narrative-transitions
Skill1 starsChanged 11 days agoApache-2.0View on GitHub
Make every section boundary in a talk, narrated video, demo, slide deck, tutorial or written explainer follow from what came before, choosing the transition by medium and auditing all boundaries together.
In this repository24 other files it gives its agents.
- .github/copilot-instructions.mdCopilot instructions
- ab-testingSkill
- artifact-inspectionSkill
- comparative-evaluationSkill
- cue-synced-demoSkill
- design-analysisSkill
- design-writingSkill
- evidence-verificationSkill
- intent-alignmentSkill
- interactive-experienceSkill
- pr-description-writingSkill
- presenter-scriptSkill
- promptlings-memory-captureSkill
- pr-videoSkill
- reasoning-first-narrationSkill
- reasoning-first-writingSkill
- review-comment-writingSkill
- review-ensembleSkill
- spec-deliverySkill
- technical-demo-productionSkill
- technical-illustrationSkill
- technical-investigationSkill
- test-analysisSkill
- tts-setupSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 37
Within code: keep one focal line or block at a time and narrate it while it is highlighted.
- Distinctive: line 60
When one transition is corrected, re-check every boundary before presenting the revision.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Narrative Transitions
Use this skill when a piece is divided into sections, scenes, slides or steps and must still play or read as one line of thought. It governs the joins. Claims, evidence and sentence-level wording belong to the writing skill in use, such as reasoning-first-narration. A transition never adds a claim, consequence or contrast that the sources do not support.
Link by cause and known-to-new
A boundary works when the opening of the next section starts from something the audience already holds and moves to what is new. End a section on the idea the next one picks up, then open the next section from that idea. A recap of the whole section is not a link.
Connect sections by reasoning, not sequence. "And then" between sections produces a list. Prefer "therefore" (this follows from what we just saw) or "but" (what we just saw runs into a problem or limit), and use "but" only where the sources establish the tension. Every section should still serve the piece's one controlling idea; a transition that cannot say how the next section serves it usually signals a structural problem, not a wording problem.
An open question can carry attention across a boundary when a later section answers it. A question the piece never answers is a teaser, not a transition.
Decide who carries the turn
The medium decides how much the words must do. Decide separately at each boundary, because mixed pieces change medium between sections.
| Medium | What carries the turn |
| Live talk or voice-only narration | The words alone. Every boundary needs a spoken bridge, because the audience never sees a heading. If a listener would be lost without the heading, the bridge is missing. |
| Narrated video | The cut and the voice together. Let the voice lead into the next picture (a J-cut) when it introduces a new idea; let it trail over the new picture (an L-cut) to finish the previous thought. When the new picture already makes the change obvious, keep narration minimal and explain what the picture cannot show. |
| Hands-on demo or screencast | The action. Narrate in step with what is on screen, neither ahead nor behind. Let visual cues such as pointer movement, an in-place highlight or the next window carry "we have moved on", and open with the result just shown or the next step the viewer is about to watch. Keep cut and fade styles few and consistent. |
| Slide deck with a presenter | The presenter. Treat each slide title as a claim to support, not a topic to announce; do not read titles aloud. Give one spoken bridge from the previous claim to the next. |
| Interactive experience with a presenter | The presenter. Bridge from the question the last exploration answered, or left open, to the question the next stop explores; say it before touching the next stop's controls. |
| Written explainer or documentation | Headings navigate; prose links. Do not stack headings without text between them, and do not fill the gap with boilerplate such as "In this section we will". Opening sentences connect the section to the reader's need. |
| Step-by-step tutorial | The sequence. Imperative steps need almost no connective text. Put connective prose at phase boundaries and state the outcome the learner should now see before moving on. |
Cross into and out of code
A code walkthrough inside a demo or video is a change of medium, so its boundaries need their own care.
- Into code: start from the behavior or claim just shown and name where it lives, for example "That rejection happens in the installer's name check." Establish the file and function before the detail using the director's selected surface. In real editor footage, meaningful visible navigation can carry the move. In source-derived presentation, retain a file/function label and reveal the relevant excerpt. Neither route requires unrelated file-tree travel.
- Within code: keep one focal line or block at a time and narrate it while it is highlighted. Name each new file or function as the view changes to it, and pause briefly after the change.
- Out of code: carry the mechanism back to its consequence, either the behavior it produces next or the next question it raises, rather than leaving the viewer in the editor. A recap of the code just shown is not a link.
The first section has no earlier section to link from. Open it with a hook that passes the opening hook check in reasoning-first-narration.
Avoid these openings
- A read-out title or heading.
- The section's thesis compressed into a headline or slogan.
- A callback aphorism to earlier framing that the audience must decode.
- An agenda line such as "Next, I will talk about".
- A recap that restates the previous section instead of carrying one idea forward.
A plain signpost is acceptable when the structure itself helps, for example "Let's start with the review lane." Plain beats clever.
Audit the whole piece
Draft and review as one continuous experience, not section by section.
- Write a one-sentence summary of each section in order, stating its idea and how it serves the controlling idea. Gaps, repeats and drift show up as jumps in that list. In an explanatory piece, a run of summaries shaped as "X does Y" is a list joined by bridges, a structural problem rather than a transition to reword.
- At every boundary, compare the last two sentences before it with the first two after it. The new idea at the end should become the known starting point of the next opening.
- Experience the piece in its medium: read spoken scripts aloud with headings removed, review video narration against the planned picture at each cut, and read written pieces straight through.
- Fix every join that fails the medium's test, except joins inside sections with structural findings. Report those findings to the caller or the structure's owner and leave those sections unchanged.
- When one transition is corrected, re-check every boundary before presenting the revision. A defect at one join usually exists at others.
Repeat the audit after material revisions, reordered sections or changed visuals.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

