Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/henryhyw/slidecraftnpx agentmods add skills/henryhyw/slidecraft/slidepoiseWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/henryhyw/slidecraft/slidepoise)<a href="https://agentmods.dev/skills/henryhyw/slidecraft/slidepoise"><img src="https://agentmods.dev/badge/skills/henryhyw/slidecraft/slidepoise/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/henryhyw/slidecraft/slidepoise"><img src="https://agentmods.dev/badge/skills/henryhyw/slidecraft/slidepoise.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00103 | $0.05917 |
| Opus 5 | $0.00051 | $0.02959 |
| Sonnet 5 | $0.00021 | $0.01183 |
| Haiku 4.5 | $0.00010 | $0.00592 |
Grade A, and why
slidepoise 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 9d 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 — 209 lines — stays where its author put it; the contents beside it link to each section on GitHub.
SlidePoise
Act as the host Agent for one information-bearing slide. Own interpretation, communication design, resource choice, image-generation orchestration, every visual judgement, semantic mapping, peer/alignment reasoning, reconstruction intent, review, and acceptance. Use scripts only to collect deterministic evidence or perform deterministic construction, including measurement, geometry, text fitting, file/library operations, asset-sheet construction, raster extraction, rendering, and objective contract facts. Scripts never issue an overall stage verdict.
Absolute authority rule
If answering a question requires looking at the slide, look at the slide and reason visually. Never replace that judgement with a score, confidence threshold, object count, whitespace metric, file-exists check, or heuristic pass/fail.
OpenCV is mandatory pixel evidence. The Agent says what an object is, which region it owns, and which measured geometry needs correction. Code executes explicit decisions without discovering visual peers or choosing alignments.
Read before working
At the start of every run:
- Resolve the SlidePoise CLI once. Use
slidepoisewhen it is onPATH. After the one-line GitHub setup, use~/.slidepoise/python/bin/slidepoiseon macOS or Linux and%USERPROFILE%\.slidepoise\python\Scripts\slidepoise.exeon Windows. RespectSLIDEPOISE_HOMEwhen it is set. Rundoctorandprofile showwith that CLI to locate the framework home and selected external profile. Use the same resolved command for every laterslidepoiseexample in this skill. - Read the external framework config, selected profile, and its selected Library Sets. Profiles, visual references, icons, and components never live inside this skill.
Read
references/profile-authoring.mdwhen the user wants to create, modify, understand, or curate a guidance profile or its persistent resources. - Resolve current-session overrides with
scripts/resolve_config.py, passing the framework config and profiles root. - Read
references/workflow.md. - Read
references/runtime-host.mdfor ChatGPT/Codex image-generation/tool adaptation. - Read
references/human-approval.mdand enforce the three hard user gates. - Read
references/resource-library.mdbefore resource selection. - Before semantic mapping/reconstruction, read
references/visual-reasoning.md,references/connectors.md, andreferences/reconstruction.md. - For raster artwork, intrinsic lettering, texture, or overlapping text, read
references/raster-composition.md. Readreferences/illustration-refinement.mdbefore considering an optional raster edit. - Read
references/case-studies.mdwhen preserving a reusable example or preparing project showcase material. Save exact image prompts and their inputs/outputs during the run so a later case study can use real evidence. - Read
references/multi-agent-review.mdwhen the host exposes subagents. Use its bounded, read-only reviews at the configured checkpoints. - Read
references/user-language.mdbefore authoring user-facing prose, slide copy, Panel summaries, activity messages, or approval questions.
