Shipped capability catalog

Installable operators for Warren.

RVBBIT ships a curated catalog with the extension. Packs are catalog rows until deployed; Warren installs the selected runtime or model backend and exposes its SQL operators.

Public packs44

SQL operators381

Runtime packs2

SQL deployment

CLI is optional once the catalog is seeded.

Fresh extension installs seed rvbbit.capability_catalog. A UI or SQL client can queue an install directly with deploy_catalog_capability. The CLI remains useful for publishing catalog changes, scaffolding packs, and local development workflows.

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'extract/gliner-medium-v2.1',
  target_selector => '{}'::jsonb
);

SELECT *
FROM rvbbit.warren_jobs
ORDER BY created_at DESC
LIMIT 5;

Showing 44 of 48 shipped catalog packs.Generated from ../rvbbit-sql/capabilities/catalog.json

mcp/apollopublic

Apollo.io (B2B Enrichment)

Apollo.io — B2B people/company search + firmographic enrichment. Needs APOLLO_API_KEY (paid plan).

mcp

Backend: runtime
Model: bundled runtime
Operators: 9
Resources: not declared

apollo_bulk_enrich_organizationsapollo_bulk_enrich_peopleapollo_enrich_companyapollo_enrich_personapollo_get_complete_organization_infoapollo_get_organization_job_postingsapollo_search_companiesapollo_search_news_articlesapollo_search_people

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/apollo',
  target_selector => '{}'::jsonb
);

summarize/bart-large-cnnpublic

BART Large CNN Summarizer

Local abstractive summarization for long notes, reports, tickets, and JSONB rows.

autosingle_gpuhf_backend

Backend: summarize_bart_cnn
Model: facebook/bart-large-cnn
Operators: 3
Resources: VRAM 1.9 GiBmodel 1.5 GiB

summarize_bartsummarize_row_localsummarize_text_local

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'summarize/bart-large-cnn',
  target_selector => '{}'::jsonb
);

rerank/bge-reranker-basepublic

BGE Reranker Base

Smaller BGE cross-encoder reranker for query/document scoring and row ranking.

autosingle_gpuhf_backend

Backend: rerank_bge_base
Model: BAAI/bge-reranker-base
Operators: 6
Resources: VRAM 1.3 GiBmodel 1.0 GiB

aboutmeansrerank_bge_base_rowrerank_bge_base_scoresemantic_matchessemantic_score

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'rerank/bge-reranker-base',
  target_selector => '{}'::jsonb
);

rerank/bge-reranker-v2-m3public

BGE Reranker v2 M3

Cross-encoder relevance scoring for query/document pairs.

autosingle_gpuhf_backend

Backend: rerank_bge_m3
Model: BAAI/bge-reranker-v2-m3
Operators: 6
Resources: VRAM 2.6 GiBmodel 2.1 GiB

aboutmeansrerank_rowrerank_scoresemantic_matchessemantic_score

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'rerank/bge-reranker-v2-m3',
  target_selector => '{}'::jsonb
);

embeddings/bge-small-en-v1.5public

BGE Small English Embeddings

CPU-friendly English sentence embeddings exposed as an Rvbbit backend.

autosingle_gpuhf_backend

Backend: embed_bge_small
Model: BAAI/bge-small-en-v1.5
Operators: 3
Resources: VRAM 159 MiBmodel 127 MiB

embed_bge_smallsemantic_embedsimilar_to

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'embeddings/bge-small-en-v1.5',
  target_selector => '{}'::jsonb
);

embeddings/bge-m3public

BGE-M3 Embeddings

High-quality multilingual BGE-M3 sentence embeddings for retrieval and semantic search.

autosingle_gpuhf_backend

Backend: embed_bge_m3
Model: BAAI/bge-m3
Operators: 3
Resources: VRAM 2.6 GiBmodel 2.1 GiB

embed_bge_m3semantic_embedsimilar_to

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'embeddings/bge-m3',
  target_selector => '{}'::jsonb
);

mcp/bravepublic

Brave Search

Brave Search (official) — web/news/image/video/local search + AI summarizer. Needs BRAVE_API_KEY (free tier ~2k/mo).

