Back to tracker
Plugin tracker
Tools
Explore what tracked Claude Connectors and ChatGPT Plugins can actually do. Search by tool, Plugin, Brand, category, verb, or access requirement.
Latest snapshot2026-09-13USmethodology registry-public-v1
Searchable tools
119,491
Authless tools
7,773
Auth required
107,252
Described tools
64,388
119,491 tools
- Search Homefinder Listings by Visual cluesvisual · Searches the homefinder database for listings given an unstructured user query that contains visual clues. Example visual clues: kitchen island, hardwoord floor, swimming poolComparisComparis
PluginnoneOperations - Get card application linkget · Always use when the user asks where or how to apply, requests an application link, or wants to continue with a Cardwell card. Returns applicationUrl, linkType, officialUrl, disclosure, and freshness date. A referral link is returned when available; otherwise applicationUrl falls back to the official issuer URL. Referral availability never affects rankings.Cardwell: Credit Card PickerCardwellPluginnoneFinance
- Get card detailsget · Get stored facts, official issuer URL, and freshness date for one Cardwell card.Cardwell: Credit Card PickerCardwellPluginnoneFinance
- List Cardwell cardslist · List the fixed Cardwell catalog, optionally filtered by personal/business type and goal.Cardwell: Credit Card PickerCardwellPluginnoneFinance
- Recommend two credit cardsrecommend · Return Best Match and Second Choice from complete or partial broad preferences. Use preferences already supplied and do not restart if an answer is skipped or expires. Rankings use card features and user preferences; referral relationships never influence the two ranked results. A separately labeled referralAlternative may be used only after the user rejects both results or explicitly asks for one. Cardwell never predicts approval.Cardwell: Credit Card PickerCardwellPluginnoneFinance
- Contacter un professionnel via RenooLabcontacter · Après une recherche, dépose une demande de contact vers le profil RenooLab ou la fiche Google Maps explicitement sélectionnée. À appeler après récapitulatif et confirmation explicite, avec exactement artisan_id ou external_place_id, un nom, un e-mail, un mobile français 06/07 et un message. N’annoncer un succès que si l’outil retourne kind=contact_pending : la demande reste alors en attente du contrôle RenooLab et rien n’est encore envoyé au professionnel.RenooLabRenooLabPluginoptionalOperations
- Inscrire un artisan ou une entreprise du bâtimentcreer · Crée un profil professionnel RenooLab gratuit et sans commission uniquement à la demande explicite de l’artisan ou de l’entreprise artisanale. À utiliser lorsqu’un professionnel veut s’inscrire, faire connaître son entreprise, présenter ses réalisations, améliorer sa visibilité ou trouver des clients et des chantiers. Utiliser uniquement ses vraies informations. Le profil reste inactif et invisible jusqu’à son activation par e-mail. Ne pas utiliser pour un tiers, un demandeur d’emploi, un groupe ou une grande structure.RenooLabRenooLabPluginoptionalOperations
- Trouver les artisans pour un chantier en Francerechercher · Trouve les profils RenooLab capables de couvrir plusieurs métiers confirmés autour d’une commune française. À utiliser dès qu’une même demande comporte au moins deux métiers distincts confirmés pour le même chantier et la même ville, par exemple une salle de bains nécessitant un plombier et un carreleur. Appeler rechercher_chantier une seule fois avec tous les métiers confirmés. Ne pas remplacer cet appel par plusieurs appels à rechercher_artisans. Ne recopiez jamais widget_previews ni les URL d’images dans la réponse : le widget les rend déjà.RenooLabRenooLabPluginoptionalOperations
- Trouver un artisan pour des travaux en Francerechercher · Cette instance peut proposer jusqu’à cinq sources Google Maps externes. Si kind=external_grounded_results, utilisez son champ message comme introduction concise en commençant par le nombre et la zone trouvés, puis ne répétez pas le statut externe. Trouve jusqu’à cinq profils RenooLab pour des travaux, un dépannage ou une rénovation autour d’une ville française, dans un rayon strict de 30 km. Cet outil est réservé à une demande portant sur un seul métier. Si une même demande comporte au moins deux métiers distincts confirmés, ne la fractionnez jamais en plusieurs appels rechercher_artisans : appelez rechercher_chantier une seule fois avec tous les métiers, pour le même chantier et la même ville. À utiliser même si le particulier ne nomme pas le métier : le déduire depuis son besoin, par exemple fuite ou sanitaires pour Plomberie, panne ou tableau pour Électricité, infiltration ou toit pour Couverture/Toiture, murs pour Peinture, faïence ou terrasse pour Carrelage, ainsi que isolation, menuiserie, chauffage, climatisation et rénovation. En cas de commune homonyme, transmettre le code postal ou le département demandé par l’outil. Ne recopiez jamais widget_previews ni les URL d’images dans la réponse : le widget les rend déjà. Ne pas utiliser pour un simple conseil DIY, un magasin ou un fournisseur.RenooLabRenooLabPluginoptionalOperations
- channel_get_conversationchannel · Get the messages in one conversation thread (read-only; LinkedIn today). Identify it by conversation_id or profile_url.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_get_emailschannel · Read your most recently received emails from your connected mailbox inbox (Gmail / Outlook / SMTP; read-only). Use this for requests like 'get my last N emails', 'check my inbox', or 'show my recent replies'. Bounces are flagged. account_id is optional when only one mailbox is connected; if the user names a specific mailbox (e.g. 'my gmail'), call linkupapi_list_accounts and pass the matching account_id. Pass unread_only to return only unread messages.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_get_profilechannel · Profile & company lookups on a connected channel (read-only; LinkedIn today). Pick an `action` and pass the matching `params`:
- get_me: no params — returns your own profile.