What ships with it
60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 312 B
- assets/icon.svg 1.2 KB
- references/case-studies.md 2.9 KB
- references/connectors.md 3.7 KB
- references/human-approval.md 4.3 KB
- references/icon-variants.md 2.5 KB
- references/illustration-refinement.md 2.3 KB
- references/maintenance.md 2.3 KB
- references/multi-agent-review.md 4.1 KB
- references/profile-authoring.md 4.4 KB
- references/raster-composition.md 6.4 KB
- references/reconstruction.md 11 KB
- references/resource-library.md 6.3 KB
- references/runtime-host.md 2.0 KB
- references/session-panel.md 11 KB
- references/user-language.md 2.7 KB
- references/visual-reasoning.md 7.0 KB
- references/workflow.md 11 KB
- runtime/js/scene_to_pptx.mjs 35 KB runs code
- runtime/scripts/audit_powerpoint_text.py 6.6 KB runs code
- runtime/scripts/measure_visual_scene.py 22 KB runs code
- runtime/scripts/postprocess_pptx.py 6.2 KB runs code
- runtime/scripts/sam_optional.py 5.1 KB runs code
- runtime/src/slidepoise/__init__.py 98 B runs code
- runtime/src/slidepoise/reconstruction/__init__.py 0 B runs code
- runtime/src/slidepoise/reconstruction/conformance.py 20 KB runs code
- runtime/src/slidepoise/reconstruction/contract.py 12 KB runs code
- runtime/src/slidepoise/reconstruction/scene.py 43 KB runs code
- runtime/src/slidepoise/reconstruction/text_fit.py 24 KB runs code
- schemas/agent-review.example.json 753 B
- schemas/generation-contract.example.json 4.0 KB
- schemas/human-approvals.example.json 772 B
- schemas/panel.example.json 336 B
- schemas/reconstruction-handoff.example.json 1.5 KB
- schemas/resource-selection.example.json 2.1 KB
- schemas/semantic-map.example.json 6.5 KB
- schemas/slide-intent.example.json 775 B
- schemas/visual-review.example.json 680 B
- scripts/apply_illustration_sources.py 1.7 KB runs code
- scripts/approval_utils.py 1.0 KB runs code
- scripts/audit_skill_boundaries.py 1.7 KB runs code
- scripts/check_approval.py 672 B runs code
- scripts/check_illustration_gate.py 874 B runs code
- scripts/collect_reconstruction_evidence.py 3.6 KB runs code
- scripts/collect_release_evidence.py 13 KB runs code
- scripts/collect_semantic_evidence.py 36 KB runs code
- scripts/component_preview.py 1.6 KB runs code
- scripts/extract_generation_region.py 2.1 KB runs code
- scripts/extract_refined_illustrations.py 1.4 KB runs code
- scripts/fetch_remix_icon_pair.py 6.0 KB runs code
- scripts/fetch_remote_asset.py 2.1 KB runs code
- scripts/fetch_wikimedia_commons_asset.py 5.0 KB runs code
- scripts/inspect_asset.py 2.4 KB runs code
- scripts/list_library.py 4.2 KB runs code
- scripts/make_asset_contact_sheet.py 14 KB runs code
- scripts/make_semantic_explanation_overlay.py 5.5 KB runs code
- scripts/make_visual_comparison.py 2.6 KB runs code
- scripts/manage_library.py 3.7 KB runs code
- scripts/normalize_generation_canvas.py 2.1 KB runs code
- scripts/package_self.py 1.6 KB runs code
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.
- 9d ago First seen · 209 lines · 103 tokens per session scan A 16709d1fc1f6
slidepoise is a skill published in the GitHub repository henryhyw/slidecraft (0 stars, last pushed 2d ago), licensed MIT. It adds 103 tokens to every session and 5,917 once invoked, about $0.0005 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 skills, from other repositories
themed-cn-pptx
Build, modify, and QA-verify Chinese + IP/character-themed + QR-embeddable EDITABLE PPTX decks using PptxGenJS, with SOTA AI image generation via OpenAI GPT Image 2 or Google Nano Banana Pro, locked aesthetic recipes, and a deterministic render-QA gate (CJK overflow, text overlap, editable-text, color contrast).…
imagegen-pptx-pipeline
End-to-end stateful workflow for designing and generating editable PowerPoint decks from briefs, templates, reference decks, brand assets, data, ImageGen comps, or user-supplied slide images. Uses multi-round content/style confirmation, reviewed per-slide comps, mandatory Real-ESRGAN 4K comp/icon upscaling, built-in…
consulting-deck
Create, revise, and audit executive-ready strategy-consulting presentations as native, editable PowerPoint files. Use when turning topics, text, Markdown, Word, PDF, spreadsheets, CSV data, URLs, research, or an existing PPTX into a board, client, strategy, transformation, market-analysis, or McKinsey-style consulting…
2slides-ppt-generator
AI-powered presentation generation via the 2slides API — create slides from text, match a reference image style, summarize documents into decks, add AI voice narration, and export pages/audio. Use for any "make slides", "create a deck", or "slides from this document" request.
2slides-ppt-generator
AI-powered presentation generation via the 2slides API — create slides from text, match a reference image style, summarize documents into decks, add AI voice narration, and export pages/audio. Use for any "make slides", "create a deck", or "slides from this document" request.
2slides-ppt-generator
AI-powered presentation generation via the 2slides API — create slides from text, match a reference image style, summarize documents into decks, add AI voice narration, and export pages/audio. Use for any "make slides", "create a deck", or "slides from this document" request.