Skip to content

lennoxsaint/Threadify-workflows / plugins/claude/skills/threadify-get-set-up

threadify-get-set-up

Verified

Set up Threadify from scratch: install workflows, sign up, connect your account and agent, find your existing knowledge, learn your voice, prepare your first week and verify an ongoing routine. Resume interrupted setup without starting over.

references/setup-guide.md

View on GitHub
# Customer setup engine

Start with **Get Set Up**. Do not run the entire catalog or ask the customer to select workflows.

## Private state and recovery

Use a new dedicated owner-only directory under `~/.threadify-workflows/customers/<session>/setup`; creator records belong in its sibling `creator` directory. Keep profile content out of command arguments, logs and release assets. The engine reuses the creator store's atomic writes, checksum, revision check and writer lock. A changed revision means read status again, not overwrite. Inspect/recover only a verified dead writer lock using the existing store recovery functions.

Public interface: `setup start`, `setup resume`, `setup status`, `setup verify`. Helpers: `setup discover`, `setup summary-hash`. Input is JSON on stdin. `start`: client, pinned version and canonical install receipt. `resume`: event plus phase-specific data and `--revision` from the last result. Existing sessions cannot change account, version or client silently.

Events in order: runtime, connection, discovery, summary, brain, offer, vault, voice, permission, week, routine-configured, routine. Later: begin-run, finish-run, pause, unpause. Evidence has `kind: observed`, `reference`, `observed_at`, explicit `account` where relevant. This is host-normalized evidence, not an independent verifier. Never use simulated evidence in customer records.

Summary facts: audience, voice, comma-separated topics, outcome, timezone, posts_per_day (1–5), times (one distinct HH:MM per post). Each source carries reference and supports. The customer confirms the exact `summary-hash`; keep their confirmation reference. A changed summary preserves history, invalidates downstream phases and pauses the routine until reconfirmed.

Brain evidence: summary_hash, processing_complete, retrieval_verified, durable_voice_verified. Vault: selection_approved, with the saved items confirmed as readback_verified. Offer: confirmed_none or readback_verified with offer_ids. Voice: sample_approved, sample_hash. All include matching summary_hash.

Week evidence: summary_hash, timezone, days:7, actual draft_count, content_verified, scheduled_count, calendar_readback_verified when scheduled. Reviewed drafts awaiting approval are shown as prepared, not scheduled. Automatic mode requires the prepared first week to have matching schedule readback.

Routine evidence: selected client/account, permission_id returned from the permission event, job_id, persisted, real_run_verified, run_id, next_run, device_must_stay_on, pause_command and runner (`native` or `os-scheduler`). Record routine-configured before the first invocation. Reconcile that invocation with finish-run, then record routine with local_run_id matching the reconciled run. Record the real provider/configuration reference, not a guessed job ID.

begin-run preflight: account, timezone, checked_at within five minutes, connection_ok, entitled, facts_current, allowance_ok, calendar_read, occupied_instants. A pending run blocks another; finish it only with verified/confirmed_no_actions/blocked evidence and its exact run_id. Unknown remains pending. Pause prevents future work but does not cancel already scheduled posts; explain that difference.

## Terms

- Setup ready: all required observations recorded; fresh provider checks still support a live claim.
- Brain ready: approved sources processed and retrievable, durable voice instructions verified.
- Standing scheduling permission: customer's revocable, account/topic/time-bound consent; not a global reply/publish switch.
- Routine verified: persistent configuration read back and one real agent invocation observed.
- First week prepared: actual drafts exist for seven days; approval and scheduling are separate states.

## Package migration

This replaces the separate v0.1 customer ZIP's one-result route. Preserve its customer files and receipts. Do not silently migrate confirmed facts into a different account. Read those files as customer-provided sources, ask to confirm the new summary, and create a separate v1 setup directory. Existing edits and old receipts remain intact.

Use the canonical account UUID in setup state and creator cards. Display the verified handle alongside it; a handle is not a substitute for the creator account_id.

## Operator proof and orchestration surfaces

Completion must produce one compact operator proof for the customer: verified release/runtime, verified account handle, Brain/retrieval/voice state, first-week draft and scheduled counts, outstanding review, routine state, next run, pause control and honest blockers. The proof is a view of existing receipts, never a new source of truth. It must not expose the canonical account UUID, private source paths, draft bodies or credentials in a public recording.

OpenAI Dots may orchestrate a supported runtime, but `dot` is not a native installation target in this release. A Dot must delegate filesystem work to an approved Codex or other supported target on a connected computer, then read the exact receipt back. Repository browsing, a task request or a claimed delegation is not installation proof.

For the first week, `draft_count` counts new drafts and `existing_occupied_count` counts preserved scheduled slots in the agreed seven-day plan. Their sum must equal the planned slots. Existing slots require `existing_slots_readback_verified`; `scheduled_count` counts only the new drafts actually scheduled. Never create extra drafts to replace occupied slots. Review mode may leave new drafts awaiting approval.

Before the initial week, `begin-batch` acquires the same persisted pending record as recurring `begin-run`, without requiring a configured routine. Both need current account/entitlement/facts/quota/calendar preflight. Standing scheduling binds that run ID; after finish-run, old approvals cannot execute. Unknown outcomes stay pending. Calendar preflight must include all scheduled posts in the seven-day range; posts at other times count toward daily frequency.

Use the next seven complete local posting days: start today only when the first agreed daily time is still at least five minutes ahead; otherwise start tomorrow. Recurring checks retain existing scheduled posts and do not backfill missed times today. Show these seven actual dates when confirming the plan.

For a recurring `finish-run`, include `job_id` and `runner_run_id` from the actual host invocation. `routine` must reference that exact reconciled job/run, local run ID and current permission. Reconfiguring a job clears previous recurring verification.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall