Tool Catalog

Once your assistant is connected, it can discover and call the tools below. This is a reference for what an agent can do to your account — the live catalog is whatever your client's listTools() (or the client's tool menu) returns, which is filtered to your permissions and the features your release has enabled.

Note

The exact set an assistant sees depends on your permissions, your subscription tier, and which features the connected release enables. Tools you aren't authorized for won't appear. Some tools are premium and return HTTP 402 until paid — see Payments.

Chat & conversations

| Tool | What it does | |------|--------------| | send_message | Send a message and get the assistant's reply (creates the transcript on first send). | | get_transcript | Read the messages in a chat. | | list_transcripts | List your chats. | | thread_list / thread_get | List and read conversation threads. |

Knowledge base & RAG

| Tool | What it does | |------|--------------| | search_knowledge | Search your knowledge base for relevant chunks. | | list_data_sources | List the RAG vector targets backing a workspace. | | get_document | Read an ingested document. | | rag_list_files / rag_file_status | List ingested files and check ingestion status. | | rag_delete_file | Remove a file from a RAG target. | | html_page_list / html_page_get / html_page_count / html_page_hosts / html_page_crawl_history | Inspect crawled HTML pages backing a knowledge base. |

Audio & transcription

| Tool | What it does | |------|--------------| | audio_upload_prepare / audio_upload_finalize | Two-step direct audio upload. | | audio_upload_to_rag | Upload audio and route its transcript into RAG. | | audio_transcript_upload_url | Get a signed URL for direct audio upload. | | audio_transcript_list / audio_transcript_get / audio_transcript_status / audio_transcript_delete | Manage audio transcripts. | | audio_transcript_generate_rag / audio_transcript_list_rag_targets | Turn transcripts into RAG content. |

Releases & configuration

| Tool | What it does | |------|--------------| | release_create / release_update / release_get / release_list / release_search | Manage releases (your deployable assistant configurations). release_update also authors the referenced sub-documents inline — threadPrefix, conversationStarters, chatDisclaimer, chatWelcomeMessage and chatCTA (see Chat CTAs). | | release_publish / release_archive | Publish or archive a release. | | release_derive_theme | Derive a chat theme from your storefront into a draft release. | | site_discover | List a site's pages (robots.txt → sitemap → sitemap index → page links). Free and read-only; creates no job. | | site_export / site_get | Screenshot a live URL and regenerate clean, editable HTML via a vision LLM (single-page returns HTML; multi-page via pageUrls from site_discover returns a jobId to poll). | | config_get | Read release configuration. | | export_config / import_config | Export or import a configuration bundle. | | mcp_server_create / mcp_server_get / mcp_server_list / mcp_server_update / mcp_server_delete | Attach and manage external MCP servers as assistant tools. |

Hosted Hermes agents

| Tool | What it does | |------|--------------| | hermes_create | Create a hosted Hermes agent — platform (vertex_ai/…, cloudflare/@cf/…) or BYOK model, optional persona, optional releaseId for per-turn RAG grounding. Returns the agent's hsk- proxy key. | | hermes_list | List the workspace's hosted agents (model, status, enabled). | | hermes_get | Read one agent, including its hsk- proxy key. | | hermes_chat | Send one message to an agent and get its reply (first turn after a sleep cold-boots the container). | | hermes_delete | Delete an agent and stop its container. | | hermes_proactive_set | Configure the proactive wake loop — whether it wakes, shadow vs live, when, how often (intervalHours), and its goals/memories. | | hermes_proactive_runs | Read recent wakes: what the agent produced, whether it was delivered, and what it cost. |

Phone billing

| Tool | What it does | |------|--------------| | phone_usage_voice | Unbilled LiveKit voice-call usage for a workspace — calls, cost (nanoUSD), oldest unbilled call. At-cost rates, no markup. | | phone_usage_sms | Unbilled Twilio SMS/MMS usage for a workspace — messages and cost (nanoUSD). At-cost rates, no markup. |

Skills

| Tool | What it does | |------|--------------| | skill_list / skill_get / skill_create / skill_update / skill_delete | Manage workspace skills. | | skill_connect_start / skill_connect_finish / skill_disconnect | Connect and disconnect skill integrations. | | skill_for_release | Read the skills active on a release. | | skill_get_catalog / skill_list_catalog | Browse the skill catalog. |

Fine-tuning & training data

| Tool | What it does | |------|--------------| | finetune_status / finetune_list / finetune_cancel | Track and manage fine-tuning jobs. | | finetune_files_copy_from / finetune_file_export_to_qa | Move training files between sources and into QA. | | training_data_generate / training_data_estimate_cost | Generate synthetic training data and estimate its cost. | | training_data_job_status / training_data_job_results / training_data_job_export | Track and export training-data jobs. |

Quality assurance & calibration

| Tool | What it does | |------|--------------| | qa_suites_list / qa_suite_run / qa_suite_copy_from | List, run, and copy QA suites. | | qa_run_status / qa_run_results | Track a QA run and read its results. | | qa_calibration_* | Submit ratings, pick winners, compute agreement, and manage calibration triples. |

Arena & A/B presets

| Tool | What it does | |------|--------------| | arena_preset_list / arena_preset_activate / arena_preset_update_variant | List, activate, and edit Arena comparison presets. |

