From auxly
Execute an accepted implementation plan inside the live, Auxly-branded multi-stage Console — phases, slices, and a progress bar that update in real time as Claude works through the plan, plus an Agents & Models panel showing which agents/subagents are active vs idle, and a Checks panel for build/test/lint gates. Blockers that need a human decision appear as RED notifications requiring action (the user types a resolution in the UI and execution resumes); warnings appear AMBER and dismissable. Renders into the same shared Auxly Console tab as plan/verify/review (no new tabs). Use whenever the user accepts a plan and wants to execute with live monitoring, or says "run the plan with a dashboard", "execute and show progress", or "track execution". Pairs with /auxly-llm-council (feed its final-plan.md) but also works standalone with any final-plan.md or plan.json.
How this skill is triggered — by the user, by Claude, or both
Slash command
/auxly:auxly-executeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Execute an accepted plan and give the user a live window into it: which phase/slice is active,
Execute an accepted plan and give the user a live window into it: which phase/slice is active, progress, the agents at work, build/test checks, and any blockers or warnings — all in the shared Auxly Console (one server, one browser tab, stage tabs: Plan ▸ Execute ▸ Verify ▸ Review).
Execution model: Claude is the executor. Claude works the plan with its own tools and pushes
progress events to the console via console.py. Nothing runs shell commands unattended.
All commands use the vendored console CLI: python3 scripts/_console/console.py <verb>.
python3 scripts/_console/console.py start --plan /path/to/final-plan.md --title "DB migration"
final-plan.md (### Phase N / #### Task N.M) or a plan.json
({ "title": "...", "phases": [ { "id":"1","name":"...","slices":["...", ...] } ] })../auxly-console/current-session.json. Tell the user the console is open.ensure instead of start to attach without a new tab.python3 scripts/_console/console.py phase 1 --status active
python3 scripts/_console/console.py slice 1.1 --status running
# ...do the real work for 1.1 with your tools...
python3 scripts/_console/console.py slice 1.1 --status done --note "migrated 59 tables"
python3 scripts/_console/console.py phase 1 --status done
python3 scripts/_console/console.py log "Phase 1 complete"
python3 scripts/_console/console.py agent lead --name "Claude (lead)" --kind claude --model opus --role executor --status active --current "slice 1.1"
python3 scripts/_console/console.py agent sub-tests --name "test-writer" --kind claude --model sonnet --role subagent --status active --current "slice 2.2"
python3 scripts/_console/console.py agent sub-tests --status idle # when it finishes
Optionally record token usage for the live meter (see /auxly-meter):
console.py meter --agent lead --model opus --tokens-in 1200 --tokens-out 3400.
Run build/test/lint as checks shown in the Execute tab's Checks panel:
python3 scripts/_console/console.py check tests --status running --name "pytest"
python3 scripts/_console/console.py check tests --status pass --output "143 passed"
python3 scripts/_console/console.py check lint --status fail --output "2 errors in app.py"
(For a dedicated, deeper verify pass use /auxly-verify; it renders into the Verify tab.)
python3 scripts/_console/console.py warning --subject "PG14 source vs PG16 target" --detail "Intentional; just noting."
python3 scripts/_console/console.py blocker --id db-pass --slice 1.2 --subject "DB password required" --detail "Need prod Postgres password."
python3 scripts/_console/console.py wait-blocker db-pass --timeout 120
wait-blocker returns {"resolved": true, "resolution": "..."} when the user clicks Resolve &
resume in the UI. If resolved:false, call it again (bounded). Raising a blocker turns the run
status red/blocked until resolved.
The Execute tab shows [Run checks ▶] and [Review ▶] buttons. When the user clicks one, it enqueues an intent — drain it and act:
python3 scripts/_console/console.py poll # -> {"actions":[{"type":"intent","name":"start_review"}]}
If you see start_review, invoke /auxly-review; start_verify, run /auxly-verify. Finish with:
python3 scripts/_console/console.py done # marks the run complete (green)
# or: console.py set --status failed
Keep the session responsive until execution finishes so the user can watch and resolve blockers.
scripts/_console/console.py| Verb | Purpose |
|---|---|
start --plan <md|json> [--title T] / ensure | open (or attach to) the console |
phase <id> --status <s> / slice <id> --status <s> [--note N] | execute progress |
check <id> --status pass|fail|running [--output O] | build/test/lint gate |
agent <id> [--name --kind --model --role --status --current] | agent/subagent roster |
meter --agent A --model M --tokens-in N --tokens-out N | token meter |
warning / blocker / wait-blocker <id> / poll | notifications + intents |
activate <stage> / log <msg> / set --status <s> / done / state | misc |
./auxly-console/runs/<ts>/state.json.scripts/_console (single source in the repo at
plugins/auxly/shared/console; maintainers run sync-console.sh to update).Provides UI/UX resources: 50+ styles, color palettes, font pairings, guidelines, charts for web/mobile across React, Next.js, Vue, Svelte, Tailwind, React Native, Flutter. Aids planning, building, reviewing interfaces.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
npx claudepluginhub tzamun-arabia-it-co/auxly-skills --plugin auxly