feat(computer-use): cua-driver backend, universal any-model schema

Background macOS desktop control via cua-driver MCP — does NOT steal the
user's cursor or keyboard focus, works with any tool-capable model.

Replaces the Anthropic-native `computer_20251124` approach from the
abandoned #4562 with a generic OpenAI function-calling schema plus SOM
(set-of-mark) captures so Claude, GPT, Gemini, and open models can all
drive the desktop via numbered element indices.

## What this adds

- `tools/computer_use/` package — swappable ComputerUseBackend ABC +
  CuaDriverBackend (stdio MCP client to trycua/cua's cua-driver binary).
- Universal `computer_use` tool with one schema for all providers.
  Actions: capture (som/vision/ax), click, double_click, right_click,
  middle_click, drag, scroll, type, key, wait, list_apps, focus_app.
- Multimodal tool-result envelope (`_multimodal=True`, OpenAI-style
  `content: [text, image_url]` parts) that flows through
  handle_function_call into the tool message. Anthropic adapter converts
  into native `tool_result` image blocks; OpenAI-compatible providers
  get the parts list directly.
- Image eviction in convert_messages_to_anthropic: only the 3 most
  recent screenshots carry real image data; older ones become text
  placeholders to cap per-turn token cost.
- Context compressor image pruning: old multimodal tool results have
  their image parts stripped instead of being skipped.
- Image-aware token estimation: each image counts as a flat 1500 tokens
  instead of its base64 char length (~1MB would have registered as
  ~250K tokens before).
- COMPUTER_USE_GUIDANCE system-prompt block — injected when the toolset
  is active.
- Session DB persistence strips base64 from multimodal tool messages.
- Trajectory saver normalises multimodal messages to text-only.
- `hermes tools` post-setup installs cua-driver via the upstream script
  and prints permission-grant instructions.
- CLI approval callback wired so destructive computer_use actions go
  through the same prompt_toolkit approval dialog as terminal commands.
- Hard safety guards at the tool level: blocked type patterns
  (curl|bash, sudo rm -rf, fork bomb), blocked key combos (empty trash,
  force delete, lock screen, log out).
- Skill `apple/macos-computer-use/SKILL.md` — universal (model-agnostic)
  workflow guide.
- Docs: `user-guide/features/computer-use.md` plus reference catalog
  entries.

## Tests

44 new tests in tests/tools/test_computer_use.py covering schema
shape (universal, not Anthropic-native), dispatch routing, safety
guards, multimodal envelope, Anthropic adapter conversion, screenshot
eviction, context compressor pruning, image-aware token estimation,
run_agent helpers, and universality guarantees.

469/469 pass across tests/tools/test_computer_use.py + the affected
agent/ test suites.

## Not in this PR

- `model_tools.py` provider-gating: the tool is available to every
  provider. Providers without multi-part tool message support will see
  text-only tool results (graceful degradation via `text_summary`).
- Anthropic server-side `clear_tool_uses_20250919` — deferred;
  client-side eviction + compressor pruning cover the same cost ceiling
  without a beta header.

## Caveats

- macOS only. cua-driver uses private SkyLight SPIs
  (SLEventPostToPid, SLPSPostEventRecordTo,
  _AXObserverAddNotificationAndCheckRemote) that can break on any macOS
  update. Pin with HERMES_CUA_DRIVER_VERSION.
- Requires Accessibility + Screen Recording permissions — the post-setup
  prints the Settings path.

Supersedes PR #4562 (pyautogui/Quartz foreground backend, Anthropic-
native schema). Credit @0xbyt4 for the original #3816 groundwork whose
context/eviction/token design is preserved here in generic form.
This commit is contained in:
Teknium 2026-04-23 16:44:24 -07:00
parent 24f139e16a
commit b07791db05
No known key found for this signature in database
23 changed files with 2861 additions and 27 deletions

View file

@ -91,6 +91,13 @@ Scoped to the Feishu document-comment handler. Drives comment read/write operati
| `ha_list_entities` | List Home Assistant entities. Optionally filter by domain (light, switch, climate, sensor, binary_sensor, cover, fan, etc.) or by area name (living room, kitchen, bedroom, etc.). | — |
| `ha_list_services` | List available Home Assistant services (actions) for device control. Shows what actions can be performed on each device type and what parameters they accept. Use this to discover how to control devices found via ha_list_entities. | — |
## `computer_use` toolset
| Tool | Description | Requires environment |
|------|-------------|----------------------|
| `computer_use` | Background macOS desktop control via cua-driver — screenshots (SOM / vision / AX), click / drag / scroll / type / key / wait, list_apps, focus_app. Does NOT steal the user's cursor or keyboard focus. Works with any tool-capable model. macOS only. | `cua-driver` on `$PATH` (install via `hermes tools`). |
:::note
**Honcho tools** (`honcho_profile`, `honcho_search`, `honcho_context`, `honcho_reasoning`, `honcho_conclude`) are no longer built-in. They are available via the Honcho memory provider plugin at `plugins/memory/honcho/`. See [Memory Providers](../user-guide/features/memory-providers.md) for installation and usage.
:::