mcp

Backend: runtime
Model: bundled runtime
Operators: 8
Resources: not declared

brave_image_searchbrave_llm_contextbrave_local_searchbrave_news_searchbrave_place_searchbrave_summarizerbrave_video_searchbrave_web_search

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/brave',
  target_selector => '{}'::jsonb
);

sql/core-workflowspublic

Core SQL Workflow Tests

SQL-only acceptance suite for rvbbit.flow, seeded operators, and SQL/code workflow nodes.

autosql_test_pack

Backend: runtime
Model: bundled runtime
Operators: 0
Resources: not declared

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'sql/core-workflows',
  target_selector => '{}'::jsonb
);

quality/data-contract-minerpublic

Data Contract Miner

Infer practical data quality checks from table samples or grouped column values.

cpuhf_backend

Backend: mine_data_contract
Model: rvbbit/data-contract-miner
Operators: 2
Resources: not declared

infer_column_contractinfer_table_contract

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'quality/data-contract-miner',
  target_selector => '{}'::jsonb
);

classify/deberta-v3-base-zero-shotpublic

DeBERTa v3 Base Zero-Shot Classifier

Smaller NLI zero-shot classifier for classifying text or JSONB rows into caller-provided labels.

autosingle_gpuhf_backend

Backend: classify_deberta_base_zero_shot
Model: MoritzLaurer/deberta-v3-base-zeroshot-v2.0
Operators: 6
Resources: VRAM 440 MiBmodel 352 MiB

bucketclassifyclassify_row_zero_shot_baseclassify_zero_shot_baseintentsemantic_classify

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'classify/deberta-v3-base-zero-shot',
  target_selector => '{}'::jsonb
);

classify/deberta-v3-zero-shotpublic

DeBERTa v3 Zero-Shot Classifier

NLI-based zero-shot classification over caller-provided labels.

autosingle_gpuhf_backend

Backend: classify_deberta_zero_shot
Model: MoritzLaurer/DeBERTa-v3-large-mnli-fever-anli-ling-wanli
Operators: 14
Resources: VRAM 1.0 GiBmodel 830 MiB

audiencebucketclassifyclassify_row_zero_shotclassify_zero_shotcomplexitycontradictsdomainformalityimpliesintentlooks_likesemantic_classifystance

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'classify/deberta-v3-zero-shot',
  target_selector => '{}'::jsonb
);

embeddings/e5-small-v2public

E5 Small Embeddings

Small, fast E5 embeddings useful for CPU-friendly retrieval experiments and lightweight semantic joins.

autosingle_gpuhf_backend

Backend: embed_e5_small
Model: intfloat/e5-small-v2
Operators: 3
Resources: VRAM 159 MiBmodel 127 MiB

embed_e5_smallsemantic_embedsimilar_to

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'embeddings/e5-small-v2',
  target_selector => '{}'::jsonb
);

classify/emotion-distilrobertapublic

Emotion Classifier

Classify text into emotion labels for support, survey, content, and conversation analytics.

autosingle_gpuhf_backend

Backend: classify_emotion_distilroberta
Model: j-hartmann/emotion-english-distilroberta-base
Operators: 4
Resources: VRAM 392 MiBmodel 313 MiB

classify_emotionclassify_row_emotionemotionemotion_confidence

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'classify/emotion-distilroberta',
  target_selector => '{}'::jsonb
);

mcp/exapublic

Exa (Neural Search)

Exa — neural/semantic web search + fetch; composes with embed() for KNN. Needs EXA_API_KEY.

mcp

Backend: runtime
Model: bundled runtime
Operators: 2
Resources: not declared

exa_web_fetch_exaexa_web_search_exa

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/exa',
  target_selector => '{}'::jsonb
);

finance/finbert-sentimentpublic

FinBERT Financial Sentiment

Financial-domain sentiment classifier for filings, news, analyst notes, and customer finance text.

autosingle_gpuhf_backend

Backend: finance_finbert
Model: ProsusAI/finbert
Operators: 3
Resources: VRAM 522 MiBmodel 418 MiB

finance_bearishfinance_sentimentfinance_sentiment_label

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'finance/finbert-sentiment',
  target_selector => '{}'::jsonb
);

mcp/firecrawlpublic

Firecrawl (MCP)

