Snowflake-Labs

Identify SQL Server migration anti-patterns — performance blockers, architecture/security considerations, and behavioral differences — by running `scai assessment anti-patterns`, which buckets SnowConvert's already-recorded issue codes by priority and writes JSON for the parent assessment multi-report. Use for anti-patterns, performance/architecture concerns in SQL Server → Snowflake migrations;

Analyze SSIS packages from SnowConvert ETL.* outputs and source .dtsx files. Classifies packages, scores migration complexity, and produces JSON for the parent assessment multi-report.

Analyze Informatica Power Center workflows/mappings from SnowConvert ETL.* outputs and source XML files. Classifies workflows, scores migration complexity, and produces JSON for the parent assessment multi-report.

Analyzes workloads to be migrated to Snowflake using SnowConvert assessment reports. Routes to specialized sub-skills for high-quality assessments. Use this skill when user wants to do an assessment of their code or ETL workload, waves generation, object exclusion, anti-patterns, sql dynamic and/or ETL analysis (SSIS)

Set up a migration project — connect to source, initialize, extract objects, convert, and assess. Covers steps 1-5 of the migration lifecycle.

End-to-end database migration to Snowflake. Orchestrates the full migration lifecycle from source connection through initial conversion. Triggers: migrate, migration, migrate to snowflake, end to end migration, e2e migration, full migration.

Convert source code to Snowflake SQL using SnowConvert. Transforms full-migration sources and code-conversion-only sources to Snowflake-compatible syntax, and optionally repoints Power BI reports. Triggers: convert, snowconvert, transform code, convert to snowflake, power bi.

Use when the migration state machine emits the `seedScript` task for a BTEQ script unit (`kind: "script"`, `source.format: "bteq"`), before baseline capture.

Use this skill when the user asks to "prepare an RFP response", "draft an RFP", "respond to an RFP", or needs to compile firm capabilities, track record, investment process, and compliance content for a proposal. Also use when the user mentions "Request for Proposal", "mandate proposal", "tender response", or "client proposal".

Use this skill when the user asks about a borrower's default probability, credit risk drivers, PD scores, risk ratings, or wants to run what-if scenarios on credit metrics like "what if revenue drops 20%" or "recalculate PD with adjusted inputs". Also use when explaining why a borrower received a particular risk rating, SHAP analysis, or comparing risk across borrowers.

Use this skill when the user asks which factors are driving returns, wants to compare factor performance across periods, or requests ad-hoc factor exploration like "which factors are working this quarter", "compare momentum vs value", or "factor correlations". Also use when the user needs analysis beyond what pre-computed factor views provide.

Use this skill when the user wants to optimize portfolio weights, find the efficient frontier, construct a maximum Sharpe or minimum variance portfolio, or rebalance with specific constraints like position caps, sector limits, or ESG minimums. Also use when the user says "what's the optimal allocation" or "rebalance for minimum risk".

Use this skill when the user asks about anomalies, risk flags, style drift, attribution drift, concentration alerts, or anything unusual in attribution patterns. Also use proactively when generating reports to check for HIGH severity items before presenting findings.

Use this skill when the user asks to "generate an attribution report", "prepare attribution for the board", "write up quarterly performance", or wants a comprehensive multi-section attribution narrative. Orchestrates multiple tools into a complete report with audience-appropriate depth.

Use this skill to adapt response depth, formatting, and language to the target audience. Activate when the user specifies an audience like "board memo", "client letter", "PM deep-dive", or when generating reports for different stakeholders, even if they don't explicitly name an audience tier.

Legacy skill — superseded by multi-level-attribution. This skill redirects to multi-level-attribution which provides full sector/country/industry drill-down, linked period attribution (QTD/YTD/12M), factor cross-reference, and interactive stopping points.

