# Wheelie docs

Agent docs index: /llms.txt

Wheelie is the public Development OS command surface for coding agents and technical founders. The installed `wheelie` CLI is the preferred path for agent automation, local harnesses, scripts, source/candidate/change operations, validation, submit, and watch. Wheelie over SSH is the zero-install human onboarding/demo/recovery path and a constrained read/status fallback. Public launch docs stay narrow: they guide a user through a one-afternoon loop from installation to an existing-checkout or public-repo task, BYO-harness support probes, fast context, a work item, a reviewable diff, and adapter-gated model/source/harness/validation/change/checkpoint commands with truthful support states.

## Pages

- [Human docs landing page](./index.html)
- [Wheelie changelog](/changelog)
- [Quickstart](./quickstart.md)
- [CLI support map](./cli.md)
- [Command taxonomy](./command-taxonomy.md)
- [Wait guard](./wait-guard.md)
- [Platform capabilities](./platform-capabilities.md)
- [Wheelie over SSH](./ssh.md)
- [Private and non-public GitHub repositories](./private-repos.md)
- FAQ (not yet published)
- [Support levels](./support-levels.md)
- [Launch support and friendly error states](./support.md)
- [BYOK and secrets guide](./byok-secrets.md)
- [Private-preview billing](./billing.md)
- [Spend support states](./spend-support-states.md)
- [Paid validation quote support-state reference](./paid-proof-quote.md)
- [Community package, example, and demo contributions](./community-contributions.md)
- [Public command manifest](/schemas/wheelie/public-command-manifest.json)
- [Feature-discovery catalog](/services/wheelie/feature-discovery.json)
- [Public discovery manifest](/services/wheelie/public-discovery-manifest.json)
- [Community contribution policy](/services/wheelie/community-contribution-policy.json)
- [Public changelog task projection](/services/wheelie/changelog.json)
- [Feature-discovery API schema](/schemas/wheelie/feature-discovery.openapi.json)
- [Paid validation quote OpenAPI fixture](/schemas/wheelie/paid-proof-quote.openapi.json)
- [Wheelie public brand asset manifest](/assets/wheelie/brand/manifest.json)

## Static brand assets

The human docs landing page uses a Wheelie-specific hero/Open Graph image and round favicon/avatar assets. These images are identity and social-preview metadata only; agents should keep using this Markdown page, `/llms.txt`, and the JSON schemas for comprehension.

Semantic asset mapping lives at `/assets/wheelie/brand/manifest.json` so asset swaps can replace selected files without hunting through docs copy.

## What is launch-supported?

The launch-preview path is install/update/doctor, auth/status probes, harness-support probes (clean public stable reports `requires_adapter` until package support or a project adapter provides projection install/status), BYO-harness use from an existing checkout, Wheelie-over-SSH pairing/TUI for zero-install human access plus command-mode read/status probes when the account and session broker report support, startup context packs, GitHub/model connection probes, public GitHub source status, work items, adapter-backed working-copy and change commands, validation only after an authorized working copy/project catalog is available, checkpoint commands when a store is attached, resource status, cleanup dry-runs, and low-noise feature-discovery hints backed by typed usage/context signals. The public growth wedge is one issue → one working-copy handoff → one reviewable diff → focused validation evidence, with submit/watch/checkpoint treated as support-state branches. For private and other non-public GitHub repositories, read [Private and non-public GitHub repositories](./private-repos.md): that path is launch-deferred until the selected repo returns a mediated source-provider adapter or allowlist receipt.

The full clean public onboarding-to-PR loop, hosted model/workspace/agent-run/validation/PR path from a clean public install, stable version/update metadata plus logout/revoke/status coherence, strong workspace/context durability, public named-workspace lifecycle/durability timing, broad performance/capacity/pricing/prod-parity numbers, and provider-backed submit/watch are not general green claims. Strong workspace/context durability is explicitly out of the current launch scope: source snapshots are authoritative for source truth and project-memory summaries are authoritative only for remembered same-project context, while VM/session continuity across reboot, stop/start, base-image update, host loss, reconnect, or partial-run recovery remains preview/adapter-gated until commands emit live named-workspace receipts. Do not publish warm-resume/cold-boot/first-setup p50/p95/p99, same-session reconnect, preview/share URL continuity, process/log continuity, cleanup SLO, host-loss, or dirty partial-run auto-recovery claims until a public/customer-facing named-workspace cohort supports them. Broad performance/capacity/pricing/prod-parity claims stay outside the next public launch scope; only narrow, bounded staging proof-wave facts may be cited with their stated limits. These claims must stay allowlist-, adapter-, preview-, or evidence-gated unless a command reports live support for the current account/project.

Billing/payment is manual invoice or contract only for the private preview. There is no self-serve checkout, card charge, billing portal, automatic overage, or money movement in the Wheelie launch path; plan changes, cancellations, credits, refunds, and overage approvals route through support unless a scoped support receipt says otherwise. Use [Spend support states](./spend-support-states.md) for public-safe spend copy.

Launch support starts from [typed friendly error states](./support.md), not raw terminal archaeology. Share `reason`, `support_state`, `next_action`, operation/evidence refs, and redacted support-bundle fields with support@continua.ai or the private-preview Discord channel named in your invite packet.

Surfaces outside that path may still be visible. They must report `support_state`, `reason`, and `next_action` so a person or agent can tell whether the surface is native, preview, requires an adapter, fixture-only, allowlist-gated, or unsupported. Fixture-only validation quote discovery and spend-capability surfaces must also report `money_movement_enabled=false` unless a scoped support receipt says otherwise.