- get: { profile_url | identifier, profile_urn? } — full public profile.
- get_contact: { profile_url } — contact info (email, phone if visible).
- search_people: { keyword?, location?, current_company?, company_url?, first_name?, last_name?, title?, industry?, school_url?, network?, offset?, limit? }.
- search_companies: { keyword?, location?, sector?, company_size?, offset?, limit? }.
- get_company: { company_url } — full company page data.
- get_candidates: { job_id, total_results? } — applicants for a job (recruiter).
- get_viewers: { count?, offset? } — who viewed your profile.
- get_reactions: { profile_url, count?, cursor? } — reactions on a profile's posts.
- get_comments: { profile_url, count?, cursor? } — comments authored by a profile.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_list_messageschannel · List your inbox on a connected channel (read-only; LinkedIn today). category in INBOX/INMAIL/JOB/UNREAD/ARCHIVED/SPAM.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_list_networkchannel · Connections & invitations lookups on a connected channel (read-only; LinkedIn today). Pick an `action`:
- list_connections: { count?, offset? } — your connections (paginated, max 1000/call).
- list_invitations: { count?, offset?, invitation_type? } — received invitations.
- list_sent: { count?, offset?, invitation_type? } — invitations you sent.
- check_invitation: { profile_url } — connection status with that profile.
- recommendations: { count? } — people the channel recommends you connect with.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_recruiter_readchannel · Recruiter / jobs lookups on a connected channel (read-only, requires recruiter access; LinkedIn today). Pick an `action`:
- get_posts: { job_id?, count?, offset?, fetch_details? } — your posted jobs. Pass job_id to fetch one specific job.
- get_candidates: { job_id, count?, offset? } — applicants for a job.
- get_cv: { application_id } — download an applicant's CV (returned base64).LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_search_contentchannel · Post search & feed on a connected channel (read-only; LinkedIn today). Pick an `action`:
- search: { keyword?, post_type?, sort_by?, post_date?, profile_url?, company_url?, author_company?, author_industry?, author_job_title?, mentions_member?, mentions_organization?, posted_by?, count?, offset? }.
- get_feed: { count?, cursor? } — your home feed.
- get_comments: { post_url, count?, offset? }.
- get_reactions: { post_url, count?, offset? }.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_send_connectionchannel · Send a connection request on a connected channel (message <=300 chars; LinkedIn today). The recipient is notified immediately — cannot be recalled even if later withdrawn.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_send_emailchannel · Send an email from your connected mailbox (Gmail / Outlook / SMTP). Delivered immediately from your mailbox and cannot be unsent. Body: pass `text` (plain text — often the best-delivering format for cold outreach), `html`, or both. account_id is optional when only one mailbox is connected; if the user names a specific mailbox (e.g. 'from my gmail'), call linkupapi_list_accounts and pass the matching account_id.LinkupAPILinkupAPIPluginrequiredSales & CRM
- channel_send_messagechannel · Send a DM on a connected channel (LinkedIn today). Delivered immediately and cannot be unsent. media_file = { data: base64, filename }. media_link = public URL.LinkupAPILinkupAPIPluginrequiredSales & CRM
- linkupapi_enrichlinkupapi · Public-data enrichment & email tools. No account_id required — uses LinkupAPI's managed pipeline. Pick an `action` and provide the matching `params`:
- enrich_person: { first_name, last_name, company_name } — finds the LinkedIn URL and returns the full profile.
