Move to a terminal driven integration to have better interactions

CLAUDE.md

@@ -6,14 +6,20 @@ changing a subsystem.
 
 ## What this is
 
-A Nova extension that puts Claude Code inside Panic's Nova editor: a sidebar for
-the session, a live Markdown transcript, and tool approvals reviewed in the
-editor.
+A Nova extension that launches and tracks Claude Code sessions for a project.
+**Chat happens in a terminal, not in Nova.** The extension prepares launch
+commands, lists the project's sessions, and manages the account behind them.
 
-It **drives the user's own `claude` executable** as a subprocess, speaking the
-CLI's newline-delimited JSON protocol over stdio. There is no bundled copy of
-Claude Code, no Node sidecar, no npm dependencies, and no network access of its
-own.
+There is no bundled copy of Claude Code, no Node sidecar, no npm dependencies,
+and no network access of its own.
+
+> **The extension used to drive the conversation itself** — `claude --print` over
+> the stream-json protocol, rendered into a Markdown tab, with approvals, plan
+> review and question answering in Nova panels. That was removed after real use:
+> modal panels truncate long questions and lose their input when you click away.
+> The implementation is preserved in commit `89d1a06`; its transport notes and
+> the editable-proposal flow are the starting point for the IDE bridge. **Do not
+> reintroduce panel-based chat.**
 
 ## Hard constraints (do not re-derive these)
 
@@ -39,16 +45,20 @@ relative `require("./x.js")` between `Scripts/` files.
 Type definitions worth having open: `npm pack @types/nova-editor-node` (it is a
 single `index.d.ts`, and is more precise than the website).
 
-**Claude Code CLI** — see [session transport](docs/features/session-transport.md)
-for the full protocol. The three facts that cost the most to find:
+**Claude Code CLI** — the facts this extension relies on:
 
-1. `--permission-prompt-tool stdio` is **required** for permission prompts to
-   reach the host, and is absent from `claude --help`. Without it every tool
-   needing approval is silently auto-denied.
-2. One assistant message id arrives multiple times (per completed content block,
-   plus streaming deltas). They must be merged, not appended.
-3. A `can_use_tool` control request **must always be answered**, including when
-   the handler throws — the CLI blocks on it and the session hangs otherwise.
+1. Sessions are stored as `~/.claude/projects/<slug>/<session-id>.jsonl`, where
+   the slug is the project path with `/` and `.` flattened to `-`. This is the
+   CLI's own store, so terminal-started sessions appear in the sidebar too.
+2. `claude --resume <id>` resumes from any terminal; `--remote-control` enables
+   driving a session from elsewhere. Nothing extra is needed to support either —
+   a launched session is an ordinary CLI session.
+3. `claude auth status --json` gives structured account state.
+
+If you ever need to drive the CLI programmatically again, commit `89d1a06`
+documents the stream-json and control protocols in detail — including that
+`--permission-prompt-tool stdio` is required and undocumented, without which
+every permission-requiring tool is silently auto-denied.
 
 **Policy** — do not implement claude.ai OAuth in this extension. Anthropic's
 Agent SDK terms bar third-party products from offering claude.ai login. Sign-in
@@ -60,15 +70,11 @@ shells out to `claude auth login`. See [authentication](docs/features/authentica
 extension.json          manifest: sidebar, commands, config, entitlements
 Scripts/
   main.js               Controller: lifecycle, commands, wiring
-  client.js             CLI process + stream-json/control protocol
-  session.js            conversation model (entries, tools, changed files)
-  permissions.js        can_use_tool → approvals, plans, questions, proposals
-  transcript.js         Markdown tab rendering
-  sidebar.js            four TreeView providers
-  diff.js               change reconstruction + unified diff
+  launch.js             building and copying the launch command
+  sessions-store.js     reading ~/.claude/projects
+  sidebar.js            Status and Sessions TreeView providers
   auth.js               claude auth status/login/logout
   cli.js                binary resolution, environment, keychain
-  sessions-store.js     reading ~/.claude/projects for resume
   util.js               conf, throttle, Signal, runCommand, processWriter
 docs/features/          one file per capability; read before changing one
 ```
@@ -96,28 +102,29 @@ nova extension activate .    # load into Nova for development
 
 Logs (`nova.inDevMode()` output) appear in Nova's Extension Console.
 
-**Node harness.** The Nova API is stubbable, so the transport and model can be
-driven under Node against the *real* CLI. This caught four real bugs and is the
-highest-value testing move available:
+**Node harness.** The Nova API is stubbable, so the extension can be exercised
+under Node — including against real session files on disk. This caught several
+real bugs and is the highest-value testing move available:
 
 ```js
 global.Process = class { /* wrap child_process.spawn; onStdout/onStderr/onDidExit */ };
 global.Range = class { constructor(start, end) { this.start = start; this.end = end; } };
 global.nova = { environment: process.env, path, fs, workspace, extension, crypto,
                 inDevMode: () => true, /* … */ };
-const { ClaudeClient } = require("./Scripts/client.js");
+const store = require("./Scripts/sessions-store.js");
 ```
 
-Then wire `onPermission` to scripted answers and send a prompt. Stub
-`showChoicePalette` / `showActionPanel` / `showInputPanel` / `notifications.add`
-to script UI decisions and assert on the resulting `PermissionResult`.
+Point `nova.workspace.path` at a directory that really has sessions, stub
+`clipboard.writeText` to capture output, and assert on the built command. Stub
+`showChoicePalette` / `showActionPanel` / `notifications.add` to script UI.
+`nova.fs.open` needs a working `readline()` — session labels depend on it.
 
 > The harness scripts from the build session lived in a temporary scratchpad and
 > are gone. Recreating them in a `Tests/` folder is the obvious next improvement.
 
-**Live tests cost real money** — they run turns on the signed-in account. Use
-`--model haiku` and one-line prompts; a full verification pass is a few tens of
-cents.
+Nothing in the current feature set spends tokens: the extension builds command
+strings and reads files. (If you add code that runs turns, use `--model haiku`
+and one-line prompts — a verification pass costs a few tens of cents.)
 
 Cheap checks that need no CLI:
 
@@ -128,9 +135,12 @@ python3 -c "import json; json.load(open('extension.json'))"
 
 ## State
 
-- Not yet committed. Not yet published. `min_runtime` is declared as `10`
-  (`nova.crypto` is the newest API touched, and it has a fallback).
+- `89d1a06` is the in-editor-chat implementation, kept for the IDE bridge.
+  The terminal-first rewrite on top of it is uncommitted.
+- Not published. `min_runtime` is declared as `10`.
 - `nova extension validate` last passed cleanly on the schema; a later run
   failed only on network reachability from a sandbox, not on the bundle.
+- **Never run inside Nova yet by this project's tooling** — the user has run
+  earlier builds; verify UI changes in the app.
 - Icons are generated by a script (a tapered starburst rendered to PNG with a
   pure-stdlib encoder); regenerate rather than hand-editing if sizes change.