@synsci/cli-darwin-x64 1.1.70 → 1.1.72

package/bin/skills/groq/SKILL.md

@@ -0,0 +1,347 @@
+---
+name: groq-inference
+description: Ultra-fast LLM inference on custom LPU hardware. OpenAI-compatible API at api.groq.com. Lowest latency in the industry (500-1000+ tok/s). Supports chat completions, vision, audio (Whisper STT + TTS), tool calling, JSON mode, and streaming. Free tier available. Inference only — no training.
+version: 1.0.0
+author: Synthetic Sciences
+license: MIT
+tags: [Inference, Groq, LPU, Low-Latency, OpenAI-Compatible, Free-Tier]
+dependencies: [groq, openai]
+---
+
+# Groq — Ultra-Fast LLM Inference
+
+Groq runs open-weight LLMs on custom LPU (Language Processing Unit) hardware, delivering the lowest inference latency in the industry — up to 1000+ tokens/sec. OpenAI-compatible API, free tier, pay-per-token pricing. Inference only, no training.
+
+## When to Use Groq
+
+**Use when:**
+- You need the fastest possible inference latency (real-time chat, agents, interactive apps)
+- You want open-weight models (Llama, Qwen, DeepSeek) without managing GPUs
+- You need an OpenAI-compatible drop-in replacement with open models
+- You want free-tier access for prototyping
+- You need fast Whisper transcription or vision with Llama 4
+
+**Don't use when:**
+- You need to fine-tune or train models (inference-only)
+- You need proprietary models (GPT-4o, Claude, Gemini)
+- You need embeddings or image generation APIs (not offered)
+
+| Feature | Groq | Together AI | Fireworks | Replicate |
+|---------|------|-------------|-----------|-----------|
+| Hardware | Custom LPU | NVIDIA GPUs | NVIDIA GPUs | NVIDIA GPUs |
+| Speed (Llama 70B) | ~280 tok/s | ~100 tok/s | ~100 tok/s | ~50 tok/s |
+| Free tier | Yes | Yes | Yes | No |
+| Training | No | Yes | Yes | Yes |
+
+## Quick Start
+
+```bash
+pip install groq
+export GROQ_API_KEY="gsk_..."   # https://console.groq.com/keys
+```
+
+```python
+from groq import Groq
+
+client = Groq()
+response = client.chat.completions.create(
+    model="llama-3.3-70b-versatile",
+    messages=[
+        {"role": "system", "content": "You are a helpful assistant."},
+        {"role": "user", "content": "Explain quantum computing in one paragraph."}
+    ],
+    temperature=0.7,
+    max_completion_tokens=512,
+)
+print(response.choices[0].message.content)
+```
+
+## Chat Completions
+
+### Streaming
+
+```python
+from groq import Groq
+
+client = Groq()
+stream = client.chat.completions.create(
+    model="llama-3.3-70b-versatile",
+    messages=[{"role": "user", "content": "Explain fast inference"}],
+    stream=True,
+)
+for chunk in stream:
+    content = chunk.choices[0].delta.content
+    if content:
+        print(content, end="")
+```
+
+### JSON Mode
+
+```python
+from groq import Groq
+import json
+
+client = Groq()
+response = client.chat.completions.create(
+    model="llama-3.3-70b-versatile",
+    messages=[
+        {"role": "system", "content": "Output valid JSON only."},
+        {"role": "user", "content": "List 3 programming languages with their paradigms."}
+    ],
+    response_format={"type": "json_object"},
+    temperature=0,
+)
+data = json.loads(response.choices[0].message.content)
+```
+
+### Tool / Function Calling
+
+```python
+from groq import Groq
+import json
+
+client = Groq()
+tools = [{
+    "type": "function",
+    "function": {
+        "name": "get_weather",
+        "description": "Get weather for a location",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "location": {"type": "string", "description": "City name"},
+                "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}
+            },
+            "required": ["location"],
+            "additionalProperties": False
+        }
+    }
+}]
+
+response = client.chat.completions.create(
+    model="llama-3.3-70b-versatile",
+    messages=[{"role": "user", "content": "Weather in San Francisco?"}],
+    tools=tools,
+    tool_choice="auto",
+)
+
+message = response.choices[0].message
+if message.tool_calls:
+    call = message.tool_calls[0]
+    args = json.loads(call.function.arguments)
+    # Execute function, then send result back as role="tool" message
+    result = {"temperature": 62, "condition": "foggy"}
+    followup = client.chat.completions.create(
+        model="llama-3.3-70b-versatile",
+        messages=[
+            {"role": "user", "content": "Weather in San Francisco?"},
+            message,
+            {"role": "tool", "tool_call_id": call.id, "content": json.dumps(result)}
+        ],
+    )
+    print(followup.choices[0].message.content)
+```
+
+## Vision
+
+Llama 4 models support image input via URL or base64. Max image size: 20 MB.
