crossai-cli 0.1.0__py3-none-any.whl

cross_ai/ai_error_handler.py

@@ -0,0 +1,15 @@
+# ai_error_handler.py — compatibility shim
+# Source of truth: cross-ai-core package (cross_ai_core.ai_error_handler)
+from cross_ai_core.ai_error_handler import *   # noqa: F401, F403
+from cross_ai_core.ai_error_handler import (   # explicit for IDE / type checkers
+    handle_api_error,
+    is_quota_error,
+    is_rate_limit_error,
+    is_transient_error,
+    get_error_type,
+    retry_with_backoff,
+    CrossAIError,
+    QuotaExceededError,
+    RateLimitError,
+    TransientError,
+)

cross_ai/ai_handler.py

@@ -0,0 +1,22 @@
+# ai_handler.py — compatibility shim
+# Source of truth: cross-ai-core package (cross_ai_core.ai_handler)
+# This file exists so st-*.py files and tests can continue to use:
+#   from ai_handler import process_prompt, get_content, ...
+from cross_ai_core.ai_handler import *        # noqa: F401, F403
+from cross_ai_core.ai_handler import (        # explicit for IDE / type checkers
+    AI_HANDLER_REGISTRY,
+    AI_LIST,
+    AIResponse,
+    _API_KEY_ENV_VARS,                        # private — needed by tests
+    check_api_key,
+    get_ai_list,
+    get_ai_make,
+    get_ai_model,
+    get_content,
+    get_data_content,
+    get_data_title,
+    get_default_ai,
+    get_usage,
+    process_prompt,
+    put_content,
+)

cross_ai/ai_url.py

@@ -0,0 +1,89 @@
+# Sorry, no AI here
+import os
+import json
+import hashlib
+import requests
+import sys
+
+AI_MAKE = "url"
+AI_MODEL = "bs4"
+
+
+def get_url_cached_response(tweet_id, verbose=False, use_cache=False):
+
+    url = f"https://api.twitter.com/2/tweets/{tweet_id}"
+    headers = get_url_headers()
+
+    if not use_cache:
+        response = requests.get(url, headers=headers)
+        json_str = json.dumps(response)  # Force structure reorganization
+        json_response = json.loads(json_str)
+        return json_response, False  # Not cached
+    else:
+        # Convert param to a string for hashing
+        param = {
+            "tweet_id": tweet_id,
+            "headers": headers,
+            "url": url,
+        }
+        param_str = json.dumps(param, sort_keys=True)
+        md5_hash = hashlib.md5(param_str.encode('utf-8')).hexdigest()
+
+        # Construct the cache file path
+        cache_dir = os.path.expanduser("~/.cross_api_cache")
+        cache_file = os.path.join(cache_dir, f"{md5_hash}.json")
+
+        # Check if the response is already in cache
+        if os.path.exists(cache_file):
+            if verbose:
+                print(f"api_cache: Using cache_file: {cache_file}")
+            with open(cache_file, 'r') as f:
+                return json.load(f), True  # Cached
+        else:
+            if verbose:
+                print("api_cache: cache miss, submitting API request")
+
+            # If not in cache, fetch the response
+            response = requests.get(url, headers=headers)
+            if (status_code := response["status_code"]) != 200:
+                print(f"x.com response failed: {status_code}")
+                sys.exit(1)
+            json_str = json.dumps(response)  # Force structure reorganization
+            json_response = json.loads(json_str)
+
+            # Save to cache
+            if not os.path.exists(cache_dir):
+                os.makedirs(cache_dir)
+                if verbose:
+                    print(f"api_cache: api_cache/ dir created: {cache_dir}")
+
+            try:
+                with open(cache_file, 'w') as f:
+                    json.dump(json_response, f)
+                    if verbose:
+                        print(f"api_cache: file created: {cache_file}")
+            except Exception as e:
+                print(f"api_cache: file write error: {str(e)}")
+
+            return json_response, False  # Fresh API call, not cached
+
+
+def get_url_headers():
+    x_bearer_token = os.environ.get('X_COM_BEARER_TOKEN')
+    headers = {"Authorization": f"Bearer {x_bearer_token}"}
+    return headers
+
+
+def get_title(data):
+    title = data.get("title")
+    return title
+
+
+def get_story(data):
+    text = data.get("text")
+    return text
+
+
+def get_ai_tag():
+    return "\n\ncross:" + json.dumps(
+        {"make": AI_MAKE, "model": AI_MODEL})

cross_ai/base_handler.py

@@ -0,0 +1,3 @@
+# base_handler.py — compatibility shim
+# Source of truth: cross-ai-core package (cross_ai_core.ai_base)
+from cross_ai_core.ai_base import BaseAIHandler, _get_cache_dir  # noqa: F401

cross_ai/commands.py

@@ -0,0 +1,54 @@
+"""
+commands.py — Entry-point dispatch for pyproject.toml [project.scripts].
+pip generates a thin wrapper that imports and calls each function here;
+runpy executes the st-*.py file directly so hyphens in filenames are never
+treated as Python identifiers.
+"""
+import os
+import runpy
+import sys
+
+_HERE = os.path.dirname(os.path.abspath(__file__))
+
+# Ensure the cross_ai/ directory is on sys.path so that st-*.py scripts can
+# use bare module imports (e.g. `from mmd_startup import require_config`).
+# When scripts are run directly as files, Python adds their directory to
+# sys.path[0] automatically — runpy.run_path() does not, so we do it here.
+if _HERE not in sys.path:
+    sys.path.insert(0, _HERE)
+
+
+def _run(name: str) -> None:
+    runpy.run_path(os.path.join(_HERE, f"{name}.py"), run_name="__main__")
+
+
+def st():           _run("st")
+def st_admin():     _run("st-admin")
+def st_analyze():   _run("st-analyze")
+def st_bang():      _run("st-bang")
+def st_cat():       _run("st-cat")
+def st_cross():     _run("st-cross")
+def st_domain():    _run("st-domain")
+def st_edit():      _run("st-edit")
+def st_fact():      _run("st-fact")
+def st_fetch():     _run("st-fetch")
+def st_find():      _run("st-find")
+def st_fix():       _run("st-fix")
+def st_gen():       _run("st-gen")
+def st_heatmap():   _run("st-heatmap")
+def st_ls():        _run("st-ls")
+def st_man():       _run("st-man")
+def st_merge():     _run("st-merge")
+def st_new():       _run("st-new")
+def st_plot():      _run("st-plot")
+def st_post():      _run("st-post")
+def st_prep():      _run("st-prep")
+def st_print():     _run("st-print")
+def st_read():      _run("st-read")
+def st_rm():        _run("st-rm")
+def st_speak():     _run("st-speak")
+def st_speed():     _run("st-speed")
+def st_stones():    _run("st-stones")
+def st_verdict():   _run("st-verdict")
+def st_voice():     _run("st-voice")
+

cross_ai/cross_stones/cross-stones-10.json

@@ -0,0 +1,57 @@
+{
+  "id": "cross-stones-10",
+  "description": "Standard 10-domain Cross-Stones benchmark set. Locked params ensure scores are directly comparable across runs over time.",
+  "n_claims": 10,
+  "max_fact_score": 200,
+  "domains_dir": "domains",
+  "domains": [
+    {
+      "id": "software_development",
+      "name": "Software Development & Programming"
+    },
+    {
+      "id": "customer_service",
+      "name": "Customer Service & Support"
+    },
+    {
+      "id": "marketing_content",
+      "name": "Marketing & Content Creation"
+    },
+    {
+      "id": "education_learning",
+      "name": "Education & Learning"
+    },
+    {
+      "id": "data_analytics",
+      "name": "Data Analytics & Business Intelligence"
+    },
+    {
+      "id": "healthcare_medical",
+      "name": "Healthcare & Medical Analysis"
+    },
+    {
+      "id": "finance_business",
+      "name": "Finance & Business Decision Making"
+    },
+    {
+      "id": "writing_editing",
+      "name": "Writing, Editing & Summarizing"
+    },
+    {
+      "id": "research_qa",
+      "name": "Research, Search & Q&A"
+    },
+    {
+      "id": "creative_media",
+      "name": "Creative Media (Images, Video, Audio)"
+    }
+  ],
+  "snapshots": [],
+  "speed_baseline": {
+    "gen_seconds": 20.07,
+    "fc_seconds": 91.87,
+    "total_seconds": 111.94,
+    "recorded_date": "2026-03-27",
+    "notes": "Baseline recorded 2026-03-27 \u2014 average across all AI providers.  Future runs faster than 111.9s total will have speed_ratio > 1.0 and cross_stone_score may exceed 1.0, reflecting genuine improvement over the 2026 benchmark era."
+  }
+}