Firecrawl MCP server (firecrawl-mcp) over stdio via npx — web scrape/search/crawl/extract. PAT read from FIRECRAWL_API_KEY, resolved by the gateway from its encrypted secret store; only the ${FIRECRAWL_API_KEY} ref is stored in PG.

mcp

Backend: runtime
Model: bundled runtime
Operators: 20
Resources: not declared

firecrawl_agentfirecrawl_agent_statusfirecrawl_check_crawl_statusfirecrawl_crawlfirecrawl_extractfirecrawl_interactfirecrawl_interact_stopfirecrawl_mapfirecrawl_monitor_checkfirecrawl_monitor_checksfirecrawl_monitor_createfirecrawl_monitor_deletefirecrawl_monitor_getfirecrawl_monitor_listfirecrawl_monitor_runfirecrawl_monitor_updatefirecrawl_parsefirecrawl_scrapefirecrawl_searchfirecrawl_search_feedback

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/firecrawl',
  target_selector => '{}'::jsonb
);

data/fletch-data-moverpublic

Fletch Data Mover

Optional ADBC import/export utility sidecar powered by Fletch.

data_movercpuruntime_sidecar

Backend: fletch_data_mover
Model: early-signal-tech/fletch
Operators: 0
Resources: not declared

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'data/fletch-data-mover',
  target_selector => '{}'::jsonb
);

llm/gemma-4-12b-it-vllmpublic

Gemma 4 12B IT via vLLM

Warren-deployed local Gemma chat provider served through vLLM's OpenAI-compatible API.

cudagpusingle_gpullm_provider

Backend: gemma_4_12b_it
Model: google/gemma-4-12B-it
Operators: 0
Resources: VRAM 44.7 GiBmodel 22.4 GiB

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'llm/gemma-4-12b-it-vllm',
  target_selector => '{"gpu":true}'::jsonb
);

mcp/githubpublic

GitHub (MCP)

GitHub MCP server (@modelcontextprotocol/server-github) over stdio via npx. PAT read from GITHUB_PERSONAL_ACCESS_TOKEN, resolved by the gateway from its GITHUB_TOKEN env; only the ${GITHUB_TOKEN} ref is stored in PG.

mcp

Backend: runtime
Model: bundled runtime
Operators: 26
Resources: not declared

github_add_issue_commentgithub_create_branchgithub_create_issuegithub_create_or_update_filegithub_create_pull_requestgithub_create_pull_request_reviewgithub_create_repositorygithub_fork_repositorygithub_get_file_contentsgithub_get_issuegithub_get_pull_requestgithub_get_pull_request_commentsgithub_get_pull_request_filesgithub_get_pull_request_reviewsgithub_get_pull_request_statusgithub_list_commitsgithub_list_issuesgithub_list_pull_requestsgithub_merge_pull_requestgithub_push_filesgithub_search_codegithub_search_issuesgithub_search_repositoriesgithub_search_usersgithub_update_issuegithub_update_pull_request_branch

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/github',
  target_selector => '{}'::jsonb
);

extract/gliner-medium-v2.1public

GLiNER Medium v2.1 Extraction

Zero-shot entity extraction over arbitrary user-provided labels.

autosingle_gpuhf_backend

Backend: extract_gliner
Model: urchade/gliner_medium-v2.1
Operators: 7
Resources: VRAM 931 MiBmodel 745 MiB

contains_entityextractextract_business_entitiesextract_entitiesextract_piiextract_row_entitieshas_pii

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'extract/gliner-medium-v2.1',
  target_selector => '{}'::jsonb
);

mcp/mapspublic

Google Maps

Google Maps Platform — geocode/reverse-geocode/places/distance-matrix/directions/timezone/elevation/weather (read-only). Needs GOOGLE_MAPS_API_KEY.