- enrich_company: { company_domain | company_url } — company info from a domain or URL.
- find_email: { linkedin_url } OR { first_name, last_name, company_name | company_domain } — professional email finder.
- validate_email: { email } — SMTP, MX & catch-all check.
- reverse_email: { email } — finds the person behind a professional email (personal providers like gmail/yahoo are rejected).
- search_jobs: { keywords, location?, country?, limit?, is_remote?, job_type? } — job board postings (Indeed/LinkedIn/Glassdoor). country drives the Indeed/Glassdoor region; if omitted it is inferred from location (e.g. 'Madrid, Spain' → spain), else defaults to france. job_type ∈ fulltime|parttime|contract|internship.
- extract_employees: { company_url, limit? } — employees of a company (LinkedIn search).
Cost: 1 credit per call.LinkupAPILinkupAPIPluginrequiredSales & CRM
- linkupapi_get_accountlinkupapi · Get full details of one account (status, country, webhook_url, timestamps).LinkupAPILinkupAPIPluginrequiredSales & CRM
- linkupapi_get_creditslinkupapi · Get the current credit balance for the authenticated API key.LinkupAPILinkupAPIPluginrequiredSales & CRM
- linkupapi_get_logslinkupapi · List API call logs for the authenticated user. Filter by action (e.g. 'send', 'invite'), account_id, or status ('success' / 'error').LinkupAPILinkupAPIPluginrequiredSales & CRM
- linkupapi_list_accountslinkupapi · List every channel account connected to your API key — each with its account_id, platform (linkedin/email), mailbox address, provider (gmail/outlook), and status. Call this to pick the right account_id when the user names a specific account or mailbox (e.g. 'my gmail').LinkupAPILinkupAPIPluginrequiredSales & CRM
- FAQfaq · Answer frequently asked questions about Lassie pet insurance. Use this when users ask about coverage, waiting periods, plans, pricing, cancellation, or other insurance-related questions. Pass the country parameter when the question is about a specific country's insurance.
TONE GUIDE — "Care from your pet's point of view":
When answering FAQ questions, embody Lassie's voice: compassionate, reliable, and inspirational.
- Talk like a caring friend — warm, helpful, never stiff or clinical
- Explain insurance concepts simply — avoid jargon, be relatable for someone new to pet insurance
- Be frank and truthful — never hide things, but frame information positively
- Keep answers short and concise — no walls of text
- Express passion for pet health and preventative care
- Do NOT sound sales-y or focus on price/being cheap — highlight Lassie's care-first qualities instead
- Never be judgmental
- COMPLIANCE: Lassie is an insurance broker — NEVER say Lassie is an insurance company. NEVER give recommendations or advice — only provide information. If the user asks what you recommend, firmly but warmly decline and redirect to factual information. NEVER use social proof (e.g. "most people choose…") or comparative framing that steers the user.
- EMOJIS: You may use sparingly: 💙 (compassion), 😊 (warmth), ☑️ (benefits), 💡 (tips), 🐾 (pets). Avoid off-palette hearts, over-expressive, or negative emojis.LassieLassiePluginnoneFinance
- Lassie Quotelassie · Get a pet insurance quote from Lassie. Use when a user wants to insure their dog or cat, get a price quote, or learn about Lassie insurance packages.
COMPLIANCE: Lassie is an insurance broker — NEVER say Lassie is an insurance company. NEVER give recommendations or advice — only provide information. Always make clear that the information provided is not a personal recommendation.
NO-RECOMMENDATION RULE (CRITICAL — applies to the ENTIRE conversation, not just pricing):
- NEVER recommend, suggest, or steer the user toward any specific plan, option, or choice — not even implicitly.
- NEVER use social proof to nudge the user (e.g. "most people choose…", "a lot of pet parents in your situation…", "popular choice for…").
- NEVER use comparative framing that implies one option is better (e.g. "if you want the best protection…", "for peace of mind…").
- NEVER answer "what do you recommend?" or "what should I pick?" with any form of guidance. Instead, warmly decline and redirect: explain that you can only provide factual information about each option, and invite the user to ask specific questions about coverage details so they can decide for themselves.
- When comparing plans, present ONLY factual differences (price, coverage limits, what's included). Let the user draw their own conclusions.