+
+```python
+from groq import Groq
+
+client = Groq()
+response = client.chat.completions.create(
+    model="meta-llama/llama-4-scout-17b-16e-instruct",
+    messages=[{
+        "role": "user",
+        "content": [
+            {"type": "text", "text": "Describe this image."},
+            {"type": "image_url", "image_url": {"url": "https://example.com/photo.jpg"}}
+        ]
+    }],
+    max_completion_tokens=1024,
+)
+print(response.choices[0].message.content)
+```
+
+Vision models: `meta-llama/llama-4-scout-17b-16e-instruct`, `meta-llama/llama-4-maverick-17b-128e-instruct`.
+
+## Audio / Speech-to-Text
+
+Whisper on LPU for near-instant transcription. Formats: flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, webm. Max: 25 MB free / 100 MB developer tier.
+
+```python
+from groq import Groq
+
+client = Groq()
+
+# Transcription
+with open("audio.mp3", "rb") as f:
+    result = client.audio.transcriptions.create(
+        file=("audio.mp3", f),
+        model="whisper-large-v3-turbo",
+        language="en",
+        response_format="verbose_json",
+    )
+print(result.text)
+
+# Translation (any language -> English)
+with open("french.mp3", "rb") as f:
+    result = client.audio.translations.create(
+        file=("french.mp3", f), model="whisper-large-v3",
+    )
+print(result.text)
+```
+
+### Text-to-Speech
+
+```python
+from groq import Groq
+from pathlib import Path
+
+client = Groq()
+speech = client.audio.speech.create(
+    model="playai-tts",
+    voice="Arista-PlayAI",
+    input="Hello from Groq!",
+)
+Path("output.mp3").write_bytes(speech.content)
+```
+
+## Model Selection
+
+### Production Models
+
+| Model | Model ID | Speed | Context | Max Output | Input $/1M | Output $/1M |
+|-------|----------|-------|---------|------------|------------|-------------|
+| Llama 3.1 8B | `llama-3.1-8b-instant` | 560 t/s | 131,072 | 131,072 | $0.05 | $0.08 |
+| Llama 3.3 70B | `llama-3.3-70b-versatile` | 280 t/s | 131,072 | 32,768 | $0.59 | $0.79 |
+| GPT-OSS 120B | `openai/gpt-oss-120b` | 500 t/s | 131,072 | 65,536 | $0.15 | $0.60 |
+| GPT-OSS 20B | `openai/gpt-oss-20b` | 1000 t/s | 131,072 | 65,536 | $0.075 | $0.30 |
+| Whisper V3 | `whisper-large-v3` | - | - | - | $0.111/hr | - |
+| Whisper V3 Turbo | `whisper-large-v3-turbo` | - | - | - | $0.04/hr | - |
+
+### Preview Models
+
+| Model | Model ID | Speed | Context | Input $/1M | Output $/1M |
+|-------|----------|-------|---------|------------|-------------|
+| Llama 4 Scout | `meta-llama/llama-4-scout-17b-16e-instruct` | 750 t/s | 131,072 | $0.11 | $0.34 |
+| Llama 4 Maverick | `meta-llama/llama-4-maverick-17b-128e-instruct` | 600 t/s | 131,072 | $0.20 | $0.60 |
+| Qwen3 32B | `qwen/qwen3-32b` | 400 t/s | 131,072 | $0.29 | $0.59 |
+| Kimi K2 | `moonshotai/kimi-k2-instruct-0905` | 200 t/s | 262,144 | $1.00 | $3.00 |
+| Safety GPT-OSS 20B | `openai/gpt-oss-safeguard-20b` | 1000 t/s | 131,072 | $0.075 | $0.30 |
