@firstpick/pi-package-webui 0.7.8 → 0.8.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (41) hide show
  1. package/README.md +10 -4
  2. package/bin/pi-webui.mjs +696 -69
  3. package/lib/git-workflow-preferences.mjs +16 -0
  4. package/lib/git-worktrees.mjs +43 -4
  5. package/lib/queue-mutation.mjs +109 -0
  6. package/lib/scoped-models.mjs +19 -0
  7. package/lib/webui-workspaces.mjs +265 -0
  8. package/package.json +4 -2
  9. package/public/app.js +2074 -188
  10. package/public/index.html +105 -4
  11. package/public/service-worker.js +1 -1
  12. package/public/styles.css +618 -73
  13. package/public/voice-conversation.mjs +1 -1
  14. package/tests/boot-failure-diagnostics.test.mjs +2 -2
  15. package/tests/codex-fast-mode-static.test.mjs +56 -0
  16. package/tests/compaction-mutation-contract.test.mjs +166 -0
  17. package/tests/compaction-resume-harness.test.mjs +113 -2
  18. package/tests/compaction-static.test.mjs +5 -0
  19. package/tests/composer-skill-tag-layout.test.mjs +70 -0
  20. package/tests/cwd-footer-reliability-static.test.mjs +5 -0
  21. package/tests/fast-mode-client-static.test.mjs +8 -8
  22. package/tests/file-selection-overlay-static.test.mjs +21 -0
  23. package/tests/file-viewer-search-static.test.mjs +65 -0
  24. package/tests/fixtures/fake-pi.mjs +216 -43
  25. package/tests/git-panel-file-preview-static.test.mjs +20 -5
  26. package/tests/git-workflow-preferences.test.mjs +8 -0
  27. package/tests/guided-git-publication-backend-static.test.mjs +40 -0
  28. package/tests/http-endpoints-harness.test.mjs +422 -2
  29. package/tests/mobile-static.test.mjs +111 -42
  30. package/tests/native-parity.test.mjs +1 -1
  31. package/tests/open-issue-wizard-static.test.mjs +1 -1
  32. package/tests/queue-edit-reorder-static.test.mjs +101 -0
  33. package/tests/queue-mutation-contract.test.mjs +143 -0
  34. package/tests/scoped-models.test.mjs +42 -0
  35. package/tests/side-panel-resize-static.test.mjs +8 -1
  36. package/tests/subagents-helper.test.mjs +216 -3
  37. package/tests/terminal-tab-workspace-static.test.mjs +4 -2
  38. package/tests/webui-workspaces-harness.test.mjs +223 -0
  39. package/tests/webui-workspaces.test.mjs +106 -0
  40. package/tests/workspace-save-load-static.test.mjs +69 -0
  41. package/webui-rpc-helper.mjs +300 -41
package/README.md CHANGED
@@ -153,13 +153,19 @@ Operations and recovery:
153
153
 
154
154
  ### Compact live output mode
155
155
 
156
- Normal output is the default. In the sidebar, open **Controls → Output processing**, select **Fast**, and click **Apply**. The same persisted server default is also available under **Settings → Browser workflow → Output processing**. Alternatively, set `--output-mode compact-v1`, `PI_WEBUI_OUTPUT_MODE=compact-v1`, or `outputModeDefault` directly. Precedence is explicit CLI flag, then environment variable, then the persisted setting, then `normal`.
156
+ Normal output is the default. In the sidebar, open **Controls → Output processing**, select **Compact**, and click **Apply**. The same persisted server default is also available under **Settings → Browser workflow → Output processing**. Alternatively, set `--output-mode compact-v1`, `PI_WEBUI_OUTPUT_MODE=compact-v1`, or `outputModeDefault` directly. Precedence is explicit CLI flag, then environment variable, then the persisted setting, then `normal`.
157
157
 
158
- The browser negotiates compact-v1 per EventSource connection (`outputMode=auto&outputModeProtocol=1`), so normal and compact clients can share one Pi tab. A browser only enables compact handling after the protocol-1 acknowledgement; an older server gets one normal-mode reconnect instead. The server default applies only to `auto` clients, and active auto streams change representation at a semantic boundary.
158
+ The browser negotiates `compact-v1` per EventSource connection (`outputMode=auto&outputModeProtocol=1`), so normal and compact clients can share one Pi tab. A browser only enables compact handling after the protocol-1 acknowledgement; an older server gets one normal-mode reconnect instead. The server default applies only to `auto` clients, and active auto streams change representation at a semantic boundary.
159
159
 
160
- Fast mode keeps live output lightweight, coalesces sustained live DOM/scroll flushes to at most once every 100 ms, preserves Markdown formatting for live and reconciled thinking, and displays only the current transient tool status as a compact labeled row with a state pill; each new tool or assistant delta replaces that status instead of retaining a tool-call history. Final reconciliation keeps visible thinking formatted and renders the final assistant response as Markdown while omitting intermediate tool calls, results, diffs, images, and raw details. The existing final `/api/messages` transcript remains authoritative. Fast mode does not change Pi generation, prompts, tools, models, providers, inference, stored transcript semantics, or the normal renderer.
160
+ Compact mode keeps live output lightweight, coalesces sustained live DOM/scroll flushes to at most once every 100 ms, preserves Markdown formatting for live and reconciled thinking, and displays only the current transient tool status as a compact labeled row with a state pill; each new tool or assistant delta replaces that status instead of retaining a tool-call history. Final reconciliation keeps visible thinking formatted and renders the final assistant response as Markdown while omitting intermediate tool calls, results, diffs, images, and raw details. The existing final `/api/messages` transcript remains authoritative. Compact mode does not change Pi generation, prompts, tools, models, providers, inference, stored transcript semantics, or the normal renderer.
161
161
 
162
- The included fast-mode metric measures deterministic post-parse serialized JSON byte-work only (`R + 2×S`) and semantic parity; it does not claim wall-clock improvement, lower DOM CPU, reduced network latency, or higher model token/s.
162
+ The included compact-mode metric measures deterministic post-parse serialized JSON byte-work only (`R + 2×S`) and semantic parity; it does not claim wall-clock improvement, lower DOM CPU, reduced network latency, or higher model token/s.
163
+
164
+ ### Codex subscription Fast mode
165
+
166
+ The optional `@firstpick/pi-extension-codex-fast-mode` companion adds `/fast-mode` and a **Normal / Fast** selector under **Codex Usage**. Fast mode is off by default and scoped to the active Pi session branch. The browser asks the extension to change mode and renders the extension-published `codex-fast-mode` status; it does not inspect ChatGPT credentials or rewrite provider requests itself.
167
+
168
+ When enabled, the extension marks only subscription-backed `openai-codex` requests using `openai-codex-responses` with `service_tier: "priority"`. Supported models may respond about 1.5× faster while spending 2× Standard credits for GPT-5.4 or 2.5× for GPT-5.5/5.6. Upstream account and model eligibility remains authoritative. Mode changes are blocked while the active tab is busy, and disabling the optional feature turns Fast mode off before hiding its integration.
163
169
 
164
170
  Optional Natural Conversation server-side voice fallback variables:
165
171