TONE GUIDE — "Care from your pet's point of view":
Lassie is the caring companion, created by and for pet parents. You are a warm, knowledgeable assistant — like a friend who happens to know everything about pet insurance. This should feel like a conversation, NOT a form.
Be COMPASSIONATE: caring, warm, and an active listener. Serve with a helping approach before they even ask for it. Put yourself in the pet parent's shoes and show genuine interest in their pet.
Be RELIABLE: everything you say should be correct but never dry or clinical. You're a trustworthy friend they can always count on.
Be INSPIRATIONAL: be forward-thinking, express passion for pet health and preventative care. Be a guiding hand with confidence — not arrogance.
DO:
- Talk as if you were a friend — be relatable, even for someone who's never dealt with pet insurance
- React to what the user shares with genuine interest (e.g., "Goldens are such great dogs! 🐶" or "Two years old — still full of energy!")
- Use the pet's name once you know it
- Explain difficult concepts in an easy way — avoid insurance jargon or overcomplicated terms
- Speak the truth and be frank — never hide things
- Keep text short and concise — use animations/widgets instead of walls of text
- Match the user's language (Swedish, French, German, or English) throughout
- Enhance Lassie's qualities: preventative care, the app, tools for improving pet health
DON'T:
- Sound sales-y or focus on price/being cheap — enhance Lassie's qualities instead
- Be judgmental — ever
- Be too playful — friendly and caring, but still reliable and trustworthy
- Use walls of text — keep it brief and warm
- Be stiff or robotic
- Act as if Lassie is the budget option
EMOJIS: You may use these sparingly: 🐶🦮 (dogs), 🐱🐈 (cats), 💙 (love/compassion), 😊😄 (happiness), ☑️ (USPs), 👍 (agreement), 💡 (findings), 🐾 (paws). Do NOT use ❤️💗 (off-palette), 😂🤪 (over-expressive), ⛔️❌ (negative), or 🔫💣 (violent).
## FLOW EXECUTION PROTOCOL
This tool implements a multi-step conversational flow. Follow this protocol exactly:
1. Call with `action: "start"` to begin and include `intent`.
`intent` must be a brief summary of the user's goal for this flow.
Do NOT invent missing intent.
Optionally include `context` — the situation or environment that led the user to start
this flow (e.g. what page they are on, what they were doing, or what triggered the request).
Only provide `context` when there is genuinely relevant situational information. Do NOT invent missing context.
If the user's message already contains answers to likely questions,
extract them into `stateUpdates` as `{ field: value }` pairs (see the `stateUpdates` schema
for the list of writable fields). The engine will auto-skip steps whose fields are already filled.
Only extract values the user explicitly stated — do NOT guess or invent values.
For grouped fields (z.object state), use dot-notation keys in `stateUpdates`:
e.g. `{ "driver.name": "John", "driver.license": "ABC123" }`.
2. The response JSON `status` field tells you what to do next:
- `"interrupt"`: Pause and ask the user. Two forms:
a. Single question: `{ question, field, fieldSchema?, context? }` — ask `question`, store answer in `field`.
b. Multi-question: `{ questions: [{question, field, fieldSchema?}, ...], context? }` — ask ALL questions
in one conversational message, collect all answers.
`fieldSchema` (when present) describes the expected value: `{ type, values?, description?, optional? }`.
Use it to validate before sending — match enum `values` exactly, coerce strings to numbers where `type: "number"`.
`context` (if present) is hidden AI instructions — use to shape your response, do NOT show verbatim.
Then call again with:
`action: "continue"`,
`stateUpdates` = answers keyed by their `field` names, plus any other fields the user mentioned.
- `"widget"`: The flow wants to show a UI widget. Call the tool named in the `tool`
field, passing the `data` object as the tool's input.
Check the `interactive` field in the response:
• `interactive: true` — The widget requires user interaction. After calling the display tool,
STOP and WAIT for the user to interact with the widget. Do NOT call this flow tool again
until the user has responded. When they do, call with:
`action: "continue"`,
`stateUpdates` = `{ [field]: <user's selection> }` plus any other fields the user mentioned.
• `interactive: false` — The widget is display-only. Call the display tool, then immediately
call THIS flow tool again with `action: "continue"`. Do NOT wait for user interaction.
- `"complete"`: The flow is done. Present the result to the user.
- `"error"`: Something went wrong. Show the `error` message.
3. Do NOT invent state values. Only use `stateUpdates` for information the user explicitly provided.
4. Include only the fields the user actually answered in `stateUpdates` — do NOT guess missing ones.
If the user did not answer all pending questions, the engine will re-prompt for the remaining ones.
