Mercurial

annotate infinite_canvas/docs/agent-sessions.md @ 280:49e9e591c9bb

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
author MrJuneJune <me@mrjunejune.com>
date Tue, 18 Aug 2026 19:14:53 -0700
parents
children c57149ad216e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
280
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
1 # Dictation and Spatial Agent Sessions
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
2
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
3 ## Prototype flow
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
4
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
5 The first orchestration slice connects three retained systems:
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
6
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
7 1. `M` reuses the single retained `Dictation` text-area scratchpad and toggles
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
8 a hidden browser transport for `http://127.0.0.1:8090/?canvas=1`.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
9 2. `Canvas_Web_Surface_Set_Dictation_Active()` deterministically starts or
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
10 stops capture on an already-negotiated WebRTC client. The hidden transport
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
11 acquires its stream, completes signaling/ICE, and opens the transcript data
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
12 channel during startup with its audio track disabled.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
13 CEF grants microphone access only to the loopback dictation origin.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
14 3. Status, partial, and final transcript events cross the platform boundary:
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
15 - native CEF encodes each event in the document title and
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
16 `OnTitleChange()` queues it for the caption;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
17 - WebAssembly uses `window.postMessage()` from a hidden iframe.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
18 4. `main.c` accumulates speech in the scratchpad and submits it to
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
19 `Canvas_Agent_Service` only when the user presses `Enter`. A selected text
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
20 area can enter the same path with `Ctrl/Cmd+Enter`.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
21 Enter sends a `commit` command through the live transcript data channel, so
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
22 the current utterance finalizes without closing the microphone session.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
23 5. `agent_service_copilot.c` sends the new thought plus
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
24 `Canvas_Scene_Build_Visible_Context()` through the shared asynchronous
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
25 `Inference_Bridge`.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
26 6. The Bazel-managed Copilot SDK sidecar uses the `canvas_orchestrator` profile
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
27 and returns JSON choosing `create` or `append` and a conversation entity ID.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
28 The canvas materializes that decision as a retained rich conversation card.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
29
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
30 The scratchpad is excluded from serialized camera context because the submitted
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
31 thought is already sent separately. Only conversation entities currently in
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
32 view are valid append targets; moving a conversation out of view removes it
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
33 from the orchestrator's routing choices.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
34
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
35 ## Service boundary
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
36
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
37 `agent_service_copilot.c` submits work to `Inference_Bridge`, whose worker
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
38 thread owns the JSON-line Python sidecar process so SDK latency never blocks
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
39 Raylib. The sidecar maintains the persistent
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
40 `infinite-canvas-orchestrator` SDK session and uses the loopback LiteLLM
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
41 gateway for authenticated GitHub Copilot inference. Only one orchestration
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
42 request is admitted at a time in this prototype.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
43
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
44 `agent_service_copilot.c` warms that orchestration session plus three reserved
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
45 worker sessions during initialization. They share the same
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
46 `canvas_orchestrator` profile and move Copilot SDK client/session setup ahead of
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
47 the first user turn. The persistent orchestrator is resumed when available;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
48 reserved workers are created fresh because they have no conversation history
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
49 to recover, avoiding expected `session.resume` errors during startup.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
50
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
51 ## Conversation entities
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
52
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
53 `CANVAS_ENTITY_CONVERSATION` is intentionally more than an editable text area.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
54 It retains:
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
55
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
56 - a stable entity/session ID used by Copilot routing;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
57 - a title;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
58 - user and agent transcript content;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
59 - asynchronous working state;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
60 - independent scroll, selection, lifecycle, z-order, and pinning state.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
61 - auto-follow to the latest appended turn and a collapsed header-only state;
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
62 - a recoverable parked state that is excluded from orchestration context.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
63
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
64 Images, browser views, and other components already appear in visible canvas
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
65 context as typed neighboring entities. A future resource list can attach those
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
66 entity IDs directly to a conversation without replacing the conversation card
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
67 or flattening rich content into one string.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
68
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
69 Archiving a conversation records its prior position and moves it to the
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
70 off-canvas parking lot. The Developer Controls parking-lot button fits the
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
71 camera to parked sessions. Restoring a parked card returns it to its prior
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
72 position and makes it eligible for routing again when visible.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
73
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
74 ## Current constraints
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
75
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
76 - `//infinite_canvas:orchestration_dev` owns the dictation server, LiteLLM
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
77 gateway, and canvas lifecycle. `//infinite_canvas:agent_dev` is an alias.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
78 - One persistent Dictation scratchpad is reused and cleared after submission.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
79 Submitted turns are retained in conversation entities after orchestration.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
80 - One Copilot request runs at a time.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
81 - Conversation transcript storage is currently bounded by
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
82 `CANVAS_ENTITY_TEXT_CAPACITY`.
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
83 - Linux uses the Copilot SDK bridge. Other platforms currently select an
49e9e591c9bb Add persistent dictation, prewarmed WebRTC speech input, Copilot SDK routing, animated conversation lifecycle controls, parking, and architecture coverage.
MrJuneJune <me@mrjunejune.com>
parents:
diff changeset
84 explicit unavailable stub.