Load html previews from psyflow-web manifest

Author: zh1peng

README.md

@@ -4,7 +4,7 @@ Public, statically deployed website that lists TaskBeacon task repositories as a
 
 - Gallery: list and card views, maturity + paradigm filters, last updated, repo/download links
 - Task detail pages: README rendering, quick start, metadata, local access panel, optional web preview panel
-- Companion detection: baseline tasks automatically pick up matching HTML repos when they share the same `slug` and the companion declares `variant: html`
+- Companion detection: baseline tasks automatically pick up matching HTML repos at runtime from the shared `psyflow-web` manifest when they share the same `slug` and the companion declares `variant: html`
 - Deployment: static export to GitHub Pages via GitHub Actions
 
 ## Tech
@@ -55,7 +55,7 @@ It:
 2. Excludes non-task repos via `DENYLIST`.
 3. Fetches README content, repo root contents, and optional metadata files.
 4. Builds a normalized task record for each task repo.
-5. Merges matching HTML companions into the main task card as `web_variant`.
+5. Keeps the canonical task index focused on task repos and excludes duplicate HTML companions when a baseline task with the same `slug` exists.
 
 Supported metadata files, in priority order:
 
@@ -84,6 +84,14 @@ If both exist, the website keeps the main task card and adds:
 
 for the matched HTML companion.
 