Analytics, metrics & billing

| Tool | What it does | |------|--------------| | analytics_metrics / analytics_trends / analytics_funnel / analytics_performance / analytics_experiments | Query usage analytics across dimensions. | | usage_stats | Aggregate usage statistics. | | metrics_pipeline / metrics_alerts / metrics_alert_ack / metrics_alert_config_set / metrics_alert_config_delete | Inspect the metrics pipeline and manage alerts. | | billing_report | Pull a billing report. |

Notifications & flags

| Tool | What it does | |------|--------------| | notification_list / notification_count / notification_mark_read | Read and triage notifications. | | notification_channel_* | Create, list, test, verify, and delete delivery channels. | | notification_trigger_create / notification_trigger_list / notification_trigger_delete | Manage notification triggers. | | flagger_create / flagger_list / flagger_test / flagger_delete | Manage content/safety flaggers. |

Feedback

| Tool | What it does | |------|--------------| | feedback_list / feedback_get / feedback_stats | Read user feedback and its aggregate stats. |

Products & merchandising

| Tool | What it does | |------|--------------| | product_add / product_update / product_search | Manage and search the product catalog. | | product_merchandising_get / product_merchandising_set | Read and set merchandising config. | | product_prominence_pin / product_prominence_recompute / product_prominence_set_pages | Control product prominence weighting. | | product_calibration_set_cadence | Set the product-calibration cadence. |

Style patterns

| Tool | What it does | |------|--------------| | style_pattern_create / style_pattern_get / style_pattern_list / style_pattern_update | Manage tone/style rewrite rules. |

Groups & access

| Tool | What it does | |------|--------------| | group_create / group_get / group_list / group_delete | Manage groups. | | group_invite / group_remove_member | Add and remove group members. | | byok_list | List bring-your-own-key provider credentials. | | connector_list / connector_status / connector_stats | Inspect external connectors. connector_status takes a provider (or a connectionId, which it resolves). | | connector_sync / connector_sync_status / connector_sync_cancel | Run a sync, read its job history, stop what's running. | | connector_disconnect | Revoke a provider's connection. Not self-recoverable — re-connecting an OAuth provider needs an interactive browser flow. | | connector_web_crawl_start | Create/update a web_crawl connector: crawl a website or sitemap into a RAG vector, optionally recurring. Pass swapOnSuccess: true when the vector backs a published release. | | connector_wordpress_discover | Inspect a WordPress site before connecting it — canonical URL, timezone, and every REST-exposed post type. A post type missing here is not REST-exposed, which is the usual cause of a sync that succeeds and ingests nothing. | | connector_wordpress_start | Create/update a wordpress connector: sync a WordPress install into a RAG vector, optionally recurring. Same swapOnSuccess rule as web_crawl. |

Terms of Service

| Tool | What it does | |------|--------------| | terms_get_for_release | Check whether the calling identity must accept a release's Terms of Service. | | terms_accept | Record acceptance of the current ToS version. | | terms_create / terms_update / terms_publish / terms_list | Author and publish versioned Terms of Service. |

Caution

terms_accept is recorded as a legal act for the authenticated identity. An agent should present the terms to its human user and get explicit agreement before calling it. See Tools → Terms of Service gate.

Memory & voiceprints

| Tool | What it does | |------|--------------| | memory_add / memory_search | Store and retrieve agent memory. | | voiceprint_list / voiceprint_get | List and read speaker voiceprints. | | tts_generate | Generate text-to-speech audio. |

TrustBench (attested evals)

| Tool | What it does | |------|--------------| | trust_run | Run an attested benchmark against a registered assistant. | | trust_verify | Verify a signed TrustRun manifest. | | trust_list_runs / trust_get_run | List and read TrustRun results. | | trust_list_benchmarks / trust_get_benchmark | Browse available benchmarks. |

Red teaming

| Tool | What it does | |------|--------------| | red_team_plan | Show the red-team corpus (categories, probe counts). No probes fire. | | red_team_run | Run one attack category against a deployed release. Check measured before citing the rate; walk categories yourself for a full sweep. |

Platform & catalog (advanced)

| Tool | What it does | |------|--------------| | divinci_api_request | Make a raw, permission-scoped Divinci API request. | | divinci_web_search / divinci_fetch_article | Search the web and fetch article content. | | divinci_create_rag_architecture / divinci_analyze_pipeline / divinci_list_rag_patterns | Design and analyze RAG architectures. | | divinci_*_catalog_node / divinci_generate_node / divinci_promote_node / divinci_test_node_sandbox | Browse and build catalog nodes. | | meta_harness_* | Inspect the self-improvement harness loop, iterations, and proposals. |

Discovery & identity

| Tool | What it does | |------|--------------| | whoami | Show the authenticated identity and its accessible workspaces. | | workspace_search | Search across your workspaces. | | search_tools / list_tool_categories | Search the tool catalog and list its categories at runtime. |

Tip

Don't hard-code this list. Call listTools() (SDK) or read your client's tool menu to get the authoritative, permission-filtered catalog at runtime — it reflects new tools as they ship. See Tools for the programmatic discovery API.