vault-ingress

name: vault-ingress description: > Parses presentation talks to catalog rhetoric patterns: opening hooks, humor style, pacing, transitions, audience interaction, slide design, and verbal signatures. Downloads YouTube transcripts and analyzes slides (from PPTX, Google Drive PDFs, or video extraction), examining HOW the speaker presents. Processes talks in parallel batches and updates the running rhetoric summary. Triggers: "parse my talks", "run the rhetoric analyzer", "analyze my presentation style", "how many talks have been processed", "update the rhetoric knowledge base", "check rhetoric vault status", "process remaining talks for style patterns". user_invocable: true

Vault Ingress — Incremental Talk Parser

Process the steps below in order; each step's output (tracking DB state, batch results, per-talk artifacts) feeds the next. Do not skip ahead.

Build a rhetoric and style knowledge base by analyzing presentation talks. Each run processes unprocessed talks, extracts rhetoric/style observations, and updates the running summary. The vault lives at ~/.claude/rhetoric-knowledge-vault/ (may be a symlink to a custom location). All paths are relative to this vault root.

Key Files & References

A talk is processable when it has video_url. Slide sources, in order of preference:

1. pptx_path — richest data (exact colors, fonts, shapes via python-pptx)
2. slides_url — download PDF from Google Drive
3. video_url — extract slides from the video using ffmpeg + perceptual dedup
4. none — transcript-only analysis (processed_partial)

The slide_source field tracks which path: "pptx", "pdf", "both", "video_extracted", or "none". The pptx_catalog array fuzzy-matches .pptx files to shownotes entries.

Step 1 — Bootstrap Vault State

Vault discovery — canonical path is always ~/.claude/rhetoric-knowledge-vault/.

1. Path exists — use as vault_root, read tracking-database.json.
2. Path missing — first-time setup: ask preferred location via AskUserQuestion, create directory (and symlink if custom path chosen), initialize empty tracking-database.json with empty config, talks, pptx_catalog.

Config bootstrapping — ask once per missing field and persist to the tracking database. Core fields: shownotes (enabled, source.type, source.path_or_url, source.talks_subdir, url.base, url.template, thumbnail_path_template, slug_convention), pptx_source_dir, python_path, template_skip_patterns. See references/schemas-db.md for the full schema and ../vault-profile/references/schemas-config.md for field-by-field semantics and migration notes.

Scan for new talks: Build the talks directory path as {shownotes.source.path_or_url}/{shownotes.source.talks_subdir}; glob *.md there; parse and add any file not yet in talks[] (title, conference, date, URLs, status "pending"). For remote_url or none source types, skip the scan — the vault ingests only the talks the speaker has already registered elsewhere. Extract video_url, slides_url from frontmatter/links. Parse IDs from URLs:

• youtube_id: extract the v= parameter from the YouTube URL
• google_drive_id: extract the file ID from the Google Drive URL

Default status is always "pending" for new entries.

Scan for .pptx files: Recursively glob **/*.pptx in pptx_source_dir; fuzzy-match to talks[] entries. Report counts. See references/schemas-db.md for the PPTX extraction output schema (per-slide visual data, shape types, global design stats). Run scripts/pptx-extraction.py for extraction.

Pattern taxonomy migration: See references/processing-rules.md for migration logic. In brief: talks with status "processed" or "processed_partial" that lack pattern_observations are marked "needs-reprocessing".

Read rhetoric-style-summary.md and slide-design-spec.md. Report: "X processed, Y remaining. PPTX: A cataloged, B matched, C extracted."

Step 2 — Select Talks to Process

• Select talks with status pending or needs-reprocessing.
• Set slide_source per the hierarchy above. Mark "skipped_no_sources" only if video_url is entirely absent — a talk with no video_url is not processable regardless of whether slides exist.
• If $ARGUMENTS specifies a talk filename or title, process ONLY that one.

Step 3 — Process Talks via Parallel Subagents (Batches of 5)

Per batch: launch 5 subagents in parallel, wait, run Step 4 (Persist Subagent Results), then run Step 5 (Update Rhetoric Summary), then move to the next batch. When all batches have finished, proceed to Step 6.

Each subagent receives the talk's DB entry and current rhetoric-style-summary.md, runs A → B → B2 → C, and returns a JSON payload. Full procedure — transcript download (YouTube auto-subs → youtube-transcript-api → Whisper fallback chain), slide acquisition per slide_source, rhetoric/style analysis, pattern-taxonomy tagging, and the return-JSON shape — lives in references/subagent-instructions.md.

Step 4 — Persist Subagent Results

Runs after each batch inside Step 3's loop (not as a separate post-loop phase). Mechanical persistence of the batch's subagent JSON returns:

• Update tracking DB — deterministic merge, NOT hand-mapping. Collect the batch's subagent JSON returns into an array file (batch-returns.json) and run scripts/persist-results.py {vault_root}/tracking-database.json batch-returns.json. The script merges each return into its matching talk entry, promotes the declared queryable scalars to the talk top level, and rewrites the DB in place; it prints a JSON merge summary to stdout and exits non-zero if a return's filename matches no talk. Do NOT hand-copy fields one at a time — that is what dropped structured data before (it was computed and reached the analysis files but never landed in the DB). Contract, the promoted-scalar allowlist, and merge semantics live in scripts/persist-results.py (top-of-file docstring and the PROMOTE list) — to make a new field queryable, extend the return schema and that list; never reintroduce manual mapping.
• Write per-talk analysis files — write {vault_root}/analyses/{talk_filename}.md for each processed talk: all 14 dimensions, structured data, verbatim examples, and a "Presentation Patterns Scoring" section. Create analyses/ directory if missing.

