| name | profile-intro-slides |
| description | Generate a reusable 2-3 slide Yue Zhao / FORTIS Lab intro insert from local website-first sources in this repo. Use when the user wants a stable identity, agenda, and representative-systems module that can be merged into other decks. Do not use this skill for full paper talks, mini-CV decks, or automatic deck insertion. |
Profile Intro Slides
Read references/source-selection.md first.
Read references/slide-recipes.md before writing any frames.
Read references/quality-checks.md before finalizing any draft.
This skill generates a small reusable slide family about Yue Zhao and FORTIS Lab. It is for a stable intro insert, not for a full opening sequence, not for a recruiting deck, not for a mini CV, and not for inserting slides into another talk automatically.
Workspace Contract
This skill is repo-specific. Treat these paths as relative to the workspace root:
README.md
theme/
templates/
yzhao062.github.io/
Before generating anything, verify that the workspace contains at least:
README.md
templates/demo.tex
templates/paper-deck-base.tex
theme/beamerthemeuscmoloch.sty
yzhao062.github.io/index.html
yzhao062.github.io/lab.html
yzhao062.github.io/files/bio.txt
yzhao062.github.io/data/open-source.json
yzhao062.github.io/data/publications.json
Useful optional sources:
yzhao062.github.io/data/lab-current-phd.json
yzhao062.github.io/data/lab-members.json
yzhao062.github.io/files/ZHAO_YUE_CV.pdf
yzhao062.github.io/images/rsz_250.jpg
yzhao062.github.io/images/lab-logo.png
If the contract is not satisfied, stop and ask the user instead of improvising a different source stack.
Default Outputs
Generate these artifacts by default:
templates/modules/yzhao-intro-slides.tex
templates/modules/yzhao-intro-facts.md
yzhao-intro-slides.tex should contain reusable frame blocks only. Do not include a document preamble, title page, section pages, bibliography appendix, or \input{...} edits to another deck.
yzhao-intro-facts.md should be the canonical fact sheet that records the stable positioning, representative systems, chosen impact signals, and the source file for each nontrivial claim or number.
Only create a standalone preview deck or compile output if the user explicitly asks for a preview.
Workflow
- Verify the workspace contract and read the source-selection reference.
- Extract stable facts from website-first sources. Use the CV only to fill gaps such as awards, service, or older chronology.
- Separate evergreen content from volatile content. Unless the user asks otherwise, omit recruiting details, news, collaboration logistics, live policy notices, and other time-sensitive site blocks.
- Build or refresh
templates/modules/yzhao-intro-facts.md first. Record:
- identity and affiliation
- one-line FORTIS expansion and mission
- the three main research directions
- 3-5 representative systems, projects, or agenda-anchor papers
- a small set of durable impact signals
- any selected sponsor or research-support logos that may appear in a secondary acknowledgment strip
- the preferred portrait and lab logo assets
- source path for each nontrivial claim or number
- Choose the slide family. Default to exactly three slides with fixed role separation:
About Yue Zhao
FORTIS Lab Research Agenda
Representative Systems and Impact
- Before drafting, assign each slide:
- exactly one role
- exactly one dominant mode:
text-first, visual-first, or compact card grid
- exactly one primary object of attention
- Write
templates/modules/yzhao-intro-slides.tex as reusable frame blocks only.
- Keep the canonical role split:
About Yue Zhao: identity and positioning
FORTIS Lab Research Agenda: agenda map
Representative Systems and Impact: execution evidence
- Do not let one slide do two or three jobs at once.
- Run the checks in
references/quality-checks.md before preview generation or handoff.
- If any density or role check fails, revise before finalizing.
- Apply slide surgery in the low-risk order defined there before attempting a larger redesign.
- Keep the slides mergeable into arbitrary decks: no talk-specific transitions, no venue-specific filler, and no dependency on surrounding slide numbers, sections, or presenter notes.
- If the user wants a variant emphasis, keep the canonical core and vary only one axis at a time: auditing and assurance, agent safety and security, anomaly-detection roots, or lab-intro flavor.
- If the website and CV disagree, prefer the more recent website wording unless the CV clearly contains the only reliable version. Note the conflict in the fact sheet.
Operating Rules
- Treat the website as the primary source of current positioning.
- Prefer stable positioning over exhaustive biography.
- Keep the module a reusable insert, not a mini-talk.
- Distinguish personal identity, agenda mapping, and execution evidence; do not collapse them into one overloaded frame by default.
- Keep the default family to exactly
3 slides unless the user explicitly asks for a shorter variant.
- Each slide must have exactly one main job, one dominant mode, and one primary object of attention.
- Prefer not to put paper-count, download, or GitHub-star callouts on the personal identity slide unless the user explicitly asks for a metrics-forward version.
- Prefer one memorable phrase and three strong pillars over long bullet lists.
- Prefer
3-4 representative systems on a systems slide. Do not try to summarize every project.
- On the
FORTIS Lab Research Agenda slide, prefer mixed examples under each pillar: named systems when tooling is the clearest anchor, and representative papers when they better communicate the current research direction.
- When a systems slide is likely to be shared as PDF, prefer clickable project names or project-title links that can route viewers to the corresponding site or repo.
- For open-source projects, default those links to the GitHub repository itself unless the user explicitly prefers docs or a project site.
- If project names are clickable, prefer subtle visible styling such as underlining rather than noisy explicit
[link] labels.
- Keep Slide 1 qualitative and slide-native: one positioning sentence max, three short bullets max, no biography prose, no metrics strip.
- If Slide 1 reads cleaner with labeled signal lines than with plain bullets, prefer short labels such as lab, team, and style.
- If Slide 1 needs a lab-scale signal, prefer the stable current/incoming PhD count and describe the broader master's or undergraduate team qualitatively rather than fixing a volatile total.
- Team-scale wording on Slide 1 should read as appreciation and research capacity, not as a roster dump or recruiting pitch.
- Keep Slide 2 as a clean pillar map: one short mission line at top, three equal pillars, one short description line plus one compact examples line per pillar.
- If Slide 2 feels visually underused, prefer a subtle relation visual such as
Observe -> Control -> Deploy across the three pillars before adding more prose.
- Keep Slide 3 as a compact project-card slide by default:
3-4 anchors, one-line payoffs, and only an optional small dated impact strip if it helps.
- If sponsor acknowledgment is desired, keep it as a thin secondary support strip on the systems slide rather than a separate sponsor slide or sponsor wall.
- If both metrics and sponsor logos are present on Slide 3, keep the project cards primary, the metrics strip secondary, and the sponsor strip quieter still.
- Never put both a long takeaway paragraph and a metrics strip on the systems slide.
- Treat volatile metrics as optional, dated, and visually subordinate. If they add clutter, remove them.
- Avoid full publication lists, full awards lists, full student rosters, or sponsor walls unless the user asks for them.
- Keep visible slide text concise, broad-CS readable, and slide-native rather than webpage-like.
- If you include numbers that can drift, record the exact source and date in
yzhao-intro-facts.md. Use an absolute date in visible text only when the date matters for interpretation.
- Preserve USC styling by generating content-only frame blocks. If a preview deck is requested, inherit styling from
templates/paper-deck-base.tex or templates/demo.tex instead of editing theme/*.sty.
- This skill generates reusable intro modules only. It does not insert them into existing decks, update presenter maps, or reorganize another talk around them.