# Duntak1 Agentworkflow Agent Workflow > agent-workflow（运行时精简版） - Skill: `tomevault-io/duntak1-agentworkflow-agent-workflow` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/duntak1-agentworkflow-agent-workflow` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/duntak1-agentworkflow-agent-workflow/raw - Safety review: pending (external: skill-scanner PASS, skillspector CAUTION) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/duntak1-agentworkflow-agent-workflow --- # agent-workflow（运行时精简版） **Truth:** [package/INVOCATION.md](package/INVOCATION.md) after install. **Reference:** [reference.md](reference.md). **Manual:** [package/AGENTWORKFLOW_MANUAL.html](package/AGENTWORKFLOW_MANUAL.html) for humans; do not load it into model context unless the user asks. ## Runtime Rule This skill is a **workflow gate**, not a documentation bundle. Keep context small: - Read only this `SKILL.md` at skill activation. - Read `package/INVOCATION.md` only when a step is ambiguous or a command is needed. - Read `reference.md` only for command lookup. - Never read the HTML manual as AI context by default. - Prefer CLI summaries (`aw status`, `aw next`, `aw task brief`, `aw context plan`, `aw sync inbox`) over opening full ledgers. ## Updating aw When the user says `更新 aw`, `update aw`, `升级 aw`, or asks to refresh an already installed skill, run the project-local updater: ```bash ./scripts/aw upgrade --from-github --adapters ``` `./scripts/aw update --from-github --adapters` is the same path. This fetches `https://github.com/duntak1/agentworkflow.git`, removes the old local skill install, installs the fresh skill, and replaces the current project's `agent-workflow/` package and `scripts/` while preserving business `docs/`, `reference/`, and workflow state. Use `--repo ` or `--ref ` only when the engineer explicitly asks for a fork, branch, or release tag. ## Startup When the user says `启动 aw`, `@aw`, `aw start`, or equivalent: 1. Declare AgentWorkflow active. 2. Ask the role before any scan or coding: - `1=产品` - `2=前端` - `3=后端` - `4=全栈` 3. After role confirmation, run/present `aw project scan`, summarize `docs/PROJECT_SCAN.md`, and ask the engineer to confirm `new` vs `existing`. 4. Ask sync-center decision and record it with `aw config init --sync-center 1|2|3`. 5. Ask code hosting provider and build target. `全栈` means frontend/backend code is in one repository by default, build target `3=fullstack`; sync center is optional unless the engineer confirms split repos, different computers, or PM-managed collaboration. ## Token Budget Default budgets per AT-T: | Context | Budget | |---------|--------| | Business files | 6 files | | Symbols | 12 symbols | | Precise searches | 3 searches | | Handoff read | latest summary only, not whole history | | Sync inbox read | relevant peer/event only, not all snapshots | | Reference docs | manifest + explicitly cited files only | Do not exceed a budget without saying why and asking the engineer to confirm the expansion. Never do an aimless full-repo scan. Avoid generated/cache/dependency/build/log directories: `.git`, `node_modules`, `dist`, `build`, `coverage`, `.next`, `.nuxt`, `target`, `vendor`, `tmp`, `logs`. Use this lean order: ```bash ./scripts/aw status --json ./scripts/aw next ./scripts/aw task brief ./scripts/aw context plan --task ./scripts/aw context gate --task ``` Read only files listed in `docs/context/tasks/CTX-.md`. Prefer CodeGraph / `aw code-map` / `aw context` symbol, caller/callee, impact, and affected-test queries. If CodeGraph is unavailable, use `CODE_MAP`, `CODE_CONTEXT_INDEX`, `FILE_INDEX`, and precise `rg`. Code map is automatic by default. `aw context plan`, `aw task start`, `aw task complete`, `aw watch index`, and `aw gate pre-commit` refresh/check `docs/context/CODE_MAP.md` unless `AW_CODE_MAP_AUTO=0` is set for an explicit exception. Before coding in existing or large projects, query the map instead of scanning the repo: ```bash ./scripts/aw code-map build ./scripts/aw code-map query "" ./scripts/aw code-map impact "" ``` `CODE_MAP.md` is a locator index, not permission to read full files. Convert findings into the task Context Plan and wait for confirmation before opening file contents. ## Required Gates | Stage | Hard rule | |-------|-----------| | DSL | DSL status not `已审` → no business code | | Plan | Plan status not `可执行` → no business code | | Confirm | Run `aw confirm ` before AT-T execution | | Task start | `aw next` → `aw task brief ` → ask scope / acceptance / non-goals → `aw task confirm "已确认：范围=...；验收=...；非目标=..."