Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/dkorobtsov/pitch-deck/auditgit clone --depth 1 https://github.com/dkorobtsov/pitch-deckWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00042 | $0.00999 |
| Opus 5 | $0.00021 | $0.00500 |
| Sonnet 5 | $0.00008 | $0.00200 |
| Haiku 4.5 | $0.00004 | $0.00100 |
Grade A, and why
audit scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 141 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Pitch Deck Audit
Run a 7-test audit battery on an existing pitch deck. Accepts Marp markdown or PDF.
Input
Read the deck file provided as argument. If PDF, read all slides. If no argument given, ask the user for the path.
Audit Battery
Run ALL 7 tests. For each, output PASS / WEAK / FAIL with specific evidence from the deck.
Test 1: Headlines Tell the Story ⭐ MOST IMPORTANT
Extract every slide headline in order. Present them as a numbered list. Then evaluate:
- Read sequentially: do they tell a complete, compelling investment story WITHOUT body text?
- Is each headline a MESSAGE not a label? ("Team" = FAIL, "Team engineered for this opportunity" = PASS)
- Does the sequence build emotional momentum? Left hook, right hook alternation?
- Does headline sequence start with biggest statement and end with compelling close?
- Could an investor read ONLY the headlines and want to take a meeting?
This is the single most important test. If headlines don't tell the story, the deck fails regardless of content quality.
Test 2: Team Placement
Evaluate whether team slide placement is optimal:
- Team early / first when: founder has personal connection to problem, industry expertise is the moat, "why us" IS the story, team credibility makes everything else believable
- Team later when: product/traction speaks louder than bios, team is strong but not the differentiator
- If founder story creates "why this person, why this problem" conviction, team should be in first 3-5 slides
- Founder origin story ("I lived this problem for X years") makes audience listen more carefully to everything after
- At early stage: 80-90% of deal is team. If team slide is buried at slide 15, that's usually wrong
Test 3: 5-Second Test (Per Slide)
For each slide: if shown for 5 seconds and removed, can viewer state the message?
- One message per slide?
- Clear visual hierarchy — eyes know where to go?
- Max 30 words (40 absolute max)?
- Text readable (>20pt) or intentionally decorative (<13pt)?
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 141 lines · 42 tokens per session scan A 3d52b0dfd541
audit is a command published in the GitHub repository dkorobtsov/pitch-deck (10 stars, last pushed 5mo ago), licensed MIT. It adds 42 tokens to every session and 999 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
wf-ads
Generate 50-100 ad copy variants from pain point research, then create renderable ad images.
vitaecontext-x
Audit or improve X/Twitter profile and posting strategy.
ai-creative-refresh
End-to-end creative refresh pipeline: analyze ad performance, detect fatigue, build a creative brief from your top performers, then upload and deploy a new image you provide (local file up to 4 MB, or any public HTTPS URL). Image generation happens outside the app — bring your own image from Photoshop, Canva…
content-repurposer
Run the content-repurposer skill.
Guided LinkedIn post workshop with copy, image, and approval.
yt
Route to the YouTube Agent Matrix orchestrator — the full pre-production pipeline.