mcp

Backend: runtime
Model: bundled runtime
Operators: 18
Resources: not declared

maps_air_qualitymaps_batch_geocodemaps_compare_placesmaps_directionsmaps_distance_matrixmaps_elevationmaps_explore_areamaps_geocodemaps_local_rank_trackermaps_place_detailsmaps_plan_routemaps_reverse_geocodemaps_search_along_routemaps_search_nearbymaps_search_placesmaps_static_mapmaps_timezonemaps_weather

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/maps',
  target_selector => '{}'::jsonb
);

memory/hindsight-slimpublic

Hindsight Memory Service

Hindsight API service for agent long-term memory, backed by a dedicated schema in the RVBBIT Postgres database and RVBBIT-compatible embedding config where possible.

system_runtimecpuruntime_sidecar

Backend: hindsight_default
Model: vectorize-io/hindsight-api-slim
Operators: 4
Resources: not declared

hindsight_memory_recallhindsight_memory_reflecthindsight_memory_retainhindsight_memory_status

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'memory/hindsight-slim',
  target_selector => '{}'::jsonb
);

mcp/ipgeopublic

ipgeolocation.io

ipgeolocation.io — IP geolocation + timezone + VPN/security + currency + ASN. Needs IPGEOLOCATION_API_KEY.

mcp

Backend: runtime
Model: bundled runtime
Operators: 16
Resources: not declared

ipgeo_bulk_lookup_ipipgeo_bulk_parse_user_agentipgeo_bulk_security_checkipgeo_check_securityipgeo_convert_timezoneipgeo_get_abuse_contactipgeo_get_astronomyipgeo_get_astronomy_time_seriesipgeo_get_my_ipipgeo_get_timezoneipgeo_lookup_asnipgeo_lookup_companyipgeo_lookup_currencyipgeo_lookup_ipipgeo_lookup_networkipgeo_parse_user_agent

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/ipgeo',
  target_selector => '{}'::jsonb
);

mcp/iplocatepublic

IPLocate (IP Geolocation)

IPLocate.io — read-only IP geolocation/privacy(VPN)/network/company/abuse lookups. Works keyless at low quota; set IPLOCATE_API_KEY for volume.

mcp

Backend: runtime
Model: bundled runtime
Operators: 6
Resources: not declared

iplocate_lookup_ip_address_abuse_contactsiplocate_lookup_ip_address_companyiplocate_lookup_ip_address_detailsiplocate_lookup_ip_address_locationiplocate_lookup_ip_address_networkiplocate_lookup_ip_address_privacy

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/iplocate',
  target_selector => '{}'::jsonb
);

catalog/join-detectivepublic

Join Detective

Suggest join keys from two table samples by combining name and value overlap evidence.

cpuhf_backend

Backend: detect_join_keys
Model: rvbbit/join-detective
Operators: 3
Resources: not declared

best_join_conditionjoin_key_confidencesuggest_join_keys

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'catalog/join-detective',
  target_selector => '{}'::jsonb
);

extract/keyphrase-kbir-inspecpublic

KBIR Keyphrase Extraction

Local keyphrase extraction for abstracts, research notes, docs, and analyst narratives.

autosingle_gpuhf_backend

Backend: extract_keyphrase_kbir
Model: ml6team/keyphrase-extraction-kbir-inspec
Operators: 3
Resources: VRAM 596 MiBmodel 477 MiB

extract_keyphrasesextract_row_keyphrasestop_keyphrase

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'extract/keyphrase-kbir-inspec',
  target_selector => '{}'::jsonb
);

mcp/linearpublic

Linear

Linear (@tacticlaunch/mcp-linear) — issues, projects, teams, cycles, comments. Reads + writes via LINEAR_API_KEY.