Primary one-command installer for the synapse-based, vehicle/industry-AGNOSTIC fleet analytics architecture on Snowflake: the FLEET_SA_APP consumer app, the FLEET_ADMIN_APP build console, the three role-scoped synapse MCP tool bundles, the neutral FLEET_APP data contract, roles, Cortex agents, and (by default) the live ORS/VROOM routing engine. Use when: install the fleet apps, deploy FLEET_SA_APP and FLEET_ADMIN_APP, set up the agnostic fleet analytics solution, install the synapse fleet stack, provision the routing engine. Self-owning and idempotent: detects-and-reuses-else-creates SPCS infra, seed data, and the ORS engine; pass --no-engine to skip the heavy engine build. Do NOT use for: changing routing maps/profiles (use routing-customization) or the legacy per-vehicle demo skills (fleet-intelligence-car, route-deviation, etc.). Triggers: install fleet apps, install-fleet-apps, deploy fleet sa app, deploy fleet admin app, install synapse fleet stack, set up agnostic fleet solution, provision routing engin

Region-generic Emergency Response evacuation-planning demo: a multi-step wizard in the FLEET_SA_APP that, for whatever region/dataset is active, colors the map by procedural hazard-zone risk (flood/wildfire), seeds participants inside the drive-time isochrone union of the region's health-anchor care centers, and solves a capacitated multi-depot evacuation VRP. Data is produced by Data Studio (generates_hazard + generates_anchors + generates_participants); the wizard binds to the FLEET_APP.EMERGENCY_RESPONSE contract + evac_seed/evac_solve verbs. Use when: setting up emergency response demo, evacuation planning, hazard risk, flood or wildfire exposure, paratransit evacuation routing, participant pickup optimization. Do NOT use for: standard fleet tracking (use fleet-intelligence-car), retail trade area analysis (use retail-catchment), route deviation analytics (use route-deviation), generic dispatch (use route-optimization). Triggers: emergency response, evacuation planning, hazard risk, wildfire risk, flood r

Deploy the routing TOOL_* procedures (AI-geocoded directions, isochrones, POI-in-isochrone, optimization, catchment, delivery/network demos) into FLEET_INTELLIGENCE.ROUTING_TOOLS. These are wrapped by the synapse ROUTING_MCP verbs that the app-level FLEET_AGENT attaches - there is no separate standalone routing agent. Use when: setting up the ORS routing tool substrate the fleet app depends on, integrating directions/isochrones/optimization with Cortex. Do NOT use for: deploying fleet intelligence demos, route deviation analysis, or changing ORS configuration. Triggers: openrouteservice demo, routing tools, ORS tool procedures, routing substrate.

Deploy the Backload Matching Engine demo: a fleet-wide VRP solve over idle-bound trailers + internal volumes + external freight-exchange offers, anchored on the OPENROUTESERVICE_APP.CORE.OPTIMIZATION function. The page picks one or many trailers, calls OPTIMIZATION once, and renders empty/loaded legs, KPI savings, and a Cortex rationale. Use when: setting up the DHL Freight backload demo, asset velocity / trailer rotation use cases, freight-exchange aggregation, internal-first vs external-second proposals, multi-trailer joint dispatch. Do NOT use for: route optimization VRP from PLACES (use route-optimization), route deviation analysis (use route-deviation), retail catchment (use retail-catchment), fleet intelligence car/e-bike demos, or single-leg directions tests (use FunctionTester). Triggers: backload, backload matching, empty mile, empty leg, asset velocity, trailer rotation, freight exchange, freight exchanges, idle trailer, idle-bound trailer, Timocom, WTransnet, Teleroute, B2P, DHL, DHL Freight, dispa

Deploy the Freight Exchange page: a dispatcher-grade marketplace cockpit that mirrors what users do today on Timocom, WTransnet, Teleroute, B2P (NA: DAT, Truckstop, Convoy, Uber Freight). Adds a parallel React route to the ORS Control App that lets dispatchers browse, filter, and inspect synthesized freight offers per active preset, with trust scores (credit/KYC/blacklist), market-rate badges (vs. weekly p25/p50/p75 USD/km from a RATE_INDEX dynamic table), and partner lane history. Reads from FLEET_INTELLIGENCE.MARKETPLACE projection views over per-preset SYNTHETIC_DATASETS.UNIFIED data, so switching preset auto-refreshes the page. Use when: setting up the Freight Exchange demo, dispatcher cockpit, freight marketplace, load board, trust-score badges, market-rate badges, Timocom-style UI parity. Do NOT use for: solver-led fleet-wide backload (use backload-matching), route-optimization VRP, route-deviation, retail-catchment, fleet-intelligence car/e-bike demos. Triggers: freight exchange, freight marketplace, l

Deploy the Route Optimization demo including Marketplace data and notebook. Use when: setting up the route optimization demo after ORS app deployment. Do NOT use for: fleet intelligence demos (use fleet-intelligence-car), route deviation analysis (use route-deviation), or retail catchment analysis. Triggers: deploy route optimization demo, setup route optimization demo, run route optimization demo.

Deploy the Dwell & Congestion Analysis demo: create a 12-step Dynamic Table pipeline for state detection, dwell sessionization, H3 congestion heatmaps, SLA alerts, facility utilization, and daily trends. Works with any vehicle type from SYNTHETIC_DATASETS.UNIFIED, configured via CONFIG table. Use when: setting up dwell analysis demo, congestion analytics, SLA breach monitoring, facility utilization tracking. Do NOT use for: route deviation analysis (use route-deviation), e-bike fleet (use fleet-intelligence-ebike), car fleet (use fleet-intelligence-car). Triggers: deploy dwell analysis, dwell analytics, congestion analysis, SLA alerts, facility utilization, dwell demo, H3 heatmap.

Generate realistic vehicle/driver location data for the Fleet Intelligence solution using Overture Maps data and OpenRouteService for actual road routes. Also supports Data Studio projection views from SYNTHETIC_DATASETS.UNIFIED for any vehicle type via CONFIG table. Configurable location (New York, London, San Francisco, etc.), number of drivers (default 80), days of simulation (default 1), and shift patterns. Use when: setting up driver location data, generating route-based simulation, deploying the fleet dashboard. Do NOT use for: e-bike fleet simulation (use fleet-intelligence-ebike), route deviation analysis (use route-deviation), or route optimization demos. Triggers: generate driver locations, create driver data, setup fleet data, fleet intelligence dashboard, car fleet simulation.

Manage the end-to-end lifecycle of the Snowflake and SAP BDC Zero-Copy Integration and connector. Use when: consuming SAP data products in Snowflake, publishing Snowflake databases to SAP BDC with minimal CSN (v1.0, SDK-compatible), analyzing shared SAP data, or troubleshooting SAP BDC connector issues. This version uses MINIMAL CSN generation only (no options, no reviews, no validation loops). Triggers: SAP BDC, SAP connector, minimal CSN, SDK CSN, zerocopy connector, SAP data product, SAP BDC Connect, SAP publish, SAP share, SAP troubleshoot.

Audits all (or selected) Cortex Agents in a Snowflake account. For each agent the skill reads the live specification, extracts every dependency (semantic views, Cortex Search services, UDFs, warehouses, tables), then checks whether a user-supplied role holds the required privilege on each object — including inherited grants via role hierarchy. Results are presented as a gap table grouped by agent, followed by an optional one-click remediation script. Triggers: check agent access, audit agent privileges, who can use my agents, role access to agents, missing agent grants, grant agent usage, agent USAGE check, PUBLIC can't call agent, fix agent permissions, check role permissions for Cortex Agent. Do NOT use for: general RBAC design (use `rbac`), warehouse credit audits, or auditing non-agent Snowflake objects.

Generate SnowPro certification study materials including study plans, domain deep-dives, flashcards, and practice exam questions for any SnowPro exam. Dynamically discovers current certifications and exam versions from Snowflake. Use when asked to study, prep for, or generate content for SnowPro certifications.