cross_ai/cross_stones/domains/creative_media.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Creative Media (Images, Video, Audio)
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in creative media production -- including image generation, video synthesis, music composition, and audio production -- as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as company announcements, legal filings, industry reports, or reputable technology and entertainment news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named tools), and half should require consulting primary sources, legal records, or detailed technical assessments published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific data points, named tools, model versions, legal case names, revenue figures, or named organizations where relevant.
+
+Distribute your 10 claims across the following aspects:
+- Capabilities and market adoption of leading AI image and video generation tools (e.g., DALL-E 3, Midjourney v6, Sora, Runway Gen-3, Stable Diffusion)
+- AI in music generation, voice cloning, and audio production: leading tools, capabilities, and commercial deployments
+- Commercial deployment of AI-generated media in advertising, film, publishing, and gaming industries
+- Copyright, intellectual property, and compensation disputes involving AI-generated content (specific cases or legislation)
+- Regulatory responses, platform policies, and industry standards addressing AI-generated deepfakes and synthetic media
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/customer_service.prompt

@@ -0,0 +1,12 @@
+# Cross-Stone Benchmark Prompt — Customer Service & Support
+Write exactly 10 specific, fact-checkable claims about the current state of AI in customer service and support automation as of 2025–2026.
+Each claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources such as industry surveys, vendor case studies, analyst reports, or reputable news outlets — as True, Partially True, Opinion, Partially False, or False.
+**Difficulty calibration:** Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed performance data published in 2024–2025.
+**Claims must not be vague generalizations.** Each claim should include specific data points, named platforms, percentages, benchmark metrics, or named organizations where relevant.
+Distribute your 10 claims across the following aspects:
+- Automation and deflection rates for AI-powered chatbots and virtual agents across industries
+- Customer satisfaction (CSAT) and first-contact resolution quality compared to human agents
+- Adoption rates across specific industries (retail, telecom, banking, healthcare)
+- Integration of large language models (LLMs) into contact center platforms (e.g., Salesforce Einstein, Zendesk AI, AWS Connect)
+- Documented cost savings, ROI metrics, and workforce impact reported by organizations deploying AI in customer service
+**Format:** Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/data_analytics.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Data Analytics & Business Intelligence
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in data analytics and business intelligence as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as vendor documentation, industry reports, academic studies, or reputable news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed technical documentation published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific data points, named tools, percentages, benchmark results, or named organizations where relevant.
+
+Distribute your 10 claims across the following aspects:
+- Adoption of AI and ML capabilities in major business intelligence platforms (e.g., Tableau, Microsoft Power BI, Looker, Snowflake Cortex)
+- Impact of large language models on natural-language querying and self-service analytics
+- AI-driven anomaly detection and predictive analytics performance in enterprise settings
+- Accuracy, reliability, and hallucination risks in AI-generated data insights vs. human analysts
+- Data governance, privacy, and explainability challenges in AI-driven analytics deployments
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/education_learning.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Education & Learning
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in education and personalized learning as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as academic research, institutional reports, or reputable news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed institutional data published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific data points, named platforms, percentages, study findings, or named institutions where relevant.
+
+Distribute your 10 claims across the following aspects:
+- Student and educator adoption rates of generative AI tools (e.g., ChatGPT, Khanmigo, Duolingo, Coursera AI)
+- Measured learning outcomes from AI-assisted instruction or tutoring systems
+- Institutional and government policies on AI use in classrooms and assessments
+- AI-powered adaptive learning platforms and the evidence base for their effectiveness
+- Academic integrity concerns, bias, and equity issues in AI-powered education
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/finance_business.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Finance & Business Decision Making
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in financial services, risk management, and business decision-making as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as regulatory filings, industry surveys, academic research, or reputable financial news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named systems), and half should require consulting primary sources, regulatory documents, or detailed financial performance data published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific data points, named platforms, dollar figures, performance metrics, regulatory citations, or named institutions where relevant.
+
+Distribute your 10 claims across the following aspects:
+- AI investment levels and adoption rates in the financial services sector (banks, insurers, asset managers)
+- Performance of AI in fraud detection, credit scoring, and AML (anti-money laundering) systems
+- AI in algorithmic trading and portfolio management: market share, strategies, and performance benchmarks
+- Regulatory frameworks governing AI in financial services (e.g., EU AI Act, U.S. OCC guidance, SEC rules)
+- Documented ROI, productivity gains, and risk reduction achieved by financial firms deploying AI
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/healthcare_medical.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Healthcare & Medical Analysis
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in healthcare, medical diagnostics, and clinical decision support as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as peer-reviewed journals, FDA databases, hospital system reports, or reputable health news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named tools), and half should require consulting primary clinical literature, regulatory filings, or detailed health system data published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific data points, named systems, FDA clearance details, clinical trial results, percentages, or named institutions where relevant.
+
+Distribute your 10 claims across the following aspects:
+- FDA-cleared or CE-marked AI medical devices: cumulative count, approval trends, and leading categories (radiology, pathology, cardiology)
+- Clinical performance of AI in medical imaging compared to human specialists (sensitivity, specificity, AUC metrics)
+- AI in drug discovery and clinical trial design: timelines, cost reductions, and named examples
+- Deployment and outcomes of AI clinical decision support systems in hospital settings
+- Regulatory, liability, bias, and health equity challenges in medical AI adoption
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/marketing_content.prompt