+| Llama Guard 4 12B | `meta-llama/llama-guard-4-12b` | - | 131,072 | $0.20 | $0.20 |
+
+### Systems & TTS
+
+| Model | Model ID | Notes |
+|-------|----------|-------|
+| Compound | `groq/compound` | Agentic: web search + code exec |
+| Compound Mini | `groq/compound-mini` | Lighter agentic system |
+| PlayAI TTS | `playai-tts` | $22/1M chars |
+| Orpheus English | `canopylabs/orpheus-v1-english` | $22/1M chars |
+
+### Selection Guide
+
+- **Fastest + cheapest**: `llama-3.1-8b-instant` — classification, extraction, simple tasks
+- **Best quality**: `llama-3.3-70b-versatile` — complex reasoning, coding, chat
+- **Best value**: `openai/gpt-oss-120b` — strong quality, 500 t/s, $0.15 input
+- **Vision**: `meta-llama/llama-4-scout-17b-16e-instruct` — 750 t/s, cheapest multimodal
+- **Long context**: `moonshotai/kimi-k2-instruct-0905` — 262K context window
+- **Safety**: `meta-llama/llama-guard-4-12b` or `openai/gpt-oss-safeguard-20b`
+
+## OpenAI Compatibility
+
+Drop-in replacement — change `base_url` and `api_key`:
+
+```python
+from openai import OpenAI
+import os
+
+client = OpenAI(
+    base_url="https://api.groq.com/openai/v1",
+    api_key=os.environ.get("GROQ_API_KEY"),
+)
+response = client.chat.completions.create(
+    model="llama-3.3-70b-versatile",
+    messages=[{"role": "user", "content": "Hello from OpenAI SDK!"}],
+)
+print(response.choices[0].message.content)
+```
+
+| Endpoint | Supported | Notes |
+|----------|-----------|-------|
+| `POST /chat/completions` | Yes | Streaming, tools, JSON mode |
+| `POST /audio/transcriptions` | Yes | Whisper |
+| `POST /audio/translations` | Yes | Whisper |
+| `POST /audio/speech` | Yes | TTS |
+| `GET /models` | Yes | List models |
+| `POST /embeddings` | No | Not available |
+| `POST /images/generations` | No | Not available |
+
+**Unsupported fields** (400 error): `logprobs`, `logit_bias`, `top_logprobs`, `messages[].name`, `n` (must be 1).
+
+## Rate Limits
+
+| Tier | Price | RPM | TPM | TPD |
+|------|-------|-----|-----|-----|
+| Free | $0 | 30 | 6,000 | 500,000 |
+| Developer | Pay-per-token | Up to 1,000 | Up to 300,000 | Unlimited |
+
+Rate limit headers returned with every response: `x-ratelimit-limit-requests`, `x-ratelimit-remaining-requests`, `x-ratelimit-limit-tokens`, `x-ratelimit-remaining-tokens`.
+
+```python
+import time
+from groq import Groq, RateLimitError
+
+client = Groq()
+
+def call_with_retry(messages, model="llama-3.3-70b-versatile", max_retries=3):
+    for attempt in range(max_retries):
+        try:
+            return client.chat.completions.create(model=model, messages=messages)
+        except RateLimitError:
+            if attempt == max_retries - 1:
+                raise
+            time.sleep(2 ** attempt)
+```
+
+## Cost Optimization
+
+1. **Use the smallest model that works.** `llama-3.1-8b-instant` is 10x cheaper than the 70B for simple tasks.