mcp

Backend: runtime
Model: bundled runtime
Operators: 152
Resources: not declared

linear_addAttachmentlinear_addIssueLabellinear_addIssueToCyclelinear_addIssueToProjectlinear_addIssueToReleaselinear_addProjectMemberlinear_addProjectToInitiativelinear_addToFavoriteslinear_addUserToTeamlinear_archiveDocumentlinear_archiveInitiativelinear_archiveIssuelinear_archiveMilestonelinear_archiveProjectlinear_archiveReleaselinear_archiveReleasePipelinelinear_archiveReleaseStagelinear_archiveRoadmaplinear_archiveTeamlinear_archiveTemplatelinear_assignIssuelinear_completeCyclelinear_completeReleaselinear_convertIssueToSubtasklinear_createCommentlinear_createCyclelinear_createDocumentlinear_createInitiativelinear_createIssuelinear_createIssueFromTemplatelinear_createIssueRelationlinear_createIssueTemplatelinear_createMilestonelinear_createProjectlinear_createProjectUpdatelinear_createReleaselinear_createReleaseNotelinear_createReleasePipelinelinear_createReleaseStagelinear_createRoadmaplinear_createSavedViewlinear_createTeamlinear_createTeamLabellinear_createWebhooklinear_createWorkflowStatelinear_deleteCommentlinear_deleteInitiativelinear_deleteReleaseNotelinear_deleteReleasePipelinelinear_deleteSavedViewlinear_deleteWebhooklinear_duplicateIssuelinear_getActiveCyclelinear_getAttachmentslinear_getAuthenticationSessionslinear_getCommentslinear_getCustomFieldslinear_getCycleByIdlinear_getCycleIssueslinear_getCycleslinear_getCycleStatslinear_getDocumentByIdlinear_getDocumentContentHistorylinear_getDocumentslinear_getFavoriteViewslinear_getInitiativeByIdlinear_getInitiativeProjectslinear_getInitiativeslinear_getIntegrationslinear_getIssueByIdlinear_getIssueCustomFieldslinear_getIssueHistorylinear_getIssueslinear_getIssueTemplateByIdlinear_getIssueTemplateslinear_getLabelslinear_getMilestoneByIdlinear_getMilestoneslinear_getNotificationslinear_getOrganizationlinear_getOrganizationAuditEventslinear_getProjectDocumentslinear_getProjectIssueslinear_getProjectMemberslinear_getProjectslinear_getProjectUpdateslinear_getRateLimitStatuslinear_getReleaseByIdlinear_getReleaseNoteByIdlinear_getReleaseNoteslinear_getReleasePipelineByIdlinear_getReleasePipelineslinear_getReleaseslinear_getReleaseStageslinear_getRoadmapByIdlinear_getRoadmapslinear_getSavedViewslinear_getServerStatuslinear_getSubscriptionslinear_getTeamLabelslinear_getTeamMembershipslinear_getTeamslinear_getTeamTemplateslinear_getUnreadNotificationCountlinear_getUserAuditEventslinear_getUserslinear_getViewerlinear_getWebhookslinear_getWorkflowStateslinear_logoutAllSessionslinear_logoutOtherSessionslinear_logoutSessionlinear_markAllNotificationsAsReadlinear_markNotificationAsReadlinear_removeFromFavoriteslinear_removeIssueFromCyclelinear_removeIssueFromProjectlinear_removeIssueFromReleaselinear_removeIssueLabellinear_removeProjectFromInitiativelinear_removeProjectMemberlinear_removeUserFromTeamlinear_searchDocumentslinear_searchIssueslinear_searchReleaseslinear_setIssuePrioritylinear_subscribeToIssuelinear_transferIssuelinear_unarchiveDocumentlinear_unarchiveInitiativelinear_unarchiveReleaselinear_unarchiveReleasePipelinelinear_unarchiveReleaseStagelinear_updateCommentlinear_updateCyclelinear_updateDocumentlinear_updateInitiativelinear_updateIssuelinear_updateIssueCustomFieldlinear_updateIssueTemplatelinear_updateMilestonelinear_updateProjectlinear_updateProjectUpdatelinear_updateReleaselinear_updateReleaseNotelinear_updateReleasePipelinelinear_updateReleaseStagelinear_updateRoadmaplinear_updateSavedViewlinear_updateTeamlinear_updateTeamMembershiplinear_updateWorkflowState

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/linear',
  target_selector => '{}'::jsonb
);

runtimes/python-runtimepublic

Managed CPython Runtime

Warren-managed CPython runtime for rvbbit operator `kind: python` nodes.