@@ -0,0 +1,72 @@
+# Cross-Stone Benchmark Prompt — Marketing & Content Creation
+Write exactly 10 specific, fact-checkable claims about the current state of AI in marketing and content creation as of 2025–2026.
+Each claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources such as industry surveys, vendor reports, academic studies, or reputable news outlets — as True, Partially True, Opinion, Partially False, or False.
+**Difficulty calibration:** Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed performance data published in 2024–2025.
+**Claims must not be vague generalizations.** Each claim should include specific data points, named platforms, percentages, benchmark metrics, or named organizations where releva# Cross-Stone Benchmark Prompt — Marketing & Content Creation
+Write exactly 10 specific, AIWrite exactly 10 specific, fact-checkable claims about the cururEach claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources sdi**Difficulty calibration:** Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed performance data published in 2024–2025.
+*AI**Claims must not be vague generalizations.** Each claim should include specific data points, named platforms, percentages, benchmark metriccat > /Users/Matt/github/cross-internal/st-stones/template/education_learning.prompt << 'EOF'
+# Cross-Stone Benchmark Prompt — Education & Learning
+Write exactly 10 specific, fact-checkable claims about the current state of AI in education and personalized learning as of 2025–2026.
+Each claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources such as academic research, institutional reports, or reputable news outlets — as True, Partially True, Opinion, Partially False, or False.
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed institutional data published in 2024-2025.
+Claims must not be vague generalizations. Each claim should include specific data points, named platforms, percentages, study findings, or named institutions where relevant.
+Distribute your 10 c# Cross-Stone Benchmark Prompt — Education & Learning
+Write exactly 10 specific, fact-checooWrite exactly 10 specific, fact-checkable claims aboutMeEach claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources such amsDifficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed institutional data published in 2024-2n Claims must not be vague generalizationpython3 << 'PYEOF'
+import os
+base = "/Users/Matt/github/cross-internal/st-stones/template"
+files = {
+"education_learning.prompt": """# Cross-Stone Benchmark Prompt — Education & Learning
+Write exactly 10 specific, fact-checkable claims about the current state of AI in education and personalized learning as of 2025-2026.
+Each claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources such as academic research, institutional reports, or reputable news outlets — as True, Partially True, Opinion, Partially False, or False.
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed institutional data published in 2024-2025.
+Claims must not be vague generalizations. Each claim should include specific data points, named platforms, percentages, study findings, or named institutionsimport os
+base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable claims about the current state of AI in educasEach claim must be a clear, declarative statement that a well-informed analyst could verify — using publicly available sources suchmsDifficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named platforms), and half should require consulting primary sources, empirical studies, or detailed institutional data published in 2024-2n Claims must not be vague generalizations. Each claim should include specific data points, named platforms, percentages, study findings, or named institutionsimport os
+base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable claims about the current state on,"education_learning.prompt" sbase = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable claims about the current state on,"education_learning.prompt" sbase = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable claims about the current state on,"education_learning.prompt" sbase = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_learning.prompt"Im"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable y "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,, "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dibase = " yfiles = {
+"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_learning.prompt"Im"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly jo"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,, "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dibase = " yfiles = {
+"education_learning.promptar"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_lear
+D"education_learning.promptar"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_lear
+D"education_learning.promptar"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_lear
+D"education_learning.promptar"edummD"education_learning.promptar"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_le tD"education_learning.promptar"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly20base = Dib L"education_leaasD"education_learning.promptar"edummD"education_learning.promptar"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific,ly"education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite exactly 10 specific, fact-checkable d "education_learning.prompt": """# Cross-Stone Bencuc"educatiptWrite ce metrics, regulatory citations, or named institutions where relevant.
