arayaroma/ether/architect-job

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use when running the recurring "review module structure and dependency direction" routine (invoked via /loop or CronCreate against this skill, or on direct request "run the architect job"). Drains pending `architect` jobs from ether-pro's job store, reviews the target repo's module boundaries/dependency direction/coupling (not line-level bugs — that's repo-review-job), proposes structural fixes when the job/action has earned autonomy, and records the outcome so the autonomy state machine can promote or demote future runs.

Install

npx vskill@latest install arayaroma/ether/architect-job

SKILL.md

Architect Job

Same job-router + crystallization mechanism as repo-review-job (docs/design/2026-08-01-job-router-crystallization-design.md), applied to an architect job type. target is a local repo path. Run ether job ... only from ether-pro's own directory, never from inside the target repo — same hard rule as repo-review-job; other ether subcommands must never run with the target as cwd (contamination risk, see that skill's "Hard rule" paragraph — applies identically here).

Distinct from repo-review-job: that skill's code-review pass finds line-level bugs/simplification/efficiency issues in a diff. This job looks at structure — module boundaries, dependency direction, layering violations, circular deps, coupling that will make future changes harder — independent of whether any single line has a bug. Run both on a target; they don't overlap and don't replace each other. A structural finding here is not itself a bug report — it's "this will hurt later," which is why its write action defaults to a stricter confirm posture (see step 4).

Workflow

Autonomy-tier check and record-run steps below follow the shared protocol in `skills/_shared/job-constitution.md` — read that first if unfamiliar; only this job type's specifics are called out here.

  1. Drain pending jobs. From ~/wrkspc/ether-pro: `ether job pending

--type architect`. If none, report "no pending architect jobs" and stop.

  1. Review (read action).
  • ether job should-confirm --type architect --actionType read.
  • If confirm, tell the user you're about to review <target>'s

structure and proceed (first runs of a new target always confirm).

  • If autonomous, proceed without asking.
  • Map the target's module/package graph (read its directory layout,

import/require graph, or use codegraph_explore if .codegraph/ exists at the target root — prefer it over manual grep for dependency edges). Look for: dependency-direction violations (a lower layer importing a higher one), circular dependencies, a module doing two unrelated jobs (mixed responsibility), duplicated structure that should be one shared boundary. Skip anything that's a line-level bug, not structural — that belongs to repo-review-job, don't duplicate it here.

  • Write findings to <target>/.ether/qa/jobs/<jobId>/report.md (Write

tool, inside target repo). Include: type, target, agent, action, outcome, confirmed-by-human, date, each finding as (boundary/module pair, what's wrong, why it matters later, suggested fix or (none obvious)).

  • Record: `ether job record-run --jobId <id> --agent architect-job

--actionType read --outcome success|failure from ether-pro's directory. failure` = the review itself couldn't run (repo missing, tool error) — not "found structural issues," that's success.

  1. If a finding has a bounded fix, propose a write action.
  • ether job should-confirm --type architect --actionType write.
  • Structural changes touch more surface than a line fix and are easy to

get subtly wrong (e.g. "fix" a layering violation by introducing a new circular dep elsewhere) — treat confirm as the expected default even once other job types on the same target have promoted past it; don't assume architect's write tier tracks repo-review's.

  • If confirm: present the finding, the proposed restructuring, and

what it touches (files/modules), wait for explicit approval.

  • If autonomous (only after real promotion history specific to this

job type): apply via a bounded cavecrew-builder delegation scoped to the exact files identified, or do it directly if it's a single-file move/rename. Run the target's full test suite (not just the touched module's) before recording success — structural changes are exactly the kind of edit that breaks something two layers away.

  • Append the write outcome (what was applied/proposed, verification

result) to the same report.md.

  • Record the write outcome the same way, with --safetyViolation for

anything that broke the build/tests or touched files outside the identified scope.

  • If a human approved a confirm-tier run, add --confirmedByHuman.
  1. Report per job: target repo, structural findings summary, fix

applied/proposed, tier used, report file path. End with ether job status --type architect and call out any tier change explicitly.

Adding a job

From ~/wrkspc/ether-pro: ether job add --type architect --target /absolute/path/to/repo [--params '{"scope":"full"}']

Scheduling

Prefer /loop <interval> /architect-job for a live session, with a longer interval than repo-review-job — structure drifts slower than line-level issues, frequent runs mostly re-find the same findings. CronCreate works for unattended cadence (session-only, expires after 7 days) but expect most runs to land on confirm given the stricter write posture in step 3.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/arayaroma/ether/architect-job/badge)](https://verified-skill.com/skills/arayaroma/ether/architect-job)

Something wrong with this skill or its page? Tell us.