manji-0

Make a git worktree usable for dagayn MCP — inherit the main checkout's graph and MCP config, then catch up the branch diff.

Systematically debug issues using graph-powered code navigation

Navigate and understand codebase structure using the knowledge graph

Build or update the code review knowledge graph. Run this first to initialize, or let hooks keep it updated automatically.

Configure, build, and verify dagayn embeddings and hybrid semantic search without losing FTS fallback behavior.

Author Markdown documents (READMEs, design docs, RFCs) so dagayn extracts correct dependency edges. Four-stage flow — outline & sort, draft & verify, polish, summary.

Evaluate architecture signals through the unified dagayn dispatcher

Register repositories, maintain multi-repo graph freshness, and search across repos with dagayn.

Formal-verify RDRA-ish models with TLA+/TLC. Use when the user asks about export --kind tla, verify --backend tlc, temporal property / always / eventually / leads_to, after.assert, when().none/has, Int/Money/Decimal/now arithmetic, multi-entity…

Review RDRA DSL files for syntax errors, semantic inconsistencies, missing relationships, staged refinement gaps, business flow coverage, requirement/NFR traceability, screen field mappings, API contracts, exports, ADR links, and lint readiness

Write RDRA DSL files from requirements using correct syntax and file structure, including staged abstract-to-concrete refinement, business flows, requirements/NFRs, conceptual models, screen fields, API contracts, data modeling annotations, ADR links,…

Choose the right RDRA-ish BUC or whole-model analysis path for refinement readiness, coverage gaps, business flow, requirement/NFR traceability, screen field mappings, API contracts, access, system boundaries, exports, and state patterns

Create a new BUC file from a requirements description, using staged refinement from BUC skeleton and business flow to conceptual/data, UI/API contracts, lifecycle, rules, NFRs, and traceability

Update an existing BUC by adding or modifying use cases, business flows, requirements, screens, fields, API contracts, concepts, NFRs, events, entities, ADR links, and predicates while preserving staged refinement

Choose the right RDRA-ish diagram, CSV review, diff, or export view for the modeling situation, then load focused references for commands and interpretation

Execute track TODOs with jj-task workspaces and agent-skill-jj ($jj) for commits. Read track status --json, start jj-task workspace, implement, use $jj for jj operations, track scrap/done for task state. Use when workflow.phase is sync_required or execute.

Advanced track workflows with agent-skill-jj — archive, handoff, multi-repo tasks, hotfixes. Assumes jj-task and $jj skill for all JJ/PR operations. Use when workflow.phase is task_complete or for cross-repo patterns.

Entry point for track CLI task management. Use when the user mentions track, todos, or task context. Assumes agent-skill-jj ($jj) for all JJ commits and PR work. Reads workflow.phase and routes to track-task-setup, track-task-execute, or track-advanced.

Set up a track task before implementation — create task, description, register repos, add TODOs and links. Assumes agent-skill-jj for jj-task repo init and workspace creation. Use when workflow.phase is setup.

DEPRECATED — use split track skills instead. Redirects to track (router), track-task-setup, track-task-execute, and track-advanced.

Robust server-side Python domain modeling with Pydantic v2 discriminated unions, Python 3.12+, uv-managed projects, frozen state models, pure state transition functions, boundary validation, explicit domain errors, PII redaction, persistence/event…

Adversarial review of server-side Python domain code for Kamae principles: Pydantic v2 discriminated unions, frozen state models, pure state transitions, explicit domain errors, validated boundaries, PII redaction, persistence/event consistency, logging and…

日本語で読みやすいソフトウェアデザインドキュメントを新規作成、改訂、構成整理、レビューする。 設計提案、技術設計書、Design Doc、RFCを日本語で執筆または改善するときに使用する。