+Distribute your 10 claims across the following aspects:
+- AI investment levels and adoption rates in the financial services sector (banks, insurers, asset managers)
+- Performance of AI in fraud detection, credit scoring, and AML (anti-money laundering) systems
+- AI in algorithmic trading and portfolio management: market share, strategies, and performance benchmarks
+- Regulatory frameworks governing AI in financial services (e.g., EU AI Act, U.S. OCC guidance, SEC rules)
+- Documented ROI, productivity gains, and risk reduction achieved by financial firms deploying AI
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.
+""",
+"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 10 specific, fact-checkable claims about the current state of AI in writing, editing, and document summarization as of 2025-2026.
+Each claim mustDistribute your 10 claims across the following aspects:
+- AI investmd - AI investment levels and adoption rates in the finanag- Performance of AI in fraud detection, credit scoring, and AML (anti-money laundering) systems
+- AI in algru- AI in algorithmic trading and portfolio management: market share, strategies, and performancs - Regulatory frameworks governing AI in financial services (e.g., EU AI Act, U.S. OCC guidance, SEC rulesui- Documented ROI, productivity gains, and risk reduction achieved by financial firms deploying AI
+Format:25Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, d """,
+"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the current state of AI in writing, edtGEach claim mustDistribute your 10 claims across the following aspects:
+- AI investmd - AI investment levels and adoption rates in the finanag-ns- AI investmd - AI investment levels and adoption rates in the finanan - AI in algru- AI in algorithmic trading and portfolio management: market share, strategies, and performancs - Regulatory frameworks governing AI in financial servi.,Format:25Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, d """,
+"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the current state of AIar"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI investmd - AI investment levels and adoption rates in the finanag-ns- AI investmd - AI investment levels and adoption rates in the finanan - AI in algru- AI in algorithmic trading and pout"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the current state of AIar"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI invests.Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentraWrite exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI investmd - AI investment levels and adoption rates in the finanag-ns- AI investmd -g Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the current state of AIar"writing_editing.prompt": """# Cross-Stone Benchmark Prompt — Writing, Editing & Summarizing
+Write exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI invests.Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable clinWrite exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI invests.Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claimseaWrite exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI invests.Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable clinWrite exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI invests.Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claimseaWrite exactly 1f Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable claims about the currentt - AI invests.Write exactly 10 spes "wrosWrite exactly 10 specific, fact-checkable clinWrite exactly 1f Write else.
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named tools), and half should require consulting primary sources, legal records, or detailed technical assessments published in 2024-2025.
+Claims must not be vague generalizations. Each claim should include specific data points, named tools, model versions, legal case names, revenue figures, or named organizations where relevant.
+Distribute your 10 claims across the following aspects:
+- Capabilities and market adoption of leading AI image and video generation tools (e.g., DALL-E 3, Midjourney v6, Sora, Runway Gen-3, Stable Diffusion)
+- AI in music generation, voice cloning, and audio production: leading tools, capabilities, and commercial deployments
+- Commercial deployment of AI-generated media in advertising, film, publishing, and gaming industries
+- Copyright, intellectual property, and compensation disputes involving AI-generated content (specific cases or legislation)
+- ReguDatClaims must not be vague generalizations. Each claim should include specific data points, named tools, model versions, legal case names, revenue figures, or named organizations where relevant.
+Distribute your 10 claims across the following aspects:
+- Capabilities anilDistribute your 10 claims across the following aspects:
+- Capabilities and market adoption of leading AI image and video generation tools (e.g., DALL-E 3, Midjourney v6, Spython3 /Users/Matt/github/cross-internal/st-stones/create_prompts.py

cross_ai/cross_stones/domains/research_qa.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Research, Search & Q&A
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in research, information retrieval, and question answering as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as benchmark leaderboards, academic papers, product documentation, or reputable technology news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported benchmark results or named tools), and half should require consulting primary literature, technical evaluations, or detailed capability assessments published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific benchmark names and scores, named systems, hallucination rates, retrieval accuracy metrics, or named organizations where relevant.
+
+Distribute your 10 claims across the following aspects:
+- Performance of leading AI models on established knowledge and reasoning benchmarks (e.g., MMLU, HLE, BrowseComp, SimpleQA)
+- Hallucination rates and factual accuracy of leading AI models in open-domain Q&A tasks
+- AI-powered research and search tools in academic and scientific settings (e.g., Perplexity, Elicit, Semantic Scholar, Google NotebookLM)
+- AI in systematic literature review, evidence synthesis, and meta-analysis workflows
+- Identified failure modes of AI in multi-step reasoning, temporal reasoning, and complex research tasks
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.

cross_ai/cross_stones/domains/software_development.prompt

@@ -0,0 +1,19 @@
+# Cross-Stone Benchmark Prompt — Software Development & Programming
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in software development and computer programming as of 2025–2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst or technical researcher could verify — using publicly available sources such as research papers, vendor reports, official documentation, or reputable news outlets — as True, Partially True, Opinion, Partially False, or False.
+
+**Difficulty calibration:** Approximately half of the claims should be verifiable with basic research (widely reported statistics or named tools), and half should require consulting primary sources, empirical studies, or detailed technical documentation published in 2024–2025.
+
+**Claims must not be vague generalizations.** Each claim should include specific data points, named tools, version numbers, percentages, benchmark names, or named organizations where relevant.
+
+Distribute your 10 claims across the following aspects:
+- Adoption rates and market penetration of AI coding assistants (e.g., GitHub Copilot, Cursor, Claude Code, Gemini Code Assist)
+- Measurable developer productivity impacts: speed, code quality, and defect rates
+- AI's share of code written in enterprise and open-source projects
+- Capabilities and current limitations in automated testing, debugging, and code review
+- Security vulnerabilities, intellectual property concerns, and license compliance issues related to AI-generated code
+
+**Format:** Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.
+

cross_ai/cross_stones/domains/writing_editing.prompt

@@ -0,0 +1,18 @@
+# Cross-Stone Benchmark Prompt -- Writing, Editing & Summarizing
+
+Write exactly 10 specific, fact-checkable claims about the current state of AI in writing, editing, and document summarization as of 2025-2026.
+
+Each claim must be a clear, declarative statement that a well-informed analyst could verify -- using publicly available sources such as usage reports, academic benchmarks, organizational policies, or reputable news outlets -- as True, Partially True, Opinion, Partially False, or False.
+
+Difficulty calibration: Approximately half of the claims should be verifiable with basic research (widely reported statistics or named tools), and half should require consulting primary sources, benchmark studies, or organizational policy documents published in 2024-2025.
+
+Claims must not be vague generalizations. Each claim should include specific data points, named tools, benchmark scores, percentages, organizational names, or policy citations where relevant.
+
+Distribute your 10 claims across the following aspects:
+- Adoption and usage statistics for AI writing assistants (e.g., ChatGPT, Grammarly, Notion AI, Microsoft Copilot, Claude)
+- Quality benchmarks comparing AI-generated text to human-written content across dimensions such as readability, accuracy, and factual consistency
+- Use of AI in professional writing contexts: legal documents, medical records, journalism, and technical writing
+- Accuracy and limitations of AI-generated content detectors (e.g., GPTZero, Copyleaks, Turnitin AI detection)
+- Regulatory requirements and organizational disclosure policies governing AI-generated content
+
+Format: Return a numbered list of exactly 10 claims with no introductory text, section headers, summaries, or commentary.