Use for ALL requests that mention: ontology, knowledge graph, semantic layer from schema, ontology-on-snowflake, abstract views, ontology metadata tables, KG_NODE, KG_EDGE, build ontology stack, generate ontology from tables, ontology layer generation, map tables to ontology, generate cortex agent config. Triggers: build ontology, create ontology stack, schema-to-ontology, OWL import, knowledge graph on snowflake, create semantic models from schema. Do NOT use for: simple semantic view creation (use semantic-view skill), standalone Cortex Agent creation (use cortex-agent skill).

Use when migrating a Snowflake Data Clean Room from V1 (SAMOOHA Provider/Consumer API) to V2 (Collaboration API). Discovers V1 setup via introspection queries, maps constructs to Collaboration API equivalents, converts JinjaSQL templates, and generates provider/consumer setup plus cleanup scripts. Output is local files only — no live DDL/DML. Triggers: DCR migration, V1 to V2, clean room upgrade, clean room migration, migrate DCR, upgrade clean room, SAMOOHA to Collaboration API.

Use when a developer or data engineer wants to assess MLOps maturity, design a promotion strategy (Code/Model/Hybrid), or implement MLOps capabilities (CI/CD, monitoring, retraining, governance) on Snowflake for traditional ML or LLM/GenAI workloads. Triggers: mlops, mlops maturity, mlops assessment, mlops strategy, mlops pattern, mlops framework, model promotion, ml ci/cd, ml monitoring, llmops, rag pipeline ops, fine-tuning ops.

Use when configuring private connectivity from Snowflake OpenFlow (running on SPCS) to AWS-hosted or on-premises data sources via AWS PrivateLink. Covers NLB + VPC Endpoint Service on AWS, then SYSTEM$PROVISION_PRIVATELINK_ENDPOINT, network rules, and External Access Integration on Snowflake. AWS only (Public Preview also exists on Azure but is out of scope here). Triggers: OpenFlow PrivateLink, SPCS private connectivity, NLB for OpenFlow, EAI for OpenFlow, SYSTEM$PROVISION_PRIVATELINK_ENDPOINT, OpenFlow RDS, OpenFlow MySQL, OpenFlow PostgreSQL, private ingestion OpenFlow, SPCS egress.

Use when a user pastes a Snowflake Quickstart URL or asks to learn/walk through a Quickstart. Fetches the source content from GitHub, parses it into stages, checks the learner's environment, and delivers a guided build experience with two modes (learner or builder). Triggers: quickstarts.snowflake.com, snowflake.com/en/developers/guides/, walk me through this quickstart, teach me this quickstart, learn this guide, I want to do this quickstart. Do NOT use for general SQL help or non-Quickstart tutorials.

Create, evaluate, and optimize custom AI functions using Snowflake Cortex AI Complete. Supports text, image, and document inputs. Use when: building LLM-powered functions, evaluating AI function performance, tuning prompts, selecting models, checking async job status. Triggers: ai function builder, custom ai function, user defined ai function, build my own llm function, evaluate ai function, tune ai function, optimize ai function, demo ai function, resume ai function job, image classification, document analysis, multimodal ai function.

Create a new custom AI function. Supports table-based or manual input specification, single or variant outputs. Direct AI_COMPLETE calls or additional pre- and post-processing.

Quick Start demo: Build a toxicity classifier and evaluate it — the fastest way to experience the core create → evaluate workflow.

Interactive demo: Generate pseudo-labels from a strong teacher model, build a cheap student function, and evaluate accuracy. Showcases pseudo-labeling and teacher-student distillation.

Interactive demo: Build a legal contract field extractor and create a weighted composite metric that scores 4 fields independently. Showcases custom evaluation metrics for multi-field AI functions.

Interactive demo: Extract structured fields from SEC 10-K filing PDFs using multimodal AI, create a custom composite metric for per-field scoring, and evaluate extraction accuracy with per-field analysis.

Interactive demo: Build a policy-conditioned ticket router where a seed prompt performs poorly, then watch prompt optimization close the accuracy gap through prompt evolution and Pareto cost/quality analysis. The canonical demo for prompt optimization.

