+- **`agent_cascade/`** + `commands/agent_cascade/` — `assembly agent-cascade`: the same live terminal conversation as `assembly agent`, but **client-orchestrated** — `engine.run_cascade` wires Streaming STT → the LLM Gateway → streaming TTS itself instead of talking to the Voice Agent endpoint, mirroring what the `agent-cascade` `assembly init` template does server-side. **Sandbox-only** (streaming TTS has no prod host; guarded via `tts.session.require_available`). Reuses the agent slice's `DuplexAudio`/`AgentRenderer` and `core.client.stream_audio`/`core.llm.complete`/`tts.session.synthesize`; the three network legs are injected through `engine.CascadeDeps` (the `tts/session.py` seam) so the cascade — greeting, clause-level streaming TTS, barge-in — is unit-tested against fakes with no sockets/mic/speaker. The LLM leg is a deepagents graph (`brain.py`) streamed token-by-token via `brain.build_streamer` (`graph.stream(stream_mode="messages")`): **context-window management is the brain's job, not the engine's** — `create_deep_agent` wires deepagents' own `SummarizationMiddleware` into the stack (summarize the oldest turns, offload the evicted history to a file), so the engine feeds the *full* untrimmed running history each turn and lets the graph compact it; the old client-side `text.trim_history`/`config.max_history` sliding window is gone from this path (`max_history` now only drives the hand-rolled `--show-code`/`assembly init` cascade, which doesn't use deepagents). The engine buffers `SpeechDelta`s, flushes complete clauses with `text.pop_clauses` (soft-separator clauses gated by `engine._MIN_CLAUSE_CHARS`), and synthesizes each clause with **streaming TTS** (`tts.session.synthesize(on_audio=…)`) so audio starts on the first frame instead of after the whole reply. The reply runs on a throwaway producer thread feeding a `queue.Queue` the worker drains under a monotonic deadline (the wall-clock backstop that replaced `_complete_within`), and an abandoned-on-timeout graph leg's langchain `ThreadPoolExecutor` worker is detached (`_detach_executor_threads_since`) so it can't wedge interpreter exit. A `ToolNotice` surfaces the "Searching the web…" affordance and drops any unspoken preamble. Under `-v` (`debuglog.active()`) `brain._stream_graph` logs each accumulated assistant line, tool call, and tool result as it streams. **Front-end:** an interactive mic session in human mode runs a **voice-only Textual TUI** (`agent_cascade/tui.py`, `LiveAgentApp`) by default — there's no text input (you can't type to it), just a transcript + an animated voice bar tracking listening/thinking/speaking. It uses its own `banner` wordmark, `messages` widgets, and `tui_status.voicebar_markup`/`VOICE_FRAMES` — all modules that now live in `agent_cascade/`; the blocking `run_cascade` runs on a worker thread and reaches the UI through a `_TuiRenderer` (the `engine.Renderer` protocol) that hops each call onto the UI thread, and a quit calls `DuplexAudio.close` to end the mic iterator and unblock that worker. `_exec._should_use_tui` gates it: file/sample input, `--json`/`-o text`, and a non-TTY all fall back to the plain `AgentRenderer` line output. **`--files`** (on by default; `--no-files` opts out) swaps the brain's in-memory backend for a real-cwd, sandbox-capable `SandboxedShellBackend` (`aai_cli/agent_cascade/sandbox.py`): file ops behave as before (traversal-blocked `virtual_mode`), and because it implements `SandboxBackendProtocol` deepagents binds a *functional* `execute` that runs commands OS-sandboxed in the real cwd — `sandbox-exec` (SBPL) on macOS, `bwrap` on Linux, refused (never an unconfined fallback) on any other platform or with the sandbox binary missing; the OS sandbox blocks the network, confines writes to cwd (+ the temp dir), and read-denies credential stores (`~/.ssh`/`~/.aws`/…, `.env*`, `.claude/`). The policy renderers are pure and the subprocess/capability boundaries injected, so the suite asserts *what we'd run* with no real sandbox. `write_file`/`edit_file`/`execute` are gated via `interrupt_on` + an `InMemorySaver`. The write gate is **path-scoped (the allow/interrupt permission model), not all-or-nothing**: `write_gate.write_interrupt_on(config.auto_write_paths)` hands `write_file`/`edit_file` one `InterruptOnConfig` whose `when` predicate fires only for writes *outside* every `--auto-write` subtree, so a write inside one runs ungated while others still pause (empty paths gate every write, the prior behavior). The matching is a transparent posix path-segment prefix check — `--auto-write DIR` is a directory subtree, and reads stay ungated, so no glob/bulk-tool machinery is needed (and no reach into deepagents internals: `create_deep_agent(permissions=…)` is unusable here — it raises `NotImplementedError` against an execute-capable backend like our `SandboxedShellBackend`). A non-string or `..`-containing target fails safe to gating. `execute` stays a plain `True` (unconditionally gated) — a command isn't a single file path to scope. `--auto-write DIR` is normalized to a cwd-rooted virtual root in `_exec._normalize_auto_write` (rejecting `..`/`~`/whole-tree aliases) and requires `--files`. `brain._stream_gated` detects the post-stream interrupt (`graph.get_state(config).interrupts`), asks an injected `Approver`, and resumes with `Command(resume=…)`, bracketing the human wait in `ApprovalPause` events so `engine._consume` suspends its reply deadline (`risk.py` surfaces a shell-risk warning on the prompt). The voice TUI supplies the approver via `agent_cascade.modals.ApprovalScreen` (`y`/`a`/`n`), which can *also* be resolved hands-free by voice: while a write awaits approval, `_consume` arms `_awaiting_approval` and `engine.on_turn` routes the next final transcript to `app.submit_voice_approval` → `ApprovalScreen.try_voice`, which applies `spoken_approval.spoken_decision` (an unambiguous affirmative approves, anything else rejects — fail-safe; destructive `risk.py`-flagged commands ignore the spoken answer and require a keypress). **Project grounding (independent of `--files`):** `_exec.run_agent_cascade` reads the launch directory's `AGENTS.md`/`CLAUDE.md` via `agent_cascade/project_context.load_project_context()` into `CascadeConfig.project_context`, which `brain.build_graph` threads into `prompt.build_system_prompt(..., project_context=…)` (appended as project background after the persona/tool guidance). `AGENTS.md` wins precedence, identical content (a symlinked `CLAUDE.md`) is de-duplicated, and the total is capped at `project_context.MAX_CONTEXT_CHARS`. It's read at the command boundary (not in `build_graph`) so the brain stays hermetic, and the `--show-code` path builds its own config without it. Headless runs auto-deny (`_exec._deny_writes`). `--files` also turns on durable per-project memory via deepagents' `MemoryMiddleware` (`memory=["./.deepagents/AGENTS.md"]`), distinct from the in-session `InMemorySaver`. The gateway-bound, sandbox-backed general-purpose subagent (deepagents' `task` tool) for delegating a focused subtask is **auto-added by deepagents** — we don't declare it. We only override its prose for a voice turn (a spoken-length summary, not the SDK's "complete answer" default) via a harness profile keyed by the gateway model's provider (`subagents.register_gp_subagent_profile`, called from `build_graph` so the deepagents import stays lazy — and kept off `brain.py`, which sits at the 500-line gate). It inherits the gateway-bound model, the sandboxed toolset, *and* the top-level `interrupt_on` (deepagents' `graph.py` merges the top-level config into the auto-added subagent), so a delegated `write_file`/`edit_file`/`execute` surfaces at the *parent* `get_state().interrupts` with no per-subagent restatement (so `_pending_writes` gates it too — verified by a HITL spike, locked in `tests/test_agent_cascade_subagents.py`). Reads (incl. `grep`) stay ungated.
0 commit comments