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 instructions/whatsonyourmind/deckforge/claude-mdgit clone --depth 1 https://github.com/Whatsonyourmind/deckforgeWrote 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/instructions/whatsonyourmind/deckforge/claude-md)<a href="https://agentmods.dev/instructions/whatsonyourmind/deckforge/claude-md"><img src="https://agentmods.dev/badge/instructions/whatsonyourmind/deckforge/claude-md.svg" alt="Measured on agentmods" 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 | $0.02210 | $0.02210 |
| Opus 5 | $0.01105 | $0.01105 |
| Sonnet 5 | $0.00442 | $0.00442 |
| Haiku 4.5 | $0.00221 | $0.00221 |
Grade A, and why
deckforge CLAUDE.md 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 4d 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 — 147 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DeckForge
API-first presentation generation platform. 32 slide types, 24 chart types, 15 themes, native PPTX + Google Slides output, AI content generation, finance vertical with 9 specialized slide types.
Tech Stack
- API: Python 3.12, FastAPI, Pydantic v2, uvicorn
- Database: PostgreSQL (prod) / SQLite (dev), SQLAlchemy 2.0 (async), Alembic migrations
- Queue: Redis + ARQ (async workers for content generation and rendering)
- Storage: MinIO (dev) / Cloudflare R2 (prod), S3-compatible via boto3
- Rendering: python-pptx (PPTX), google-api-python-client (Google Slides), Plotly + Kaleido (static chart PNGs)
- SDK: TypeScript, tsup build, vitest tests (in
sdk/directory) - Billing: Stripe (subscriptions, 3 tiers), x402 (USDC per-call for AI agents on Base L2)
- Auth: Unkey (prod) / DB-based SHA-256 key hashing (dev), API key prefixes
dk_test_/dk_live_ - Layout: kiwisolver constraint solver, 12-column grid, Pillow text measurement
- Themes: 15 YAML definitions in
src/deckforge/themes/data/ - Content: Model-agnostic LLM orchestration (Claude, OpenAI, Gemini, Ollama) with 4-stage pipeline
- MCP: FastMCP server with 6 tools for AI agent integration
Project Structure
src/deckforge/
config.py # Pydantic Settings, DECKFORGE_ env prefix, all config fields
main.py # App entrypoint, lifespan, uvicorn runner
ir/ # Pydantic IR models
slides/ # 32 slide types (23 universal + 9 finance), discriminated on slide_type
elements/ # Element types (title, text, metric, chart, image, table, etc.)
charts/ # 24 chart types, discriminated on chart_type
api/ # FastAPI routes + middleware
routes/ # 15 route modules: render, generate, health, billing, onboarding, etc.
middleware/ # Auth (Unkey/DB), rate limiting (Redis token bucket), credit check, x402
db/ # Database layer
base.py # SQLAlchemy Base with naming conventions
engine.py # Async engine, session factory, get_db() dependency
models/ # 8 models: User, ApiKey, Job, Deck, BatchJob, UsageRecord, PaymentEvent, WebhookRegistration
repositories/ # Repository pattern, singleton instances, session-as-argument
rendering/ # Output renderers
pptx/ # PPTX renderer with ELEMENT_RENDERERS registry, transitions, Open XML
gslides/ # Google Slides renderer with ELEMENT_BUILDERS dispatch, batchUpdate
charts/ # Static chart renderers (Plotly -> PNG at 300 DPI), 24 chart types
thumbnails/ # 3-tier fallback: LibreOffice -> pdf2image -> Pillow placeholder
layout/ # Constraint-based layout
solver.py # kiwisolver constraint system
grid.py # 12-column grid
patterns/ # PATTERN_REGISTRY: 9 layout patterns + GenericPattern fallback
overflow.py # Adaptive: font reduce (2pt steps) -> reflow -> slide split
themes/ # Theme system
data/ # 15 YAML themes (arctic-clean, corporate-blue, finance-pro, etc.)
resolver.py # 3-tier resolution: colors -> palette -> slide_masters
brand_kit.py # BrandKitMerger (protected keys: spacing, typography.scale)
content/ # NL-to-IR pipeline
pipeline.py # 4-stage: intent -> outline -> expand -> refine
intent.py # Prompt classification
outline.py # Slide structure planning
expand.py # Content generation per slide
refine.py # QA and polish
llm/ # LLM provider adapters
router.py # Provider selection + fallback chain
adapters/ # Claude (tool_use), OpenAI (json_schema), Gemini (response_schema), Ollama (retry)
charts/ # Chart recommender (data analysis -> chart type suggestion)
finance/ # Financial formatting, conditional formatting, data ingestion
qa/ # 5-pass QA pipeline, auto-fix engine (contrast, overflow, alignment), scorer
billing/ # Stripe client, credit reservation/deduction, tier definitions, x402 config
services/ # Business logic: DeckOperations, SlideTypeRegistry, CostEstimator
workers/ # ARQ workers: ContentWorkerSettings, RenderWorkerSettings, task definitions
mcp/ # MCP server: 6 tools (render, generate, themes, slide_types, estimate, preview)
sdk/ # TypeScript SDK (@deckforge/sdk)
src/ # Immutable builder pattern, SSE streaming, full type coverage (32 slide + 24 chart types)
demos/ # 5 demo deck IRs (mckinsey-strategy, pe-deal-memo, startup-pitch, board-update, product-launch)
alembic/ # Database migrations (5 migrations)
scripts/ # bootstrap-db.sh (migrations + seed data)
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.
- 4d ago First seen · 147 lines · 2,210 tokens per session scan A 6f5b4a3af3cb
deckforge CLAUDE.md is an instructions file published in the GitHub repository Whatsonyourmind/deckforge (7 stars, last pushed 9d ago), licensed MIT. It adds 2,210 tokens to every session, about $0.0111 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 instructions, from other repositories
mcp-ms-office-documents AGENTS.md
Instructions for ForLegalAI/mcp-ms-office-documents, covering agents.md, project overview, architecture, key conventions and filename generation.
logic CLAUDE.md
Instructions for lemur47/logic, covering claude.md, project overview, commands, install dependencies and run api server (hot reload).
loopback-training-server CLAUDE.md
Instructions for aderaaij/loopback-training-server, covering training api, tech stack, project structure, frontend (web dashboard) and development.
emer-ai-tools AGENTS.md
AGENTS.md instructions for emercoin/emer-ai-tools, covering the problem, what this gives you, how it works, trust model and tools (mcp).
reed AGENTS.md
AGENTS.md instructions for simonives/reed, covering reed — agent context, scope, precedence, instructions and conventions.
reed CLAUDE.md
Claude Code instructions for simonives/reed, covering reed — claude context, scope, precedence, instructions and conventions.