Interactive demo [Experimental]: Build a PII redaction function using Agent Research mode — the agent searches the web for techniques, proposes SQL UDF architectures with pre/post-processing, and you pick the approach. Then optimize the full function body.

Set up real-time ADS-B aircraft position ingestion from adsb.lol: create tables, network rules, external access integrations, Python ingestion procedures, flight-matching enrichment, scheduled tasks, and optional historical backfill. Subskill of aviation-installer — must be invoked from the router, not independently. Use when: configuring ADS-B data pipeline as part of installation. Do NOT use for: standalone execution, flight schedules (use flight-schedules), derived analytics (use derived-analytics). Triggers: ADS-B ingestion, aircraft tracking setup, adsb.lol, realtime aircraft data.

Create airport database infrastructure, schemas, cost-attribution tags, airport properties from Overture Maps, gate and runway geometry, and airline reference dimension. Subskill of aviation-installer — must be invoked from the router, not independently. Use when: setting up base airport infrastructure as part of installation workflow. Do NOT use for: standalone execution, ADS-B ingestion (use adsb-ingestion), derived analytics (use derived-analytics). Triggers: base setup, airport infrastructure, airport properties, setup airport database.

Create the airport analytics Dynamic Table pipeline: gate analysis (6 DTs), traffic facts (4 DTs), runway crossings, flight tracker, monitoring views, operational KPIs, and the full task DAG. Subskill of aviation-installer — must be invoked from the router after ADS-B ingestion completes. Use when: deploying derived analytics layer as part of installation. Do NOT use for: standalone execution, base setup, ADS-B ingestion. Triggers: derived analytics, deploy derived analytics, dynamic tables, setup dynamic tables, gate analysis, runway crossings, runway crossings detection, traffic analysis, airport analytics pipeline, DT pipeline.

Set up Aviationstack flight schedule ingestion: create schedule tables, network rule, external access integration with API key secret, Python ingestion procedure, ETL procedure, and scheduled task. Conditional subskill of aviation-installer — only runs when user provides an Aviationstack API key. Use when: configuring flight schedule data as part of installation with API key. Do NOT use for: standalone execution, ADS-B ingestion (use adsb-ingestion), derived analytics (use derived-analytics). Triggers: flight schedules, Aviationstack setup, schedule ingestion, flight timetable.

Install and configure an airport analytics platform in Snowflake. Routes to sub-skills for base infrastructure setup, ADS-B real-time ingestion, flight schedule ingestion, TSA checkpoint throughput ingestion, and derived analytics pipelines. Use when: installing airport analytics, setting up a new airport, deploying aviation platform, provisioning airport database. Do NOT use for: deploying the Streamlit dashboard (use aviation-dashboard), cleaning up objects (use aviation-cleanup), viewing flight data. Triggers: install airport, setup airport, deploy aviation, provision airport, aviation installer, new airport setup, airport analytics platform.

Set up TSA checkpoint throughput ingestion: create PDF stages, throughput table, network rule, external access integration, Python PDF-fetch and AI_EXTRACT procedures, and weekly scheduled tasks. Optional subskill of aviation-installer. Use when: configuring TSA throughput data as part of installation. Do NOT use for: standalone execution, ADS-B ingestion (use adsb-ingestion), flight schedules (use flight-schedules). Triggers: TSA throughput, passenger throughput, checkpoint data, TSA pipeline.

Deploy the Airport Analytics dashboards: Streamlit-in-Snowflake plus the React dashboard via SPCS (Docker build + CREATE SERVICE). Create the Streamlit app, build/push the Docker image, create the SPCS service, and verify the URLs. Use when: deploying dashboard, setting up airport analytics UI, installing flight tracker, monitoring page, SPCS dashboard, React dashboard. Do NOT use for: installing airport data pipeline (use aviation-installer), cleaning up objects (use aviation-cleanup). Triggers: deploy dashboard, aviation dashboard, airport analytics UI, streamlit airport, install dashboard, flight tracker app, react dashboard, SPCS dashboard.

Discover and remove all Snowflake objects created by the aviation-ops-intelligence skills. Uses the COMMENT tracking tag (sf_sit-is-aviation) to find objects, generates DROP statements in dependency-safe order, and optionally executes them. Use when: cleaning up after a demo, removing airport analytics objects, tearing down an environment, removing a specific airport. Do NOT use for: dropping objects not created by these skills, production cleanup without review, provisioning or deploying airports. Triggers: aviation-cleanup, cleanup airport, teardown aviation, remove airport analytics, remove airport objects, drop aviation objects, reset aviation environment.

Draft on-brand marketing copy – product announcement emails and slide outlines – that follows Peak Outfitters' brand voice and style guidelines. Use whenever the user wants to write marketing content, a launch email, an announcement, or a slide/deck outline, even if they don't mention the brand explicitly. Triggers: brand content, marketing copy, launch email, announcement, deck outline, slide outline, on-brand, style guide.

Detect anomalies in hot food sales data. Use this skill when the user asks about unusual patterns, spikes, drops, outliers, or unexpected changes in revenue, quantity sold, or transaction counts across stores, categories, states, or items.

Generate a comprehensive sales report by store or by state. Use this skill when the user asks for a sales report, performance summary, store report, state report, executive summary, or full analysis of a specific store or state.

**[REQUIRED]** Use for ALL tasks combining Snowflake Cortex AI functions with Ibis (Python dataframe library): AI_COMPLETE, AI_SENTIMENT, AI_CLASSIFY, AI_EXTRACT, AI_FILTER, AI_SUMMARIZE_AGG, SNOWFLAKE.CORTEX.COMPLETE, SNOWFLAKE.CORTEX.EMBED_TEXT_768/1024, vector similarity search, semantic search, EnrichmentPipeline, direct Cortex REST API calls (PAT auth, JWT auth, streaming SSE, tool calling), cortex_ibis, cortex_rest.

Analyze Snowflake AI cost data and generate insights and reports.

Deploy AI Cost Dashboard stored procedures and Streamlit app to Snowflake.

Build and manage a Snowflake AI Cost Dashboard using stored procedures and Streamlit. Use for **ALL** requests involving: AI cost, AI spending, cost dashboard, spending dashboard, Snowflake AI credits, Cortex cost tracking. DO NOT attempt AI cost dashboard work manually - invoke this skill first.

Generate Cortex REST API usage report as PDF. Use when: user wants usage report, token consumption report, API usage summary, cortex usage PDF. Triggers: usage report, cortex usage, token report, API consumption.

Interactive tutorial teaching Snowflake Cortex Agents. Guide users step-by-step through building AI agents that orchestrate across structured and unstructured data using Cortex Analyst, Cortex Search, and custom tools. Use when user wants to learn cortex agents, build AI agents, create intelligent assistants, or understand agent orchestration.

Advanced tutorial teaching Snowflake Cortex Agents with custom tools. Extends the base tutorial by adding stored procedure-based tools (inventory lookup, price calculator, product summary) alongside Cortex Analyst and Cortex Search. Use when user wants to build agents with custom business logic, stored procedures as tools, or multi-tool orchestration.

Interactive tutorial that teaches Cortex Code CLI from scratch. Guides non-technical users step-by-step through slash commands, skills, special syntax, sessions, and more. Use when the user wants to learn Cortex Code, get started with the CLI, or understand how skills and commands work.

Interactive tutorial teaching Snowflake Cortex Search for building semantic search and RAG applications. Guide users through creating search services and querying them using SQL and Python. Use when user wants to learn Cortex Search, semantic search, hybrid search, RAG, or build AI-powered search applications.

Deploy the Cortex CLASSIFY_TEXT tutorial notebook to the user's Snowflake account and provide a link to open it in Snowsight. Use when user wants to learn text classification through a Jupyter notebook experience.

Interactive tutorial teaching Snowflake Cortex CLASSIFY_TEXT for categorizing unstructured text. Guide users through classifying customer reviews using Python and SQL. Use when user wants to learn text classification, Cortex LLM functions, or analyze unstructured feedback data.

Interactive tutorial that teaches Snowflake Dynamic Tables hands-on. The agent guides users step-by-step through building data pipelines with automatic refresh, incremental processing, and CDC patterns. Use when the user wants to learn dynamic tables, build a DT pipeline, or understand DT vs streams/tasks/materialized views.

Audit an existing prompt for IDD alignment. Scores it across four dimensions (Goal, Requirements, Constraints, Output), identifies gaps, and offers to hand off to the rewrite skill. Use when: "evaluate this prompt", "audit my prompt", "is this IDD", "score my prompt".

Calculate the Intent Compression Ratio (ICR) for a workflow, script, or CI/CD pipeline. Counts total operations vs intent expressions and classifies the result. Recommends what to compress next to increase agentic readiness. Use when: "measure ICR", "calculate ICR", "what is ICR of this workflow", "how much compression", "intent compression ratio".

Guided rewrite of a prompt using the IDD Goal/Requirements/Constraints/Output structure. Collects each section interactively, generates the rewritten prompt, and optionally writes it to a file. Use when: "rewrite this prompt", "improve my prompt", "apply IDD to this", "IDD-ify this prompt".

Apply Intent-Driven Development (IDD) principles from Kamesh Sampath's IDD series to prompts, workflows, and CI/CD automation. Evaluate existing prompts for IDD alignment, rewrite them using the Goal/Requirements/Constraints/Output structure, or measure the Intent Compression Ratio (ICR) of a workflow. Use when: "review my prompt", "is this IDD-compliant", "rewrite this prompt", "measure ICR", "what is my ICR", "IDD audit", "intent compression ratio", "apply IDD", "improve my cortex exec prompt", "IDD principles".

Guided 6-step scaffold: set up a new GitHub Actions CoCo agent project from the github-coco-agent template. Provisions Snowflake OIDC resources, sets GitHub secrets, and optionally pushes a sample app to trigger the scan->issue->fix automation loop end-to-end. Use when the user chose the GitHub path, or invoked $devops-coco-agents:scaffold-for-github directly.

Guided 6-step scaffold: set up a new GitLab CI CoCo agent project from the gitlab-coco-agent template. Provisions Snowflake OIDC resources, sets GitLab CI/CD variables, and optionally pushes a sample app to trigger the scan->issue->fix automation loop end-to-end. Use when the user chose the GitLab path, or invoked $devops-coco-agents:scaffold-for-gitlab directly.

Guided walkthrough for scaffolding a CoCo agent project on GitHub Actions or GitLab CI, provisioning Snowflake OIDC resources, and running the scan->issue->fix automation loop end-to-end. Use when: "scaffold a project", "scaffold from template", "set up coco agent", "scaffold github", "scaffold gitlab", "set up CI/CD automation", "automate scan-fix loop", "set up github-coco-agent", "set up gitlab-coco-agent".

Build and manage Meta Conversions API (CAPI) pipelines in Snowflake. Triggers: set up Meta CAPI, create CAPI pipeline, discover tables, find CAPI candidates, get signal recommendations, run pipeline, send events to Meta, check CAPI status, pipeline health, debug CAPI, CAPI errors, resend failed events.

One-command installer for the Snowflake vector-tile demo: dynamic ST_AsMVT tiles served by Martin on SPCS (arm 1), a precomputed PMTiles archive baked from the same data (arm 2), and native H3 aggregation for client-side deck.gl (arm 3). The viewer is Martin's built-in Web UI on the public ingress. Idempotent and self-owning: detect-reuse-else-create for the Snowflake Postgres + PostGIS instance, SPCS infra (image repo, compute pool, stages, secret, EAIs), and the Martin service. Use when: install the tileserver, deploy Martin vector tiles on Snowflake, set up the Snowflake vector-tile demo, serve ST_AsMVT tiles, dynamic tiles on Snowflake Postgres, bake PMTiles from Snowflake, PostGIS MVT function source. Pass --no-pmtiles to skip the heavy bake. Triggers: install tileserver, deploy-tileserver, deploy martin, martin vector tiles, snowflake vector tiles, ST_AsMVT tiles, PMTiles bake, dynamic tiles snowflake postgres, postgis tile server.

One-command installer for Snowflake-native geocoding (forward + reverse) built on the Overture Maps Marketplace share, with an optional libpostal SPCS service for international addresses. Deploys the GEOCODING.PUBLIC UDFs and stored procedures, auto-detecting whichever Overture Addresses share is installed. Idempotent and self-owning: detects-and-reuses existing objects. Use when: install geocoding, deploy geocoding, set up address geocoding, geocode addresses in Snowflake, reverse geocode lat/lon, build the geocoding service, deploy Overture geocoding. Do NOT use for: high-accuracy/routing-grade geocoding (use a partner solution) or changing the Overture data source. Triggers: install geocoding, deploy geocoding, set up geocoding, address geocoding, forward geocode, reverse geocode, geocode my addresses, Overture geocoding, libpostal, international geocoding.

Build a first-party identity resolution pipeline on any set of source tables in Snowflake. Discovers source schemas, maps columns to canonical identity fields, generates a fully incremental Dynamic Table pipeline with deterministic and probabilistic matching, graph resolution, a keyring, and a golden record with field-level survivorship. Use for: identity resolution, IDR, identity graph, customer deduplication, identity stitching, master customer ID, golden record, customer 360 identity, first-party data graph, merge duplicate customers, link customer records across systems. Triggers: identity resolution, build IDR, identity graph, deduplicate customers, stitch identities, master customer, golden record, link records, merge duplicates, resolve identities, first-party identity, 1PD identity, identity pipeline, customer dedup.

Build an end-to-end Cloudflare HTTP log analytics pipeline in Snowflake with a Cortex Agent. Ingests Logpush data from R2 into a parsed view, semantic view, and queryable agent. Use when: cloudflare logs, R2 logpush, http log analytics, web traffic agent, cloudflare to snowflake. Triggers: cloudflare, logpush, R2, http logs, web analytics agent.

Routes Snowflake-related operations to Cortex Code CLI for specialized Snowflake expertise. Use when user asks about Snowflake databases, data warehouses, SQL queries on Snowflake, Cortex AI features, Snowpark, dynamic tables, data governance in Snowflake, Snowflake security, or mentions "Cortex" explicitly. Do NOT use for general programming, local file operations, non-Snowflake databases, web development, or infrastructure tasks unrelated to Snowflake.

Routes Snowflake-related operations to Cortex Code CLI for specialized Snowflake expertise. Use when the user asks about Snowflake databases, warehouses, schemas, tables, SQL on Snowflake, Cortex AI features, Snowpark, dynamic tables, streams, tasks, governance, or Snowflake security. Do not use for general programming, local file operations, non-Snowflake databases, web development, or infrastructure unrelated to Snowflake.

Routes Snowflake-related operations to Cortex Code agent for specialized Snowflake expertise. Use when querying databases, checking data quality, or asking about Snowflake features.

Routes Snowflake-related operations to Cortex Code CLI for specialized Snowflake expertise. Use when user asks about Snowflake databases, data warehouses, SQL queries on Snowflake, Cortex AI features, Snowpark, dynamic tables, data governance in Snowflake, Snowflake security, or mentions "Cortex" explicitly. Do NOT use for general programming, local file operations, non-Snowflake databases, web development, or infrastructure tasks unrelated to Snowflake.

Cortex-specific behavioral guidelines — always-on inner constraint layer for all CocoPlus personas

Relevance-ranked context loading — Tier 2 async subagent (Haiku, <5s) that fires after Tier 1 deterministic checks in UserPromptSubmit. Injects ranked context from CocoGrove, CocoContext, Environment Inspector, Prompt Studio, and CocoDream.

Developer engagement observer — non-blocking, always-on observational layer that tracks Delegation Intensity, Review Depth, and Engagement Zone throughout a session. Summary surfaced at $ship and FULL checkpoints.