Align Daytona computer-use shims

Author: homanp

AGENTS.md

@@ -117,8 +117,9 @@ Useful flags:
 - `--sandbox-mode docker|local` selects static agent isolation. Use `local` only in a disposable
   outer sandbox such as Daytona, ephemeral CI, or a throwaway VM.
 - `--execute-app` enables opt-in live verification in a disposable runtime workspace.
-- `--computer-use daytona` exposes Daytona Toolbox-backed `cu-info`, `cu-screenshot`,
-  `cu-click`, `cu-type`, `cu-key`, `cu-scroll`, and `cu-drag` commands to live Pi agents.
+- `--computer-use daytona` exposes Daytona Toolbox-backed `cu-*` commands to live Pi agents,
+  including display/window inspection, screenshots, click aliases, mouse movement/button control,
+  keyboard input/hotkeys, scrolling, dragging, and raw API calls.
   PITHOS calls the already-running sandbox Computer Use API on `127.0.0.1:2280`;
   do not require a client-provided Daytona Toolbox URL.
 - `--runtime-profile <path>` points at an explicit `.pithos/runtime.yaml` profile.
@@ -140,17 +141,21 @@ When Daytona computer use is requested, enforce all of the following behavior:
 - Require local runtime execution: `--sandbox-mode local` or `environment.sandbox: local` must be used
   so `127.0.0.1:2280` resolves to the outer Superagent sandbox daemon. Treat Docker runtime execution
   as a preflight issue for Daytona computer use, not as a soft warning.
-- Install `cu-info`, `cu-screenshot`, `cu-click`, `cu-type`, `cu-key`, `cu-scroll`, and `cu-drag`
-  before starting the live verification agent, and prepend the shim directory to the agent `PATH`.
+- Strip Daytona and Superagent sandbox-creation credentials from the live-agent environment
+  (`DAYTONA_*`, `PITHOS_DAYTONA_*`, and `SUPERAGENT*`). The agent should only receive local
+  `127.0.0.1:2280` computer-use access, never credentials that can create more sandboxes.
+- Install the full `cu-*` shim set before starting the live verification agent, and prepend the shim
+  directory to the agent `PATH`.
 - Run or require a smoke check (`cu-info` or direct `GET /computeruse/display/info`) before relying on
   Daytona Computer Use. If the daemon is unreachable, mark runtime verification blocked with a clear
   setup error instead of silently falling back to non-UI verification.
 - Route findings that need browser, desktop, terminal UI, OAuth/device flow, or visual interaction
   through the live-agent path when Daytona computer use is requested. Deterministic source-oracle
   plugins may still run for purely static/source-verifiable findings, but they must not be presented as
   evidence that Daytona computer use was exercised.
-- The live-agent prompt must require `cu-info` and `cu-screenshot` before UI actions, and another
-  screenshot after UI actions. The verdict must cite the relevant computer-use evidence.
+- The live-agent prompt may allow normal `read`/`bash` investigation, but it must require `cu-info`
+  and `cu-screenshot` before UI actions, another screenshot after UI actions, and computer-use
+  evidence before accepting a verdict.
 - If a Daytona-enabled live-agent verdict does not include transcript or artifact evidence of at least
   one `cu-*` command, treat that verdict as blocked or inconclusive rather than confirmed or
   not-reproduced.

README.md

@@ -103,9 +103,10 @@ pithos run /path/to/repo \
 ```
 
 PITHOS does not start Daytona, Xvfb, XFCE, x11vnc, or noVNC. It creates
-per-run `cu-info`, `cu-screenshot`, `cu-click`, `cu-type`, `cu-key`,
-`cu-scroll`, and `cu-drag` shims that call the already-running sandbox Computer
-Use API on `127.0.0.1:2280`.
+per-run `cu-*` shims for display/window inspection, screenshots, mouse clicks
+and movement, keyboard input/hotkeys, scrolling, dragging, and raw Computer Use
+API calls. The shims call the already-running sandbox Computer Use API on
+`127.0.0.1:2280`.
 
 If setup is incomplete, PITHOS writes `verify/RUNTIME-SETUP.md` with the missing
 env vars, personas, mocks, or profile fields. Live verification does not load

docs/configuration.md

@@ -159,9 +159,12 @@ pithos run /path/to/repo \
   --computer-use daytona
 ```
 
-PITHOS creates per-run `cu-info`, `cu-screenshot`, `cu-click`, `cu-type`,
-`cu-key`, `cu-scroll`, and `cu-drag` commands and prepends them to the live
-agent's `PATH`. The commands call the already-running sandbox Computer Use API
-on `127.0.0.1:2280`. The live-agent prompt tells Pi to screenshot before and
-after UI actions and to use `cu-info`, `DISPLAY`, and `xdpyinfo` diagnostics when
-desktop actions do not affect the expected screen.
+PITHOS creates per-run `cu-*` commands and prepends them to the live agent's
+`PATH`. These include display/window inspection, screenshots, click aliases,
+mouse movement/button control, keyboard input/hotkeys, scrolling, dragging, and
+`cu-api` for raw Computer Use API calls. The commands call the already-running
+sandbox Computer Use API on `127.0.0.1:2280`. The live-agent prompt tells Pi to
+use normal investigation tools as needed, but always verify through `cu-*`
+computer-use evidence before accepting a runtime verdict. PITHOS strips
+Daytona/Superagent sandbox-creation credentials from the live-agent environment
+so the agent can control only the existing local computer-use daemon.