runtimeoperator_runtimecpuruntime_sidecar

Backend: python_default
Model: rvbbit/python-runtime
Operators: 0
Resources: not declared

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'runtimes/python-runtime',
  target_selector => '{"docker":true}'::jsonb
);

runtimes/mcp-gatewaypublic

MCP Gateway Runtime

Warren-managed MCP gateway for rvbbit operator `kind: mcp` nodes and SQL MCP calls.

runtimeoperator_runtimecpuruntime_sidecar

Backend: mcp_default
Model: rvbbit/mcp-gateway
Operators: 0
Resources: not declared

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'runtimes/mcp-gateway',
  target_selector => '{"docker":true}'::jsonb
);

rerank/ms-marco-minilm-l6-v2public

MS MARCO MiniLM Reranker

Lightweight cross-encoder reranker for query/document scoring and ranking table rows by relevance.

autosingle_gpuhf_backend

Backend: rerank_ms_marco_minilm
Model: cross-encoder/ms-marco-MiniLM-L6-v2
Operators: 6
Resources: VRAM 108 MiBmodel 87 MiB

aboutmeansrerank_ms_marco_rowrerank_ms_marco_scoresemantic_matchessemantic_score

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'rerank/ms-marco-minilm-l6-v2',
  target_selector => '{}'::jsonb
);

embeddings/openrouterpublic

OpenRouter Embeddings

Remote OpenAI-compatible embeddings through OpenRouter for faster hosted semantic search.

apihf_backend

Backend: embed_openrouter
Model: openai/text-embedding-3-small
Operators: 1
Resources: not declared

embed_openrouter

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'embeddings/openrouter',
  target_selector => '{}'::jsonb
);

mcp/openweatherpublic

OpenWeatherMap

OpenWeatherMap — current/forecast/hourly/daily/alerts/air-pollution + geocoding. Needs OPENWEATHER_API_KEY (32-hex). SDK pinned to 1.12.1 in args (avoids a FastMCP completions crash with sdk 1.29).

mcp

Backend: runtime
Model: bundled runtime
Operators: 11
Resources: not declared

openweather_geocode-locationopenweather_get-air-pollutionopenweather_get-current-air-pollutionopenweather_get-current-weatheropenweather_get-daily-forecastopenweather_get-hourly-forecastopenweather_get-location-infoopenweather_get-minutely-forecastopenweather_get-onecall-weatheropenweather_get-weather-alertsopenweather_get-weather-forecast

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/openweather',
  target_selector => '{}'::jsonb
);

security/prompt-injection-deberta-v3-base-v2public

ProtectAI Prompt Injection Detector

Local classifier for detecting prompt-injection and jailbreak attempts before they reach an LLM.

autosingle_gpuhf_backend

Backend: security_prompt_injection
Model: protectai/deberta-v3-base-prompt-injection-v2
Operators: 2
Resources: VRAM 880 MiBmodel 704 MiB

detect_prompt_injectionis_prompt_injection

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'security/prompt-injection-deberta-v3-base-v2',
  target_selector => '{}'::jsonb
);

classify/twitter-roberta-sentimentpublic

RoBERTa Sentiment Classifier

Practical positive/neutral/negative sentiment classifier for reviews, tickets, social text, and notes.

autosingle_gpuhf_backend

Backend: classify_twitter_roberta_sentiment
Model: cardiffnlp/twitter-roberta-base-sentiment-latest
Operators: 4
Resources: VRAM 597 MiBmodel 478 MiB

classify_row_sentimentclassify_sentimentsentimentsentiment_confidence

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'classify/twitter-roberta-sentiment',
  target_selector => '{}'::jsonb
);

qa/roberta-base-squad2public

RoBERTa SQuAD2 Extractive QA

Extractive question answering over a supplied context paragraph.

autosingle_gpuhf_backend

Backend: qa_roberta_squad2
Model: deepset/roberta-base-squad2
Operators: 2
Resources: VRAM 591 MiBmodel 473 MiB

answer_from_contextanswer_text

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'qa/roberta-base-squad2',
  target_selector => '{}'::jsonb
);

forecast/seasonal-trend-baselinepublic

Seasonal Trend Forecast

Lightweight local forecasting Warren for JSONB series produced by SQL aggregates.