+The companion buttons are attached at runtime by fetching:
+
+```text
+/psyflow-web/task-manifest.json
+```
+
+This keeps `taskbeacon.github.io` independent from HTML companion rebuilds. Only the shared `psyflow-web` runner needs to refresh when new `Hxxxxxx-*` repos appear.
+
 By default, HTML companions run through the shared `psyflow-web` GitHub Pages runner:
 
 ```text

changelog.md

@@ -0,0 +1,8 @@
+# Changelog
+
+## 2026-03-11
+
+- Switched HTML companion discovery from build-time index merging to runtime attachment via the shared `psyflow-web` manifest.
+- Added a shared client-side companion loader that fetches `/psyflow-web/task-manifest.json` and merges matching HTML previews onto canonical TaskBeacon tasks by `slug` and `T/H` task id pairing.
+- Updated the gallery, featured carousel, homepage preview counters, footer summary, and task detail pages to use runtime companion data without changing the visible card layout.
+- Simplified the indexer so `tasks_index.json` now stays focused on canonical task records, while HTML preview metadata lives in the `psyflow-web` manifest.

public/readmes/ANT.md

@@ -0,0 +1,27 @@
+# H000002 BART
+
+HTML/browser preview of the Balloon Analogue Risk Task built with `psyflow-web`.
+The within-trial pump/cash loop, deterministic explosion sampler, risk-condition semantics, instruction text, and feedback logic are aligned to the local `T000002-bart` task; only the block and trial count are shortened for preview use.
+
+## Layout
+
+- `main.ts`: task-level orchestration
+- `config/config.yaml`: declarative task config
+- `src/run_trial.ts`: trial-level logic
+- `src/utils.ts`: explosion sampler and block summary helpers
+- `assets/`: task-specific balloon and audio assets
+
+## Run
+
+From [psyflow-web](/e:/xhmhc/TaskBeacon/psyflow-web):
+
+```powershell
+npm install
+npm run dev
+```
+
+Then open:
+
+```text
+http://127.0.0.1:4173/?task=H000002-bart
+```

public/readmes/H000002-bart.md

@@ -0,0 +1,27 @@
+# H000012 SST
+
+HTML/browser preview of the Stop-Signal Task built with `psyflow-web`.
+The within-trial procedure, go/stop condition semantics, instruction text, and adaptive SSD controller are aligned to the local `T000012-sst` task; only the block and trial count are shortened for preview use.
+
+## Layout
+
+- `main.ts`: task-level orchestration
+- `config/config.yaml`: declarative task config
+- `src/run_trial.ts`: trial-level logic
+- `src/controller.ts`: adaptive SSD controller
+- `src/utils.ts`: constrained condition generator and block summary helpers
+
+## Run
+
+From [psyflow-web](/e:/xhmhc/TaskBeacon/psyflow-web):
+
+```powershell
+npm install
+npm run dev
+```
+
+Then open:
+
+```text
+http://127.0.0.1:4173/?task=H000012-sst
+```

public/readmes/H000012-sst.md

@@ -1,16 +1,15 @@
 # AX-CPT Task
 
-![Maturity: smoke_tested](https://img.shields.io/badge/Maturity-smoke_tested-d97706?style=for-the-badge&labelColor=c2410c)
-
+![Maturity: smoke_tested](https://img.shields.io/badge/Maturity-smoke_tested-d97706?style=flat-square&labelColor=111827)
 | Field | Value |
 |---|---|
 | Name | AX-CPT Task |
-| Version | main (1.0) |
+| Version | main (1.1.0) |
 | URL / Repository | https://github.com/TaskBeacon/T000001-ax-cpt |
 | Short Description | A task measuring cognitive control, context processing, and goal maintenance. |
 | Created By | Zhipeng Cao (zhipeng30@foxmail.com) |
-| Date Updated | 2025/07/25 |
-| PsyFlow Version |0.1.0|
+| Date Updated | 2026/02/16 |
+| PsyFlow Version | 0.1.8 |
 | PsychoPy Version |2025.1.1              |
 | Modality     |Behavior/EEG                  |
 | Language | Chinese |
@@ -20,6 +19,9 @@
 This task is an AX-Continuous Performance Test (AX-CPT) designed to assess cognitive control and context processing. Participants are presented with a series of letter pairs (cue followed by probe). They must make a specific target response only when the cue is an 'A' and the probe is an 'X' (AX trials). For all other combinations (AY, BX, BY), a different response (or no response) is required.
 
 ## 2. Task Flow
+
+![Task flow](./task_flow.png)
+
 ### Block-Level Flow
 | Step | Description |
 |---|---|
@@ -38,6 +40,13 @@ This task is an AX-Continuous Performance Test (AX-CPT) designed to assess cogni
 | 4. Feedback | Feedback (`Correct`, `Incorrect`, or `Too Slow`) is displayed. |
 | 5. Inter-Trial Interval (ITI) | A blank screen is shown before the next trial. |
 
+### Runtime Context Phases
+| Phase Label | Meaning |
+|---|---|
+| `context_cue` | context cue stage in `src/run_trial.py` responder context. |
+| `delay_fixation` | delay fixation stage in `src/run_trial.py` responder context. |
+| `probe_response` | probe response stage in `src/run_trial.py` responder context. |
+
 ## 3. Configuration Summary
 ### a. Subject Info
 | Field | Meaning |
@@ -66,9 +75,9 @@ This task is an AX-Continuous Performance Test (AX-CPT) designed to assess cogni
 | cue_B | text | The cue letter 'B'. |
 | probe_X | text | The probe letter 'X'. |
 | probe_Y | text | The probe letter 'Y'. |
-| correct_feedback | text | "正确" (Correct) in white. |
-| incorrect_feedback | text | "错误" (Incorrect) in white. |
-| no_response_feedback | text | "未反� (No response) in white. |
+| correct_feedback | text | "正确" (Correct) in white. |
+| incorrect_feedback | text | "错误" (Incorrect) in white. |
+| no_response_feedback | text | "未反? (No response) in white. |
 | instruction_text | textbox | Instructions explaining the task. |
 | instruction_text_voice | voice | Voice instructions for the task. |
 | block_break | text | Feedback shown between blocks. |
@@ -82,7 +91,7 @@ This task is an AX-Continuous Performance Test (AX-CPT) designed to assess cogni
 | isi | 0.5 | isi_duration |
 | probe | 1.0 (max response time) | probe_duration |
 | feedback | 0.5 | feedback_duration |
-| iti | random 0.8�.2 | iti_duration |
+| iti | random 0.8?.2 | iti_duration |
 
 ### e. Triggers
 | Event | Code |
@@ -104,9 +113,9 @@ This task is an AX-Continuous Performance Test (AX-CPT) designed to assess cogni
 | feedback_no_response | 53 |
 | feedback_onset | 60 |
 
-## 4. Methods
+## 4. Methods (for academic publication)
 Participants perform an AX-CPT task, a continuous performance paradigm used to measure cognitive control and context processing. Each trial consists of a cue letter (A or B) followed by a probe letter (X or Y). Participants are instructed to press the 'f' key if the cue is 'A' and the probe is 'X' (AX trials). For all other cue-probe combinations (AY, BX, BY), they are instructed to press the 'j' key.
 
 The task begins with a fixation cross, displayed for 500ms. The cue letter is then presented for 500ms, followed by a brief inter-stimulus interval (ISI) fixation of 500ms, and then the probe letter for up to 1000ms or until a response is made. Feedback (correct, incorrect, or no response) is provided for 500ms, followed by a randomized inter-trial interval between 800ms and 1200ms.
 
-The task is structured into 4 blocks of 40 trials each (total 160 trials). Trial types (AX, AY, BX, BY) are presented in a randomized order within each block, with a higher proportion of AX trials to establish a strong context. This design allows for the examination of proactive and reactive cognitive control strategies.
+The task is structured into 4 blocks of 40 trials each (total 160 trials). Trial types (AX, AY, BX, BY) are presented in a weighted randomized order within each block (`AX=40%`, `AY=10%`, `BX=10%`, `BY=40%`) to preserve strong context expectancy while retaining critical low-frequency AY/BX control trials. This design allows for the examination of proactive and reactive cognitive control strategies.