+2. **Cap output with `max_completion_tokens`.** Output tokens cost more than input.
+3. **Try `openai/gpt-oss-20b`** — best price/performance at $0.075 input and 1000 t/s.
+4. **Cache at `temperature=0`.** Deterministic output means identical prompts yield identical results.
+5. **Use Whisper Turbo.** `whisper-large-v3-turbo` is 64% cheaper than v3 with minimal quality loss.
+6. **Preprocess audio** to 16KHz mono WAV to reduce upload size.
+
+## Common Issues
+
+| Problem | Solution |
+|---------|----------|
+| `401 Unauthorized` | Check `GROQ_API_KEY`. Get key at https://console.groq.com/keys |
+| `400 Bad Request` | Remove unsupported fields: `logprobs`, `logit_bias`, `top_logprobs` |
+| `429 Rate Limited` | Exponential backoff. Upgrade to developer tier for 10x limits |
+| Empty stream chunks | Check `chunk.choices[0].delta.content` for `None` |
+| Invalid JSON output | Add "Output valid JSON only" to system prompt, set `temperature=0` |
+| Tool calls missing | Not all models support tools. Use Llama 3.3 70B or GPT-OSS |
+| Audio too large | Free: 25 MB max. Use `url` param or upgrade to dev tier (100 MB) |
+| Vision fails | Only Llama 4 models support vision. Max 20 MB |
+| `n` param error | Groq only supports `n=1` |
+| `max_tokens` warning | Use `max_completion_tokens` instead |
+
+## Resources
+
+- **API Docs**: https://console.groq.com/docs/overview
+- **Models**: https://console.groq.com/docs/models
+- **API Keys**: https://console.groq.com/keys
+- **Pricing**: https://groq.com/pricing
+- **Rate Limits**: https://console.groq.com/docs/rate-limits
+- **OpenAI Compat**: https://console.groq.com/docs/openai
+- **Tool Use**: https://console.groq.com/docs/tool-use/local-tool-calling
+- **Vision**: https://console.groq.com/docs/vision
+- **Speech-to-Text**: https://console.groq.com/docs/speech-to-text
+- **Structured Outputs**: https://console.groq.com/docs/structured-outputs
+- **Python SDK**: https://github.com/groq/groq-python (`pip install groq`)
+- **Status**: https://status.groq.com
+- **LPU Architecture**: https://groq.com/blog/inside-the-lpu-deconstructing-groq-speed

package/bin/skills/hypothesis-generation/SKILL.md

@@ -0,0 +1,293 @@
+---
+name: hypothesis-generation
+description: "Generate testable hypotheses. Formulate from observations, design experiments, explore competing explanations, develop predictions, propose mechanisms, for scientific inquiry across domains."
+allowed-tools: [Read, Write, Edit, Bash]
+---
+
+# Scientific Hypothesis Generation
+
+## Overview
+
+Hypothesis generation is a systematic process for developing testable explanations. Formulate evidence-based hypotheses from observations, design experiments, explore competing explanations, and develop predictions. Apply this skill for scientific inquiry across domains.
