Epic: grand tunnel restructuring and CLI UX (#640)

* feat: restructure tunnel handling around provider-based service model" -m "Introduce tunnel service/registry/provider architecture and move Cloudflare handling behind provider adapter." -m "Add canonical tunnel modes (quick, managed-remote, managed-local) with legacy named/try-cf-tunnel compatibility mapping." -m "Add managed-local config-path support, normalized API response fields, tunnel-focused tests, and shell aliases for tunnel test workflows.

* feat(tunnels): harden managed startup and decouple runtime APIs

Improve managed Cloudflare startup reliability with explicit config validation, YAML diagnostics, and readiness detection based on process output instead of fixed delay assumptions.

Refactor server tunnel lifecycle around provider-aware runtime state and API responses while keeping legacy Cloudflare token endpoint compatibility, and add coverage for unsupported mode validation plus managed-local startup cases.

* feat: remove named tunnel mode and standardize managed modes

Replace named tunnel terminology with managed-remote and managed-local across API, server state, and UI settings without legacy aliases.

Add provider capability discovery endpoint and descriptor-based mode validation, including explicit mode_unsupported errors for removed mode values.

* feat(tunnels): finalize provider-aware tunnel UX and managed-local safety

Restructure tunnel settings with provider selection, mode chips, persisted managed-local config path, and clearer session badges while preserving existing tunnel flows.

Add legacy named-data migration, provider discovery CLI, and user-friendly managed-local config validation/error messaging with updated API/CLI/server tests.

* Add provider icon to tunnel settings

* Add control+C to stop tunnel

* feat(cli): add tunnel lifecycle profiles and preserve preset naming

Replace legacy tunnel flags with explicit tunnel lifecycle commands, daemon-by-default startup, and file-backed log tailing so tunnel operations are predictable and provider-agnostic.

Add managed-remote profile storage/migration for start-by-name workflows and propagate preset summaries to settings so user-defined profile names are preserved instead of falling back to Default.

* feat: improve tunnel CLI safety and startup UX

Add interactive TTL support and per-start TTL overrides for tunnel start
Strengthen port safety and instance validation with clearer startup and error guidance
Refine tunnel doctor and CLI output formatting for clearer, less noisy diagnostics

* feat: add TTL support, safety gates, and polished tunnel CLI output

* fix: harden tunnel doctor checks and CLI port handling

* fix: improve tunnel CLI diagnostics and profile output

* fix: streamline tunnel profile UX and doctor diagnostics

* fix: clarify tunnel replacement behavior across CLI and UI

* Upd docs

* docs: add mandatory clack CLI skill guidance. cleanup

* fix: standardize tunnel CLI mode parity and prompt UX

* fix: align CLI quiet and JSON output behavior

* feat/web-serve: in-progress animation

* fix: tunnel doctor managed remote validation

* Fix: security tightening

* fix: instance restart ux

* fix: tighten tunnel doctor input handling and CLI port/prompt validation

* chore: remove tunnel test suites per owner request

---------

Signed-off-by: Iuliia Ivashko <yulia.ivashko@gmail.com>

AGENTS.md

@@ -119,6 +119,50 @@ All scripts are in `package.json`.
 - No new deps unless asked.
 - Never add secrets (`.env`, keys) or log sensitive data.
 
+## CLI Parity and Safety Policy (MANDATORY)
+
+### Principle: policy-first, UX-second
+
+All safety and correctness rules MUST be enforced in core command logic, independent of output mode.
+
+Interactive/pretty UX (`@clack/prompts`) is a presentation layer only.
+It must never be the only place where validation or restriction is enforced.
+
+### Required parity across modes
+
+The same functional outcome and safety gates MUST hold for all execution modes:
+
+- Interactive TTY (full Clack UX)
+- Non-interactive shells (piped/stdin-less automation)
+- `--quiet`
+- `--json`
+- Fully pre-specified flags (no prompts)
+
+In all modes, invalid operations MUST fail with non-zero exit code and deterministic error semantics.
+
+### Non-negotiable rule
+
+Do not rely on prompts to enforce policy.
+
+- Prompts MAY help users choose valid inputs.
+- Core validators MUST run even when prompts are unavailable or skipped.
+- `--quiet` suppresses non-essential output only; it does not weaken validation.
+- `--json` changes output shape only; it does not weaken validation.
+
+Detailed Clack UX patterns (primitives, prompt gating, and implementation checklist)
+are defined in the `clack-cli-patterns` skill and should not be duplicated here.
+
+## Clack CLI Skill (MANDATORY for terminal CLI work)
+
+When working on terminal CLI commands, prompts, or output formatting, agents **MUST** study the Clack CLI skill first.
+
+**Before starting terminal CLI work:**
+```
+skill({ name: "clack-cli-patterns" })
+```
+
+Scope: terminal CLI only (for example `packages/web/bin/*`). Do not apply this requirement to VS Code or web UI work.
+
 ## Theme System (MANDATORY for UI work)
 
 When working on any UI components, styling, or visual changes, agents **MUST** study the theme system skill first.