G13 gap taxonomy. Use when running /harness-revise, or when the user explicitly asks to trim/dedupe/validate .claude/. For broader reliability/cost/throughput scoring, use /harness-govern (its conform step applies the official…

hmj1026/dhpk

Agent

Database migration safety specialist. Reviews schema-migration files for up/down symmetry, idempotency, FK / index naming collision across multi-tenant deploy footprints, large-ALTER strategy on high-volume tables, engine/charset explicitness, and rollback executability. Sentinel-driven (.pending-migration-review).…

hmj1026/dhpk

Agent

Performance specialist. Owns latency / index usage / query count for relational data-access (N+1, EXPLAIN, index coverage, query-count regressions) AND — when a frontend or native trap sheet is active — client/runtime performance (bundle size, render / re-render, memory leaks, Web Vitals). Use proactively on…

hmj1026/dhpk

Agent

Opt-in /dhpk:do --plan consultant: critique or sketch an implementation plan, then perform a manually requested warm or cold diff review. Return a verdict-first, END-terminated result under the mode's token cap; bounded read-only discovery uses at most 2 Explore children and 12 planner reads.