Five-skill presentation system: ingest talks into a rhetoric vault, run interactive clarification, generate a speaker profile, create presentations that match your documented patterns, and produce the deck illustrations + thumbnail visual layer. Includes a 102-entry Presentation Patterns taxonomy (91 observable, 11 unobservable go-live items) for scoring, brainstorming, and go-live preparation.
93
95%
Does it follow best practices?
Impact
93%
1.20xAverage score across 34 eval scenarios
Advisory
Suggest reviewing before use
{
"context": "Tests the negative-case behavior of the SKILL.md Step 2 freshness check: when the outline's `**Model:**` line was set against a current flagship, the check should be entered AND should pass without inserting friction. Baseline agents (no tile) skip the freshness check entirely, so 'web-search precedes generation' is the tile-specific lift here; the remaining criteria are defensive — they catch regressions where the tile starts inserting friction where none was warranted.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Web-search precedes generation",
"description": "The agent calls WebSearch (or an equivalent live-web lookup) to identify current flagship image-generation models before invoking generate-illustrations.py. This is the tile-specific behavior — without the skill loaded, an agent would jump straight to generation. A run that skips the live-web lookup fails this criterion.",
"max_score": 32
},
{
"name": "No spurious model-change proposal",
"description": "The agent does NOT propose editing `skills/illustrations/scripts/generate-illustrations.py` to change `COMPARE_MODELS`, and does NOT propose changing the outline's `**Model:**` line, when the web-search outcome shows the outline's model is among current flagships. Surfacing such a proposal anyway fails this criterion.",
"max_score": 27
},
{
"name": "Proceeds to generation without re-litigating",
"description": "After the freshness check completes, the agent moves on to invoke `generate-illustrations.py` (or otherwise produce slide images) as the next concrete action. Halting at the freshness check to ask the speaker to reconfirm a model choice that has no surfaced gap fails this criterion.",
"max_score": 23
},
{
"name": "Generation uses the outline's declared model",
"description": "If image generation is invoked, the command uses the model declared in the outline header — not a different model picked or substituted by the agent.",
"max_score": 18
}
]
}evals
scenario-1
scenario-2
scenario-3
scenario-4
scenario-5
scenario-6
scenario-7
scenario-8
scenario-9
scenario-10
scenario-11
scenario-12
scenario-13
scenario-14
scenario-15
scenario-16
scenario-17
scenario-18
scenario-19
scenario-20
scenario-21
scenario-22
scenario-23
scenario-24
scenario-25
scenario-26
scenario-27
scenario-28
scenario-29
scenario-30
scenario-31
scenario-32
scenario-33
scenario-34
rules
skills
illustrations
presentation-creator
references
patterns
build
deliver
prepare
scripts
vault-clarification
vault-ingress
vault-profile