+
+## When to Use This Skill
+
+This skill should be used when:
+- Developing hypotheses from observations or preliminary data
+- Designing experiments to test scientific questions
+- Exploring competing explanations for phenomena
+- Formulating testable predictions for research
+- Conducting literature-based hypothesis generation
+- Planning mechanistic studies across scientific domains
+
+## Visual Enhancement with Scientific Schematics
+
+**⚠️ MANDATORY: Every hypothesis generation report MUST include at least 1-2 AI-generated figures using the scientific-schematics skill.**
+
+This is not optional. Hypothesis reports without visual elements are incomplete. Before finalizing any document:
+1. Generate at minimum ONE schematic or diagram (e.g., hypothesis framework showing competing explanations)
+2. Prefer 2-3 figures for comprehensive reports (mechanistic pathway, experimental design flowchart, prediction decision tree)
+
+**How to generate figures:**
+- Use the **scientific-schematics** skill to generate AI-powered publication-quality diagrams
+- Simply describe your desired diagram in natural language
+- Nano Banana Pro will automatically generate, review, and refine the schematic
+
+**How to generate schematics:**
+```bash
+python scripts/generate_schematic.py "your diagram description" -o figures/output.png
+```
+
+The AI will automatically:
+- Create publication-quality images with proper formatting
+- Review and refine through multiple iterations
+- Ensure accessibility (colorblind-friendly, high contrast)
+- Save outputs in the figures/ directory
+
+**When to add schematics:**
+- Hypothesis framework diagrams showing competing explanations
+- Experimental design flowcharts
+- Mechanistic pathway diagrams
+- Prediction decision trees
+- Causal relationship diagrams
+- Theoretical model visualizations
+- Any complex concept that benefits from visualization
+
+For detailed guidance on creating schematics, refer to the scientific-schematics skill documentation.
+
+---
+
+## Workflow
+
+Follow this systematic process to generate robust scientific hypotheses:
+
+### 1. Understand the Phenomenon
+
+Start by clarifying the observation, question, or phenomenon that requires explanation:
+
+- Identify the core observation or pattern that needs explanation
+- Define the scope and boundaries of the phenomenon
+- Note any constraints or specific contexts
+- Clarify what is already known vs. what is uncertain
+- Identify the relevant scientific domain(s)
+
+### 2. Conduct Comprehensive Literature Search
+
+Search existing scientific literature to ground hypotheses in current evidence. Use both PubMed (for biomedical topics) and general web search (for broader scientific domains):
+
+**For biomedical topics:**
+- Use WebFetch with PubMed URLs to access relevant literature
+- Search for recent reviews, meta-analyses, and primary research
+- Look for similar phenomena, related mechanisms, or analogous systems
+
+**For all scientific domains:**
+- Use WebSearch to find recent papers, preprints, and reviews
+- Search for established theories, mechanisms, or frameworks
+- Identify gaps in current understanding
+
+**Search strategy:**
+- Begin with broad searches to understand the landscape
+- Narrow to specific mechanisms, pathways, or theories
+- Look for contradictory findings or unresolved debates
+- Consult `references/literature_search_strategies.md` for detailed search techniques
+
+### 3. Synthesize Existing Evidence
+
+Analyze and integrate findings from literature search:
+
+- Summarize current understanding of the phenomenon
+- Identify established mechanisms or theories that may apply
+- Note conflicting evidence or alternative viewpoints
+- Recognize gaps, limitations, or unanswered questions
+- Identify analogies from related systems or domains
+
+### 4. Generate Competing Hypotheses
+
+Develop 3-5 distinct hypotheses that could explain the phenomenon. Each hypothesis should:
+
+- Provide a mechanistic explanation (not just description)
+- Be distinguishable from other hypotheses
+- Draw on evidence from the literature synthesis
+- Consider different levels of explanation (molecular, cellular, systemic, population, etc.)
+
+**Strategies for generating hypotheses:**
+- Apply known mechanisms from analogous systems
+- Consider multiple causative pathways
+- Explore different scales of explanation
+- Question assumptions in existing explanations
+- Combine mechanisms in novel ways
+
+### 5. Evaluate Hypothesis Quality
+
+Assess each hypothesis against established quality criteria from `references/hypothesis_quality_criteria.md`:
+
+**Testability:** Can the hypothesis be empirically tested?
+**Falsifiability:** What observations would disprove it?
+**Parsimony:** Is it the simplest explanation that fits the evidence?
+**Explanatory Power:** How much of the phenomenon does it explain?
+**Scope:** What range of observations does it cover?
+**Consistency:** Does it align with established principles?
+**Novelty:** Does it offer new insights beyond existing explanations?
+
+Explicitly note the strengths and weaknesses of each hypothesis.