cpuhf_backend

Backend: forecast_seasonal_trend
Model: rvbbit/seasonal-trend-baseline
Operators: 3
Resources: not declared

forecast_next_valueforecast_seriesforecast_values

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'forecast/seasonal-trend-baseline',
  target_selector => '{}'::jsonb
);

governance/semantic-column-typepublic

Semantic Column Type and PII Scanner

Classify column samples and redact obvious sensitive values from text.

cpuhf_backend

Backend: classify_semantic_column
Model: rvbbit/semantic-column-type
Operators: 3
Resources: not declared

classify_column_sampleclassify_column_typedetect_pii_text

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'governance/semantic-column-type',
  target_selector => '{}'::jsonb
);

llm/sqlcoder-7b-vllmpublic

SQLCoder 7B via vLLM

Warren-deployed local text-to-SQL provider with a Postgres SQL drafting operator.

cudagpusingle_gpullm_provider

Backend: sqlcoder_7b
Model: defog/sqlcoder-7b-2
Operators: 1
Resources: VRAM 26.1 GiBmodel 13.0 GiB

draft_postgres_sql

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'llm/sqlcoder-7b-vllm',
  target_selector => '{"gpu":true}'::jsonb
);

tables/tapas-wtqpublic

TAPAS Table Question Answering

Local question answering over small JSONB tables for BI-style table exploration.

autosingle_gpuhf_backend

Backend: table_qa_tapas
Model: google/tapas-base-finetuned-wtq
Operators: 2
Resources: VRAM 525 MiBmodel 420 MiB

answer_table_questiontable_answer_text

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'tables/tapas-wtq',
  target_selector => '{}'::jsonb
);

mcp/tavilypublic

Tavily (Web Search)

Tavily — web search / extract / crawl / map as read-only enrichment operators. Needs TAVILY_API_KEY.

mcp

Backend: runtime
Model: bundled runtime
Operators: 5
Resources: not declared

tavily_crawltavily_extracttavily_maptavily_researchtavily_search

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/tavily',
  target_selector => '{}'::jsonb
);

mcp/timepublic

Time / Timezone

time-mcp — timezone conversion, relative time, week-of-year, days-in-month. Deterministic local clock + IANA tz DB, no API key.

mcp

Backend: runtime
Model: bundled runtime
Operators: 6
Resources: not declared

time_convert_timetime_current_timetime_days_in_monthtime_get_timestamptime_get_week_yeartime_relative_time

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'mcp/time',
  target_selector => '{}'::jsonb
);

classify/toxic-bertpublic

Toxicity Classifier

Multi-label toxicity classifier for moderation, support triage, and data quality workflows.

autosingle_gpuhf_backend

Backend: classify_toxic_bert
Model: unitary/toxic-bert
Operators: 4
Resources: VRAM 522 MiBmodel 418 MiB

classify_row_toxicityclassify_toxicitytoxicitytoxicity_score

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'classify/toxic-bert',
  target_selector => '{}'::jsonb
);

sql/workflow-visualspublic

Workflow Visual Operator Tests

SQL-only acceptance suite for deterministic rvbbit.flow UI artifact operators.

autosql_test_pack

Backend: runtime
Model: bundled runtime
Operators: 11
Resources: not declared

action_buttonbar_chartbind_filterfilter_controllayout_gridline_chartmetric_cardscatter_plottable_viewtile_namevega_lite

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'sql/workflow-visuals',
  target_selector => '{}'::jsonb
);

classify/language-detection-xlm-robertapublic

XLM-R Language Detection

Detect the language of text for routing, normalization, and multilingual analytics.

autosingle_gpuhf_backend

Backend: detect_language_xlm_roberta
Model: papluca/xlm-roberta-base-language-detection
Operators: 2
Resources: VRAM 1.3 GiBmodel 1.0 GiB

detect_languagelanguage

SQL deployment

SELECT rvbbit.deploy_catalog_capability(
  catalog_id => 'classify/language-detection-xlm-roberta',
  target_selector => '{}'::jsonb
);