AI & ML

samraddhashrivastavatech Skill

Agent-device messaging channel — the private comms pathway between the host agent and the device for confirmation prompts, status pings, and task results.

samraddhashrivastavatech Skill

Operate the persistent local digital-person runtime and coordinate its approved goals with OpenCode and a separate computer-use MCP.

samraddhashrivastavatech Bundle

Plan, construct, critique, correct and export professional editable artwork using separate Artist and Visual MCPs.

samraddhashrivastavatech Skill

Grease Pencil 3 (Blender 4.3+) object model, 2D drawing and animation, GP materials and looks, classic 2D inside Blender, and hybrid 2D-3D pipelines.

samraddhashrivastavatech Skill

Read the full device profile — model, Android/API level, build, screen, battery, memory, and capability flags. Use before automation that depends on device capabilities, or to report what a device actually is.

ilog3 Skill

Use when matching an education research topic, question, or variable model to appropriate theories, conceptual frameworks, and prior research traditions. Supports ResearchRabbit/Litmaps/Connected Papers-style citation exploration, OpenAlex/Semantic Scholar search, and theory-fit scoring.

ilog3 Skill

Use when generating, refining, comparing, or validating education research questions and hypotheses. Supports qualitative, quantitative, mixed-methods, intervention, design-based research, survey, classroom observation, and systematic review questions using Elicit-style evidence search, Semantic Scholar/OpenAlex, and LLM reasoning.

ilog3 Skill

Use when education research uses scales, questionnaires, tests, rubrics, or latent constructs and needs reliability and validity analysis, including Cronbach's alpha, McDonald's omega, item-total correlations, EFA, CFA, CR, AVE, convergent validity, discriminant validity, factor loadings, model fit, and measurement reporting. This second-layer execution skill supports quantitative, mixed-methods, psychometric, program evaluation, and survey studies.

ilog3 Bundle

Apply Herbert J. Rubin and Irene S. Rubin's responsive interviewing model to design, conduct, analyze, and report in-depth qualitative interviews. Use when selecting knowledgeable conversational partners, building credible iterative studies, combining main, follow-up, and probe questions, or developing concepts, themes, mechanisms, and theory from interview evidence.

ilog3 Bundle

Plan, conduct, report, and appraise educational research using the 2013 教育研究方法 textbook edited by Chen Xiangming and jointly written by major Chinese universities. Use when choosing among positivist, interpretive, and integrated approaches; developing a topic, literature review, design, data plan, and report; or applying philosophical inquiry, surveys, educational measurement, experiments, statistics, qualitative inquiry, narrative inquiry, case study, historical research, action research, or comparative education. Match method to question and paradigm, make evidence and inference auditable, and verify current statistical, ethical, reporting, and policy standards.

samraddhashrivastavatech Skill

Builds local desktop UIs. Use for tkinter or when the user wants a desktop app. Native Windows admin is a different MCP.

ilog3 Bundle

Analyze informal groups, leadership, status, mobility, politics, and community organization through William Foote Whyte's Street Corner Society, and apply its long-term participant-observation lessons. Use for urban ethnography, community field entry, work with key informants, relational mapping, observation of interaction patterns, iterative redesign, researcher participation, reciprocity, and ethical critique. Do not generalize Cornerville as a timeless model of poverty or treat Whyte's contested field relations as a universal ethical template.

samraddhashrivastavatech Skill

Handles the CATLX memory architecture: episodic, semantic, workspace, and knowledge-graph memory stores, the Memory Broker as the single access point, the coherence protocol, memory compression, RAG-based retrieval optimization, and long-term persistence/portability. Use when the user asks about CATLX memory, remembering conversations, facts/preferences, workspace state, knowledge graphs, conflict resolution between memories, or how memory feeds AI calls.

samraddhashrivastavatech Skill

Safely inspect and transform local media through the media-master MCP using capability checks, exact plans, approval, bounded jobs, and output verification.

samraddhashrivastavatech Bundle

Analyze raster edges and create verified hybrid/vector EPS, SVG, PDF, or PNG through the Hybrid EPS MCP.

samraddhashrivastavatech Skill

Persistent, interruptible Windows perception and control through MCP; use only when GUI interaction is necessary.

samraddhashrivastavatech Bundle

Discover and compose safe raster, vector, CV, OCR, video, procedural and optional model capabilities.

samraddhashrivastavatech Skill

Handles the CATLX AI provider architecture: the Provider Abstraction Layer (PAL), current and future providers, provider routing and failover protocol, the cost optimization router, and local LLM integration (llama.cpp/vLLM with OpenAI-compatible API). Use when the user asks about which AI provider CATLX uses, provider failover, routing policies, the cost router, local models, embeddings, or how AI inference is abstracted.

samraddhashrivastavatech Skill

Context-safe chunking for large/small-model docs (`pdf-chunk`)

samraddhashrivastavatech Bundle

Safely reason about and operate VMs through the virtualization-master MCP with explicit HOST/GUEST context, exact approval, readiness checks, and independent verification.

samraddhashrivastavatech Skill

Handles the CATLX security architecture: least-privilege model, the credential vault (DPAPI on Windows), Windows Credential Manager integration, the permission router, the capability firewall (AppContainer + Windows Filtering Platform), the runtime sandbox, audit logs, risk detection, and safe/recovery modes. Use when the user asks about CATLX security, credentials, permissions, sandboxing, audit logging, risk detection, safe mode, or how CATLX protects privileged operations.

samraddhashrivastavatech Skill

Handles the CATLX Workflow Engine: the DAG execution model, the four scheduling modes (immediate/deferred/event/background), checkpointing, replay, rollback, multi-layer recovery and resilience, state machines, and the YAML workflow definition DSL. Use when the user asks about CATLX workflows, running tasks in steps, scheduling, checkpoint/resume, replaying/rolling back a run, or defining a workflow in YAML.

samraddhashrivastavatech Skill

android_ui

samraddhashrivastavatech Skill

evidence-extraction

samraddhashrivastavatech Skill

government-research

samraddhashrivastavatech Skill

android_remote_touchpad

samraddhashrivastavatech Skill

competitive-research

samraddhashrivastavatech Skill

contradiction-search

samraddhashrivastavatech Skill

presentation-research

samraddhashrivastavatech Skill

android_adb

samraddhashrivastavatech Skill

Handles CATLX hardware adaptation: hardware tier detection (T0/T1/T2/T3), the capability matrix, the boot-time hardware scan, the Capability Router state machine that emits the CapabilityMap, and runtime re-adaptation when resources change. Use when the user asks about hardware tiers, capability scaling, the 50ms boot scan, CapabilityMap fields, adaptive reconfiguration, or how CATLX picks OCR/STT/TTS/LLM backends per hardware.

samraddhashrivastavatech Skill

android_apps

samraddhashrivastavatech Skill

android_files

samraddhashrivastavatech Skill

android_input

samraddhashrivastavatech Skill

android_shell

tristan-hauber Bundle

Capture one compact, evidence-bounded record of an agent, skill, review, validation, or workflow failure for later improvement assessment. Use when a user explicitly asks to report feedback, or when current-run evidence shows a material missed safeguard, incorrect claim, unsupported reversal, wrong workflow result, or avoidable workflow inefficiency.