Run P.808 family of tests on Prolific platform + agent specification#96
Open
babaknaderi wants to merge 29 commits into
Open
Run P.808 family of tests on Prolific platform + agent specification#96babaknaderi wants to merge 29 commits into
babaknaderi wants to merge 29 commits into
Conversation
Contributor
babaknaderi
commented
May 22, 2026
- Multiple fixes and updates.
- Support Prolific Crowdsourcing platform
- Added Agent support for creating and parsing results.
…and added agent instruction to create study.
- Rewrite .github/create.instruction.md for the P.808 repo with correct paths, column names, method mappings, and end-to-end automation steps - Add src/utils/download_clips.py to download clips from CSV URLs locally - Add src/utils/select_training_clips.py to select training clips from rating clips - Update AGENTS.md with task-specific instruction file references for auto-discovery by agents - Fix librosa API compatibility in create_trapping_stimuli.py (get_duration path= and resample orig_sr=/target_sr= kwargs) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Add 'upload' mode that uses 'az storage blob upload-batch' with az login credentials for zero-manual-step uploads - Stages only needed files (not CSVs/txt) via temp directory - Falls back to 'upload-local' (azcopy command) when az CLI unavailable - Update create.instruction.md to use direct upload as the default - Existing upload-local and copy-remote modes are preserved Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…P.804 ver assignment - Split source clip note into gold-specific (requires clean reference audio) and trapping-specific (quality does not matter, agent can download directly) - P.804: replace two independent gold sets with single set approach, assigning ver=1 for ovrl_ans=5 and ver=2 for ovrl_ans=1 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Move language.instructions.md to .github/instructions/ for proper CLI discovery - Add Custom Agents section to copilot-instructions.md - Add create-study agent definition in .github/agents/ - Update AGENTS.md and README.md Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Points Claude Code users to coding standards and the create-study agent runbook so the agent is discoverable across both Copilot CLI and Claude Code. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Instructs any AI agent to adapt PowerShell commands and Windows paths to the user's OS and shell, making the runbook usable from Claude Code on macOS/Linux as well as Copilot CLI on Windows. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
All instances of C:\my\repos\P.808 are now REPO_ROOT, which the agent resolves to the actual Git root at runtime. This makes the runbook portable across clones, machines, and operating systems. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The quick-reference was a condensed copy of the main workflow (sections 3-10). All six known issues were already documented inline in their respective workflow steps. Removing both eliminates maintenance risk from duplicated information. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ates - Rename 'src/trapping clips/' to 'src/trapping_clips_assets/' (avoid spaces) - Keep sample source WAVs (s1-s12.wav) in new location - Update all config files and docs to reference new directory name - Add .gitkeep to output/ subdirectory - Fix CSV double line endings in select_training_clips.py - Update agent instructions (create-study.agent.md): - Add PowerShell execution policy bypass guidance - Add training_gold_clips concept for P.804 and pp835 - Ask for manual training clip selection first, fall back to auto - Add [ASK] review step after generating gold/trapping/training clips - Make --check_urls and --create_local_test mandatory - Use random opaque dir names for rating clip uploads - Fix cleanup/review messaging after file deletion - Update all trapping directory references Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Add --sas_token parameter to download_clips.py for private storage - Clarify training_gold_clips priority over training_clips for P.804/pp835 - Add URL mapping CSV generation step (step 9b) - Document check_urls_in_files_exist for validation with SAS token support - Expand azcopy guidance: prefer SAS token over azcopy login - Condense agent file to fit within 30,000 char limit Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Agent: generate SAS tokens with az CLI instead of asking user for them - Agent: use SAS tokens on both source and destination for azcopy transfers (never rely on azcopy login which expires after 90 days) - Agent: add note about azcopy --recursive preserving directory structure - P.804 template: remove outdated information Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Add CCR score mappings (-3 to 0) to create_trapping_stimuli.py - Add trapping_ccr.cfg with adj_ccr_score_ message prefix - Update agent doc: CCR trapping config in method tables - Add reuse-confirmation workflow for multi-study sessions Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
DCR and CCR do not use traditional trapping clips with overlaid spoken scores. The TP field is a reference clip used as a gold/control question (legacy naming). Update agent doc to skip create_trapping_stimuli.py for these methods and use reference clips directly. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
New agent runbook for processing crowdsourced subjective test results: - Runs result_parser.py with config, answers, and optional Prolific data - Validates inputs, checks rejection rates (>35% alarming) - Analyzes gold question performance (>20% rejection alarming) - Provides structured summary and points to output files - Documents scale suffixes per method and output file patterns Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Updated docstring in check_tps() and inline comment in data_cleaning() to note that CCR and DCR use gold questions coded as trapping questions. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Corrected user-facing text and accessibility attributes across all 15 HTML templates. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
Author
|
@copilot resolve the merge conflicts in this pull request |
Co-authored-by: babaknaderi <15108406+babaknaderi@users.noreply.github.com>
Merge conflicts resolved in commit
|
Copilot stopped reviewing on behalf of
babaknaderi due to an error
May 22, 2026 07:40
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.