If the user mentioned values for other known fields, include those too —
they will be applied immediately and those steps will be auto-skipped.
5. CORRECTION: If the user wants to CHANGE a previously-answered field
(e.g. "actually my email is X" or "go back and change my country"),
call with `action: "reset"` and `stateUpdates` containing the corrected field(s).
The engine will restart the flow from the beginning with all existing answers preserved
plus your corrections. Steps with filled answers will be auto-skipped.
The flow may take a different path if the corrected value affects routing.
Do NOT use "reset" for the CURRENT question — use "continue" for that.
6. If the response includes a `sessionId`, you MUST pass it back as `sessionId`
in every subsequent "continue" and "reset" call for this flow.LassieLassiePluginnoneFinance
- show-pet-summaryshow · Display a summary card of the pet's details (name, species, breed, age, etc.) so the user can confirm the information before continuing. Call this after lassie-quote has collected the pet data, passing the returned pet fields as arguments. Use it whenever the pet's information needs to be reviewed or re-confirmed by the user.LassieLassiePluginnoneFinance
- show-pricingshow · Display the available Lassie pet insurance plans with their prices and coverage details so the user can compare options. Call this after lassie-quote has produced a pricing quote, passing the returned pricing fields (country, insurance type, plans, etc.) as arguments. Use it whenever the user needs to see or revisit the available plans for their pet.LassieLassiePluginnoneFinance
- Browse reviewed image and video modelslist · Browse the fixed, reviewed ChatGPT v1 catalog of text-to-image and text-to-video models. Use it before generating when the user names a model or wants to compare supported sizes, durations or displayed credit estimates. The final request-specific charge is calculated server-side. This lookup creates no job and consumes no credits.FlixlyFlixlyPluginrequiredContent & Design
- Check a generationget · Fetch the status and output URL of an allowed image or video job created with this same connection credential. The mounted Flixly result widget uses this read-only helper to update one card in place. It creates no job and consumes no generation credits, so the same id is safe to check.FlixlyFlixlyPluginrequiredContent & Design
- Check available creditsget · Show only the connected Flixly account's currently available generation-credit balance. The response contains no plan, email, API key, token, payment details, prompts, media or internal account identifiers. It creates no job and consumes no credits.FlixlyFlixlyPluginrequiredContent & Design
- Generate a videogenerate · Create one paid short video from a text prompt with a model from the fixed ChatGPT v1 catalog. The input is text only; reference and source media are not accepted. It consumes server-calculated Flixly credits. The attached result widget automatically monitors the original task id and displays the completed video without resubmitting or charging again.FlixlyFlixlyPluginrequiredContent & Design
- Generate an imagegenerate · Create one paid image from a text prompt with a model from the fixed ChatGPT v1 catalog. The input is text only; reference and source media are not accepted. It consumes server-calculated Flixly credits. The attached result widget automatically monitors the original task id and displays the completed image without a second paid submission.FlixlyFlixlyPluginrequiredContent & Design
- ack instructionsack · Mark instructions as claimed/done so future polls don't re-surface them. Pass either a list of ids, or all=true to ack every queued instruction in the workspace.ElodusElodusPluginrequired
- add data model fieldadd · Add a field to a data-model entity. `fk_to_entity_id` wires this field as a foreign key to another entity in the same workspace — drives the connection arrows on the ER diagram. `composition=true` marks the FK as ownership-style (rendered as a filled diamond in class diagrams).ElodusElodusPluginrequired
- add phase dependencyadd · Wire a 'this phase depends on that phase' edge into the workspace's phase board. Drives the dependency arrows on /phases. Idempotent — re-adding an existing edge is a no-op.ElodusElodusPluginrequired
- add to collectionadd · File one result into a collection. Take source+ref straight from the search result you are filing (source='nih_reporter', ref=core project number; source='crossref', ref=DOI). Do NOT curate a collection the researcher made without being asked — the set exists because a person chose it.ElodusElodusPluginrequired
- append sectionappend · Append a new section to a document (write). For a PRESENTATION document a section IS a slide: prefer a reference-token body ('@present:document:42', '@present:section:99', '@present:widget:<slug>', '@present:bpmn:<slug>') to show an existing object live. If the content you want on the slide does NOT exist yet, FIRST create a document for it (create_document) and link THAT — don't write a free-text slide; text slides go stale while linked ones stay live. Use plain markdown only for title/intro/divider slides.ElodusElodusPluginrequired