+
+### 6. Design Experimental Tests
+
+For each viable hypothesis, propose specific experiments or studies to test it. Consult `references/experimental_design_patterns.md` for common approaches:
+
+**Experimental design elements:**
+- What would be measured or observed?
+- What comparisons or controls are needed?
+- What methods or techniques would be used?
+- What sample sizes or statistical approaches are appropriate?
+- What are potential confounds and how to address them?
+
+**Consider multiple approaches:**
+- Laboratory experiments (in vitro, in vivo, computational)
+- Observational studies (cross-sectional, longitudinal, case-control)
+- Clinical trials (if applicable)
+- Natural experiments or quasi-experimental designs
+
+### 7. Formulate Testable Predictions
+
+For each hypothesis, generate specific, quantitative predictions:
+
+- State what should be observed if the hypothesis is correct
+- Specify expected direction and magnitude of effects when possible
+- Identify conditions under which predictions should hold
+- Distinguish predictions between competing hypotheses
+- Note predictions that would falsify the hypothesis
+
+### 8. Present Structured Output
+
+Generate a professional LaTeX document using the template in `assets/hypothesis_report_template.tex`. The report should be well-formatted with colored boxes for visual organization and divided into a concise main text with comprehensive appendices.
+
+**Document Structure:**
+
+**Main Text (Maximum 4 pages):**
+1. **Executive Summary** - Brief overview in summary box (0.5-1 page)
+2. **Competing Hypotheses** - Each hypothesis in its own colored box with brief mechanistic explanation and key evidence (2-2.5 pages for 3-5 hypotheses)
+   - **IMPORTANT:** Use `\newpage` before each hypothesis box to prevent content overflow
+   - Each box should be ≤0.6 pages maximum
+3. **Testable Predictions** - Key predictions in amber boxes (0.5-1 page)
+4. **Critical Comparisons** - Priority comparison boxes (0.5-1 page)
+
+Keep main text highly concise - only the most essential information. All details go to appendices.
+
+**Page Break Strategy:**
+- Always use `\newpage` before hypothesis boxes to ensure they start on fresh pages
+- This prevents content from overflowing off page boundaries
+- LaTeX boxes (tcolorbox) do not automatically break across pages
+
+**Appendices (Comprehensive, Detailed):**
+- **Appendix A:** Comprehensive literature review with extensive citations
+- **Appendix B:** Detailed experimental designs with full protocols
+- **Appendix C:** Quality assessment tables and detailed evaluations
+- **Appendix D:** Supplementary evidence and analogous systems
+
+**Colored Box Usage:**
+
+Use the custom box environments from `hypothesis_generation.sty`:
+
+- `hypothesisbox1` through `hypothesisbox5` - For each competing hypothesis (blue, green, purple, teal, orange)
+- `predictionbox` - For testable predictions (amber)
+- `comparisonbox` - For critical comparisons (steel gray)
+- `evidencebox` - For supporting evidence highlights (light blue)
+- `summarybox` - For executive summary (blue)
+
+**Each hypothesis box should contain (keep concise for 4-page limit):**
+- **Mechanistic Explanation:** 1-2 brief paragraphs (6-10 sentences max) explaining HOW and WHY
+- **Key Supporting Evidence:** 2-3 bullet points with citations (most important evidence only)
+- **Core Assumptions:** 1-2 critical assumptions
+
+All detailed explanations, additional evidence, and comprehensive discussions belong in the appendices.
+
+**Critical Overflow Prevention:**
+- Insert `\newpage` before each hypothesis box to start it on a fresh page
+- Keep each complete hypothesis box to ≤0.6 pages (approximately 15-20 lines of content)
+- If content exceeds this, move additional details to Appendix A
+- Never let boxes overflow off page boundaries - this creates unreadable PDFs
+
+**Citation Requirements:**
+
+Aim for extensive citation to support all claims:
+- **Main text:** 10-15 key citations for most important evidence only (keep concise for 4-page limit)
+- **Appendix A:** 40-70+ comprehensive citations covering all relevant literature
+- **Total target:** 50+ references in bibliography
+
+Main text citations should be selective - cite only the most critical papers. All comprehensive citation and detailed literature discussion belongs in the appendices. Use `\citep{author2023}` for parenthetical citations.