Proceed immediately to Step 5.

Step 5 — Update Rhetoric Summary

Still per-batch (continues Step 3's loop). The summary update is a separate step from Step 4's persistence because it requires a speaker-review gate — unlike DB writes, edits to rhetoric-style-summary.md change the speaker's ground-truth narrative and must not be applied silently.

1. Speaker-review gate. Present the subagent's proposed summary_updates and new_patterns as a section-by-section diff and wait for explicit speaker confirmation. Silent application erodes the speaker's sense of ownership of their own style summary; pattern-taxonomy additions in particular drift if applied unreviewed. Only bypass the gate if the speaker pre-authorized this batch ("just apply everything, don't ask").
2. Apply approved changes. Integrate confirmed new_patterns and summary_updates into rhetoric-style-summary.md. Sections 1–14 map to the 14 dimensions; Sections 15–16 are the cross-talk improvement & adherence baseline and speaker-confirmed intent — structure defined in references/processing-rules.md Rhetoric Summary — Improvement & Adherence Sections. Recount status from the DB every time — never increment manually.
3. Report. Output: talks processed, new patterns, current state, skipped talks. Flag structural changes prominently (new presentation mode, new workflow pattern).

When Step 3's batch loop finishes, proceed to Step 6.

Step 6 — Extract Remaining PPTX Visual Data

Runs once after all Step 3 batches have completed.

Process PPTX files not yet extracted during Step 3: unmatched catalog entries, talks that used PDF as primary but have a PPTX available, or entries with pptx_visual_status: "pending". Skip if already "extracted". Run scripts/pptx-extraction.py <path.pptx> for each file.

PPTX matching rules: The .pptx files are in Conference/Year/TalkName.pptx and shownotes entries have conference and title fields. Fuzzy-match by: normalize conference names (strip year, "Days", "Conference"), match by date proximity and title substring. Skip files with "static" in name, conflict copies matching (N).pptx, and files matching config.template_skip_patterns. Some talks have multiple .pptx files (one per delivery) — match to the closest date.

After 3+ extractions, populate slide-design-spec.md; after 5+, analyze cross-talk patterns (colors, fonts, footers).

Proceed immediately to Step 7.

Step 7 — Regenerate Speaker Profile

If {vault_root}/speaker-profile.json exists, invoke Skill(skill: "vault-profile") with the updated tracking database. Report the diff of changes (added fields, changed values) so the speaker can verify.

If the profile doesn't exist, skip this step silently.

Proceed immediately to Step 8.

Step 8 — Verify Improvement Goals

If the tracking DB has no improvement_goals in a verifiable state (none whose status is outside achieved/retired), skip this step silently. Otherwise, with the final Section 15 baseline now current, verify each such goal: compute its current_value, set status (achieved|improving|stalled|regressed), and write the verification fields back — full rubric in references/processing-rules.md Improvement Goal Verification. Report each goal's status in the run summary; regressed or stalled goals are the speaker's own priorities — surface them first.

Proceed immediately to Step 9.

Step 9 — Same-Week Clarification Trigger

If no talks were newly processed in this run, finish here without further action.

Otherwise, scan the newly-processed talks for delivery date and bucket each by how long ago it was delivered (today − date). The handoff strength is tiered by recency — clarification quality decays fast, so the freshest talks get an active handoff, not a footnote. For every bucket, first compute that talk's candidate clarification topics:

• Each per-talk areas_for_improvement entry.
• Any pattern_observations the subagent flagged as unverifiable from transcript alone (low confidence, heavy reliance on visual cues, non-English dialogue without captions).

≤7 days (same-week) — hand off inline, don't just recommend. This is the freshest-possible clarification window: memory of the delivery is sharpest right after the talk, and verbal beats that didn't appear in auto-captions (bilingual jokes rendered in a non-primary language, improvised asides, fly-bys that weren't in the deck) are only recoverable now. Do NOT bury this as a closing recommendation. Use AskUserQuestion to offer to run vault-clarification right now, showing the candidate topics you computed so the speaker sees exactly what the session would cover. If they accept, invoke Skill(skill: "vault-clarification") immediately, carrying those candidate topics as the session's seed agenda. If they decline, note it and finish.

7–30 days — recommend the full session. Recommend running Skill(skill: "vault-clarification"), listing the candidate topics, but note that some verbatim details may already be lost. Do not auto-invoke.

30+ days — recommend the compressed session. Memory has decayed and detailed recall is unreliable; recommend the compressed clarification instead of the full one. Do not auto-invoke.

Error Handling

Important Notes

• Create transcripts/, slides/, analyses/ dirs if missing.
• Re-read tracking DB before writing (single source of truth).
• Preserve all summary content — add/refine, never delete.
• After 10+ scored talks, produce per-talk adherence assessments against the Section 15 baseline — definition in references/processing-rules.md Adherence Assessment.

For input-quality edge cases that require non-default handling — wide-angle room recordings, Whisper hallucination on bad audio, non-speaker talks slipping into playlists — see references/known-issues.md.