Staging Golden Tier

Phase 1 — Inventory

List the staging folder and cross-reference against the manifest.

# List staging files
gws drive files list --params '{"q": "\"1W169BsPkexomhf6Qeyz-6dFOkTn-Z2UW\" in parents and trashed=false", "includeItemsFromAllDrives": true, "supportsAllDrives": true, "fields": "files(id,name,modifiedTime,mimeType,size)", "pageSize": 100}'

For each file:

1. Parse the filename to extract person, session type, and stage
2. Check manifest.yaml for an existing entry matching that person + stage
3. Flag duplicates for comparison, mark new sessions for processing

Present the inventory table to the user for approval before proceeding.

Phase 2 — Download & Convert

For each approved file:

# Download .docx files
gws drive files get --params '{"fileId": "<ID>", "alt": "media", "supportsAllDrives": true}' --output "/tmp/staging/<filename>"

# Convert .docx to .md
pandoc -f docx -t markdown --wrap=none "/tmp/staging/<filename>.docx" -o "/tmp/staging/<filename>.md"

For files already in .md format, download directly — no conversion needed.

Phase 3 — Parse & Extract

Identify the source type and extract the transcript content.

Gemini Notes exports (most common in staging):

• The .docx contains both AI summary notes AND an embedded transcript
• The transcript section starts after a line containing Transcript (often preceded by a clipboard emoji)
• Everything before that line is the notes section (Summary, Details, Suggested next steps)
• Extract both sections: transcript → _transcript.md, notes → _notes.md

Read AI exports (.txt files):

• Pure transcript, no notes section
• Format: TIMESTAMP - Speaker Name followed by dialogue on next lines

Pre-cleaned files (already .md with some processing done):

• Validate existing structure against the spec
• Apply remaining cleaning steps as needed

Phase 4 — Clean

Run the deterministic cleanup script first:

python3 scripts/clean_transcript.py "/tmp/staging/<file>.md" -o "/tmp/staging/<file>_cleaned.md"

This handles: pandoc artifacts, Unicode normalisation, empty spacer lines, Gemini disclaimer removal, timestamp anchor cleanup.

Then apply AI judgment (Claude does this directly):

1. Read references/known-speakers.md to map display names to canonical names and roles
2. Standardise speaker labels: map Gemini display names (e.g., Bails18 Wills → keep as-is for Gemini format; resolve Unidentified Speaker where possible from context)
3. For Read AI sources: convert to **Speaker Name** [Role] (H:MM:SS): format
4. For Gemini sources: preserve Speaker Name: text format with standalone timestamp blocks
5. Merge consecutive same-speaker turns only when clearly a single thought interrupted by the transcription
6. Correct transcription errors using the known errors table in known-speakers.md
7. Verify session metadata: confirm the date, participant name, and session type from transcript content (not just filename)

Phase 5 — Format & Place

Generate the YAML frontmatter:

python3 scripts/generate_frontmatter.py \
  --participant "Name" \
  --role "Support Coordinator" \
  --stage 4 \
  --session-type "MVP Testing" \
  --date "2026-04-09" \
  --source "Gemini embedded transcript" \
  --source-file "original.docx" \
  --content-type transcript \
  --has-companion-notes true \
  "/tmp/staging/<file>_cleaned.md"

This prepends the frontmatter and document header, calculates word count, and writes the final file.

Place the output file:

• Transcript: Golden_Tier/{SC_or_Participants}/{name}/{session_id}_{date}_transcript.md
• Notes: Golden_Tier/{SC_or_Participants}/{name}/{session_id}_{date}_notes.md

Phase 6 — Lock & Update Index

# Lock processed files
chmod 444 "Golden_Tier/{path}_transcript.md"
chmod 444 "Golden_Tier/{path}_notes.md"

# Update manifest
python3 scripts/update_manifest.py \
  --manifest "Golden_Tier/manifest.yaml" \
  --transcript "Golden_Tier/{path}_transcript.md" \
  --notes "Golden_Tier/{path}_notes.md"

Also update Golden_Tier/README.md:

• Update the folder structure section if new people added
• Update session counts in person entries
• Update gap register (remove filled gaps, note remaining ones)
• Update totals in the "What is this?" section

IMPORTANT: Every file write to the Golden Tier is a curated corpus modification. Present the diff for user review.

Phase 7 — Final Verification

Re-read each produced Golden Tier file and validate:

Present a verification summary table. Any failures get flagged for manual review.

Safety

• All Google Drive operations follow the GWS write guard: read operations run freely, write operations require explicit user approval
• Golden Tier files are set read-only after processing — to modify, the user must explicitly unlock
• The skill is idempotent: re-running on an already-processed file detects it via the manifest and skips
• Original staging files are NOT deleted from Drive — that requires a separate approval