+
+**LaTeX Compilation:**
+
+The template requires XeLaTeX or LuaLaTeX for proper rendering:
+
+```bash
+xelatex hypothesis_report.tex
+bibtex hypothesis_report
+xelatex hypothesis_report.tex
+xelatex hypothesis_report.tex
+```
+
+**Required packages:** The `hypothesis_generation.sty` style package must be in the same directory or LaTeX path. It requires: tcolorbox, xcolor, fontspec, fancyhdr, titlesec, enumitem, booktabs, natbib.
+
+**Page Overflow Prevention:**
+
+To prevent content from overflowing on pages, follow these critical guidelines:
+
+1. **Monitor Box Content Length:** Each hypothesis box should fit comfortably on a single page. If content exceeds ~0.7 pages, it will likely overflow.
+
+2. **Use Strategic Page Breaks:** Insert `\newpage` before boxes that contain substantial content:
+   ```latex
+   \newpage
+   \begin{hypothesisbox1}[Hypothesis 1: Title]
+   % Long content here
+   \end{hypothesisbox1}
+   ```
+
+3. **Keep Main Text Boxes Concise:** For the 4-page main text limit:
+   - Each hypothesis box: Maximum 0.5-0.6 pages
+   - Mechanistic explanation: 1-2 brief paragraphs only (6-10 sentences max)
+   - Key evidence: 2-3 bullet points only
+   - Core assumptions: 1-2 items only
+   - If content is longer, move details to appendices
+
+4. **Break Long Content:** If a hypothesis requires extensive explanation, split across main text and appendix:
+   - Main text box: Brief mechanistic overview + 2-3 key evidence points
+   - Appendix A: Detailed mechanism explanation, comprehensive evidence, extended discussion
+
+5. **Test Page Boundaries:** Before each new box, consider if remaining page space is sufficient. If less than 0.6 pages remain, use `\newpage` to start the box on a fresh page.
+
+6. **Appendix Page Management:** In appendices, use `\newpage` between major sections to avoid overflow in detailed content areas.
+
+**Quick Reference:** See `assets/FORMATTING_GUIDE.md` for detailed examples of all box types, color schemes, and common formatting patterns.
+
+## Quality Standards
+
+Ensure all generated hypotheses meet these standards:
+
+- **Evidence-based:** Grounded in existing literature with citations
+- **Testable:** Include specific, measurable predictions
+- **Mechanistic:** Explain how/why, not just what
+- **Comprehensive:** Consider alternative explanations
+- **Rigorous:** Include experimental designs to test predictions
+
+## Resources
+
+### references/
+
+- `hypothesis_quality_criteria.md` - Framework for evaluating hypothesis quality (testability, falsifiability, parsimony, explanatory power, scope, consistency)
+- `experimental_design_patterns.md` - Common experimental approaches across domains (RCTs, observational studies, lab experiments, computational models)
+- `literature_search_strategies.md` - Effective search techniques for PubMed and general scientific sources
+
+### assets/
+
+- `hypothesis_generation.sty` - LaTeX style package providing colored boxes, professional formatting, and custom environments for hypothesis reports
+- `hypothesis_report_template.tex` - Complete LaTeX template with main text structure and comprehensive appendix sections
+- `FORMATTING_GUIDE.md` - Quick reference guide with examples of all box types, color schemes, citation practices, and troubleshooting tips
+
+### Related Skills
+
+When preparing hypothesis-driven research for publication, consult the **venue-templates** skill for writing style guidance:
+- `venue_writing_styles.md` - Master guide comparing styles across venues
+- Venue-specific guides for Nature/Science, Cell Press, medical journals, and ML/CS conferences
+- `reviewer_expectations.md` - What reviewers look for when evaluating research hypotheses