Enforce a two-mode QA gate for PPT decks using evidence ledgers and ai-tone scoring.
npx skills add https://github.com/AllenZhou/MIRISE --skill evidence-keeper
Please help me install this Agent Skill. Skill: evidence-keeper Source: https://github.com/AllenZhou/MIRISE/tree/main/agents/ppt-pipeline/skills/evidence-keeper Command: npx skills add https://github.com/AllenZhou/MIRISE --skill evidence-keeper
PPT decks often lack a verifiable evidence trail and consistent tone; this skill provides a two-mode QA gate to enforce a 5-level evidence ledger and ai_tone scoring for slides.
Analyze the current PPT deck in your workspace to initialize the ledger and compute the ai-tone score.
High-intent search queries and answers about installing and using this skill.
You can verify PPT slide claims by maintaining an evidence ledger with tiered classifications like Verified, Single-source, Calculated, Pinned, and Unknown to ensure all deck content is auditable.
Detecting AI generated tone in presentation slides involves scanning SlideSpec and slide content against anti-ai-tone rules to compute a deck-wide ai_tone_score for automated quality control.
A slide evidence ledger is a per-page QA record that maps deck claims to source classifications. You need it when planning or reviewing PPT decks to enforce a verifiable evidence trail and consistent tone.
You can automate PPT quality assurance checks by ingesting deckplan.json, evidence_sources[], and slides.json to build per-page claims, a structured claim_tree, and a deck-wide ai_tone_score.
Handling unknown or suspicious evidence in a slide deck requires applying QA gate decisions to generate follow-on tasks, ensuring required actions are executed before slide approval.
yzhao062Ingest NSF solicitation text into a normalized requirements baseline.
toddwardGenerate branded slide decks from .pptx templates and content outlines.
lisposterConvert content into knowledge comics with storyboards, prompts, and PDFs.
genesis-agentsIdentify and verify factual statements in slide content with credibility scoring.