` | | Context | `aw context plan --task ` → engineer reviews allowed files → `aw context gate --task ` | | Coding | Only after `aw task start ` may you request/use `aw paste task` or edit business code | | Completion | `aw task complete ` verifies; failed verification logs Bug and keeps task open | | Checkpoint | Completed AT-T needs Git decision, compact/handoff evidence, and `docs/FILE_INDEX.md` refresh when business files changed | When the user says “执行研发任务”, “开始开发”, “做下一个任务”, or equivalent, do **not** start coding. Start with `aw next` and `aw task brief`. ## Context Continuity | Lane | Use for | Do not use for | |------|---------|----------------| | Handoff (`docs/handoff/PROJECT_HANDOFF.md`) | Current goal, progress, blockers, next 1-3 steps | Long-term reusable facts | | Memory (`docs/memory/`) | Stable decisions, preferences, reusable procedures, recurring risks, summarized chat episodes | Raw chat logs or transient task status | Use `aw compact "focus" --write --snapshot` before Codex context compaction, new chats, model switches, long pauses, or after a large requirement/AT-T batch. Add `--memory-summary` only for reusable decisions or follow-ups. Do not store secrets. New session lean resume: ```bash ./scripts/aw handoff --check ./scripts/aw memory inject ./scripts/aw status --json ./scripts/aw next ``` Read handoff/memory selectively. Do not read `ENGINEERING_INDEX.md` into AI context; it is for human engineers. ## DSL / Plan / PM - DSL suite: `aw dsl suite "title"` creates multi-file DSL coverage for requirements, pages, interactions, events, boundaries, and acceptance. - `aw dsl review --write` creates the engineer review package before `aw approve dsl`. - `aw approve dsl --plan` may guide Plan generation; PM/three-side projects should use `aw pm plan --write` before dispatch. - `aw plan change`, `aw plan task-add`, and `aw task split` handle development-time changes without bloating the active task. - Product-led work uses `aw pm start` and `aw pm init `; PM owns shared references, Pencil intake, shared DSL/Plan, dispatch, dashboard, and lifecycle gates. ## Cross-Project Sync For split frontend/backend repositories: - Configure sync center with `aw sync init --project ... --agent ... --role ...`. - Start dependent work with `aw sync pull` + `aw sync gate --task `. - Read only `aw sync inbox --from ` summaries and relevant event/contract files. - `aw sync pull` imports read-only inbox files; never treat inbox as local truth until adopted through REQ / Bug / Plan / Handoff. - API changes use `aw contract change`, `aw contract test`, and `aw contract gate`. - Long-lived worker identities use `aw agents register`; common roles can be bootstrapped with `aw agents register --defaults`. - Multi-agent implementation uses `aw agents claim`, `aw agents heartbeat`, and `aw agents gate --strict`. ## Engineering Principles | Principle | Meaning | |-----------|---------| | Think before coding | Clarify authority, scope, acceptance, and non-goals | | Simplicity first | Keep changes small and directly tied to the AT-T | | Mature solutions first | Prefer proven libraries, SDKs, framework features, and official patterns | | Surgical changes | Avoid unrelated refactors and noisy diffs | | Goal-driven execution | Verify against real project commands and recorded acceptance | Also maintain completeness, traceability, maintainability, and handoffability. Use `aw audit`, `aw policy`, `aw security`, `aw service-catalog`, `aw release`, `aw trace check`, `aw metrics summary`, `aw ops gate`, `aw report handoff`, `aw report release`, `aw score record`, and `aw recover ...` when the task touches risk, production, dependencies, services, release, multi-agent work, or handoff. ## File / Bug / Git Rules - Record spoken new requirements with `aw req new`; record development-time changes with `aw req change`, then update DSL/Plan/ATOMIC before continuing. - Every bug or suspected bug goes to `aw bug add`. - Update `docs/FILE_INDEX.md` when adding/deleting/renaming business files. - Ask before Git commit/push. Use `aw commit --task --changelog "..."`; do not commit automatically. - Comment AI-written code for non-obvious business rules, boundaries, tradeoffs, side effects, and temporary decisions; avoid obvious comments. ## Pencil Pencil design files are formal inputs. Store `.pen` sources under `global/references/design/pencil/source/`, exports under `exports/`, screenshots under `screenshots/`, and changes in `DESIGN_CHANGELOG.md`. Do not parse encrypted `.pen` files as plain text; use Pencil tooling or exported artifacts. --- > Source: [duntak1/agentworkflow](https://github.com/duntak1/agentworkflow) — distributed by [TomeVault](https://tomevault.io).