lecture-ftg

name	lecture-ftg
description	Fill gaps in a lecture-notetaker .md using the corresponding slide PDF — detect weak explanations, research enhancements, compress to minimum viable insertions, and add cited callouts. Runs end-to-end through all phases with no approval gates once invoked.
disable-model-invocation	true

Lecture FTG (Fill The Gaps)

Invocation

Manual only. Apply this skill only when the user invokes /lecture-ftg or explicitly selects it. Do not load or run it based on conversation context alone.

Identify weak points in a lecture .md, research enhancements, compress to minimum viable insertion, deliver enhanced .md. Auto-proceeds end-to-end — no approval gates.

Before starting

1. Read /mnt/skills/public/pdf-reading/SKILL.md for slide handling.
2. Read all four references below — each governs one workflow phase.

Inputs

Input	Format	Required?
Lecture note	1 .md from lecture-notetaker	Yes
Slides	.pdf of the lecture	Yes

Multiple .md files: run the skill once per file. Gap detection is local to a single lecture.

Workflow (auto-proceed end-to-end)

Phase	What happens	Reference
1. Identify	Run gap detector across .md + slides; produce internal flag list	references/gap-detector.md
2. Plan	Pick category for each flag; research with web_search; collect candidates with sources	references/enhancement-palette.md
3. Compress	Apply four-clause-cut test, then drop test, to each candidate	references/compression-rules.md
4. Insert	Add surviving enhancements as [!info]+ 🤖 callouts at appropriate locations	references/callout-format.md

Phase 3 is load-bearing — there is no hard cap on enhancement count, so compression must self-limit. If after compression an enhancement is generic with no specific anchor, drop it.

Cross-check with slides — important

lecture-notetaker deliberately drops content the slide already covers. If the .md is thin in a section but the slide is rich, the gap is intentional. Always check the slide before flagging. Re-introducing slide-redundant content reverses the notetaker's filter and is the most common failure mode of this skill.

Output identifiability

Every enhancement is a [!info]+ 🤖 Enhancement — <category> callout with a mandatory italicised source line. Format and source-line variants in references/callout-format.md. Do not insert AI content as inline prose; do not modify source content; only add callouts.

When gaps and enhancements are sparse

A .md that's already comprehensive will produce few or no enhancements. That is correct behaviour, not a failure. Present the .md mostly unchanged and note in delivery.

References

File	When to read	Governs
references/gap-detector.md	Phase 1	Six gap signals, anti-patterns, user-disposition defaults, edge cases
references/enhancement-palette.md	Phase 2	Eight categories, when each fits, decision rule for overlapping fits
references/compression-rules.md	Phase 3	Four-clause-cut test, drop test, worked compression cycle
references/callout-format.md	Phase 4	Exact callout format, sourcing rule variants, insertion location