- build business processbuild · Build a business process diagram from a structured spec — the easy way for an agent to author one. You give nodes + flows (+ optional swimlanes); the server generates valid BPMN 2.0 XML AND lays it out (positions, edge routing) so you never compute coordinates. Renders in the modeler and is fully editable by humans. Pass `process_id` to rebuild an existing diagram. Prefer this over create_business_process (which needs raw XML). (write)ElodusElodusPluginrequired
- create business processcreate · Create a business process diagram from raw XML (the OMG BPMN 2.0 schema the modeler reads/writes, including diagram-interchange coordinates). Most agents should use build_business_process instead — it takes a node/flow spec and lays it out for you. Use this only when you already have full BPMN XML. (write)ElodusElodusPluginrequired
- create collectioncreate · Start a new collection. Use when the researcher wants to keep a set of results together — 'the papers I will cite', 'calls worth applying to this cycle'.ElodusElodusPluginrequired
- create data model entitycreate · Create a table in a workspace's database (the Database designer). Pass `database_slug` (or `database_id`) to put it in a specific database; omit both to use the workspace's default database. Use `create_database` to make a new one. (`kind` is accepted for back-compat but ignored — everything is a table now.)ElodusElodusPluginrequired
- create databasecreate · Create a database (a container for tables) in a workspace, for the Database designer. Tables are added with `create_data_model_entity` targeting this database's slug.ElodusElodusPluginrequired
- create documentcreate · Create a new document in a workspace (write). The 'kind' MUST be one of THIS workspace's document_kinds — call list_object_types first and use a kind from its `document_kinds` list; do not assume generic kinds like 'blueprint'. To build a SLIDE DECK / presentation, pass kind='presentation' (a universal utility kind always accepted, even if it's not in the ontology): the document's sections become the slides — add each with append_section. PREFER REFERENCE SLIDES: to put existing workspace content on a slide, set the section body to a SINGLE reference token — '@present:document:42', '@present:section:99', '@present:widget:<slug>', '@present:bpmn:<slug>' — so the slide shows the LIVE object and stays in sync. If the slide's content does NOT exist yet, FIRST create a document for it (create_document) and link THAT — don't write a free-text slide. Do NOT retype an existing document/section's content into a text slide (it duplicates and goes stale). Use a plain markdown body only for title / intro / divider slides that have no underlying object. SLIDE SIZING: each slide is one full 16:9 screen shown from a distance, NOT a scrolling page — the presenter does not shrink to fit. Budget each slide to about a short heading plus ~6 bullet lines (~40 words), ~10 lines / ~90 words absolute max; split dense content across MORE slides rather than overflowing one. Prefer short bullets over paragraphs; one idea per slide.ElodusElodusPluginrequired
- create objectcreate · Create an instance of an ontology object type. `attributes` are validated against the type's schema. For document types pass optional doc_kind; for record types optional entity_kind.ElodusElodusPluginrequired
- create phasecreate · Create a new SDLC phase in a workspace.ElodusElodusPluginrequired
- create saved searchcreate · Save a named search so the team (and you) can re-run it. Stores CRITERIA, never results. Draft one when the researcher lands on a query worth keeping, then tell them you saved it. Defaults to 'available' — do NOT pin unless the researcher explicitly asks, because pinning makes it standing context in every future conversation. Criteria must match the source's schema (see the source's own search tool for the allowed keys).ElodusElodusPluginrequired
- create widgetcreate · Create a Widget — a small self-contained interactive frontend mini-app (e.g. a mortgage calculator) to illustrate an interface/interaction. Seed any data with widget_store_set.ElodusElodusPluginrequired
- create workordercreate · Queue a new workorder in a workspace (write, requires per-user key).ElodusElodusPluginrequired
- crossref get bibtexcrossref · Resolve a DOI to a ready-to-paste BibTeX entry (or a formatted citation). Feed the content straight into an Overleaf .bib via overleaf_write_file.ElodusElodusPluginrequired
What is Tool Explorer?
Tool Explorer indexes the callable tool names and descriptions attached to public registry profiles. It is useful for seeing what agents can actually invoke, not just which profile exists.
How do category and verb filters work?
Category filters use the live registry category rollup. Verb filters use the public tool insights rollup, so the page stays backed by the same read models as the tracker charts.
Why do auth requirements matter?
Auth requirements show whether a tool is likely usable without account connection, requires authentication, is private, or is unknown in the current snapshot.