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 snapshot2026-09-12USmethodology registry-public-v1
Searchable tools
113,018
Authless tools
7,424
Auth required
100,766
Described tools
61,167
113,018 tools
- Get Review Batch / QC Statusget · Review batch and QC status snapshot for the connected eDiscovery workspace. Returns the number
of review phases, the total and reviewed document counts, the total review batches, and the
batch pipeline by state — pending (not started), active (in progress / accepted), and completed.
Use this when the user asks about review/QC progress, batch or QC status, how many phases or
batches exist, or how far along review is. Takes no parameters — it reports on the connected
workspace.Casepoint GovCasepoint
PluginrequiredOperations - Get Review Team Productivityget · Review team productivity report for the connected eDiscovery workspace, grouped by reviewer.
Returns one row per reviewer with documents reviewed, image counts and edits summed across
the date range, total review time (HH:MM:SS), and the hourly review rate (total documents
reviewed per review-hour).
Use this when the user asks things like: "how productive is the review team", "reviewer
productivity", "review rate / docs per hour", "who reviewed the most documents", "how much
time did reviewers spend". All parameters are optional — defaults give the full, unfiltered
report for every reviewer.Casepoint GovCasepointPluginrequiredOperations
- Get Workspace Summaryget · Orientation snapshot of the connected eDiscovery workspace (matter). Returns these
workspace-level fields: workspaceName, workspaceStatus (Active / Closed / Archived),
matterType, openDate, and totalCustodians.
Document counts cover EVERY datastore the caller can access (not just the default one) and are
returned in two parts:
- 'aggregate' — the default overview: accessibleDataStores (how many datastores the caller
can access) plus totalDocuments and producedDocuments summed across all of them.
- 'dataStores' — an array with one entry per datastore, each carrying dataStoreId,
dataStoreName, totalDocuments and producedDocuments.
Workspace name/status/matter type and custodian count are workspace-level (not per datastore).
By DEFAULT present the 'aggregate' overview. Only break the numbers down per datastore (from
'dataStores') when the user explicitly asks for individual / per-datastore data — e.g.
"documents per datastore", "breakdown by datastore", "counts for each datastore".
Use this when the user asks things like: "what is this matter / workspace", "give me an
overview", "matter type / status / open date", "how many documents are in here", "how many
custodians", "documents per datastore". Takes no parameters — it reports on the connected workspace.Casepoint GovCasepointPluginrequiredOperations
- Search Requestssearch · OUTPUT FORMAT — READ FIRST
══════════════════════════
Present results as a concise Markdown table/summary; never paste raw JSON to the user.
Use fully spelled-out, human-readable column headers — never a bare symbol. Every DATE
field's header MUST include the word "Date". Use exactly these headers per field:
tracking_number → "Tracking Number" (NEVER "Tracking #" or "#")
request_type → "Request Type"
requester_name → "Requester"
received_date → "Received Date"
perfected_date → "Perfected Date"
adjusted_due_date → "Adjusted Due Date"
state → "State"
status → "Status"
processing_track → "Processing Track"
component → "Component"
days_remaining → "Days Remaining"
closed_by → "Closed By"
closed_date → "Close Date"
close_remark → "Close Remarks"
deliver_mode → "Delivery Mode"
estimated_completion_date → "Estimated Completion Date"
RENDER EVERY FIELD RETURNED — DO NOT CHERRY-PICK COLUMNS
════════════════════════════════════════════════════════
The row shape is fixed — every row the tool returns contains ALL 16 fields listed
above, in that order. Your default rendering MUST include EVERY one of those fields
as its own Markdown-table column, using the headers exactly as mapped. Do NOT drop
columns just because the values look uninteresting for a specific query (e.g. don't
omit "Request Type", "Perfected Date", "Adjusted Due Date", "State", "Status",
"Processing Track", "Days Remaining", "Close Remarks", or "Estimated Completion Date"
merely because the user's prompt was about closures).
Rules:
• The column count in the header row equals 16 (one per field above). Same count
on the delimiter row. Same count on every data row.
• Render null / empty values as "—" (em dash). Never collapse a column just
because every row has a null value.
• Do NOT reorder columns. The order above is the required order.
• The ONLY exception is when the user's prompt explicitly narrows the columns
("show only tracking number and close date", "just requester and status") — in
that case render exactly the requested subset in the requested order.
• If the table is wide, that's fine — do not truncate, wrap by omitting columns,
or promote fields into prose narration.
PURPOSE
═══════
Search FOIA Requests using natural-language-derived filter clauses.
READ-ONLY. Scoped to the caller's connected organization + role visibility.
HOW FIELDS ARE ADDRESSED
════════════════════════
Every field is referenced by its human DISPLAY LABEL (e.g. "Due Date", "Status",
"Component"). There are no internal identifiers in this contract — `fieldname` is
always a display label, both in your input and in any warning echoed back. Use the
standard labels in the trigger table below. Never invent or guess a label, and never
ask for or expect a field list — the available fields are exactly the standard labels.
WHEN TO USE
═══════════
Prompt asks for a LIST of requests filtered, sorted, or counted.
WHEN NOT TO USE
═══════════════
- Single-request detail (correspondence log, all custom fields) — out of scope.
- Aggregate metrics (FY report, fee-waiver totals, avg processing days) — out of scope.
- Mutations (create / update / close a request) — this tool is read-only.
- Cross-organization queries — scoped to the caller's connected organization only.
TRIGGER TABLE — apply ALL matching rows (use these labels exactly as written)
═════════════════════════════════════════════════════════════════════════════
REQUEST-TYPE FILTERING (add a "Request Type" clause in cases 1–2; ALWAYS combine it
with any OTHER filters the prompt also gives):
1. The prompt names a request TYPE as a descriptor of the requests — "FOIA requests",
"appeals", "consultation requests", "Privacy Act requests". This holds even when
other filters are present, and you then send BOTH clauses:
• "in-progress FOIA requests" → "State" Equal "In Progress" AND "Request Type" Equal "FOIA"
• "overdue Appeal requests" → "Due Date" LessThan <today> AND "Request Type" Equal "Appeal"
2. Bare "give me all requests" / "list requests" with NO other filter → "Request Type"
Equal "FOIA" (a clause is required; "all requests" defaults to the FOIA type).
Do NOT add a "Request Type" clause when:
- "FOIA" only names the APPLICATION / system — "requests in the FOIA application",
"in-progress requests in FOIA" — that is the app, not the type; send only the real
filters so all types return.
- the prompt gives a filter (State, Status, Date, Requester, etc.) and names NO type at all —
"in-progress requests", "overdue requests" — send only those clauses.
Litigation is excluded server-side and is never available, so never filter by it.
| Prompt trigger | Clause (field label · operator · value) |
|-----------------------------------------|------------------------------------------|
| A type describes the requests: "FOIA | "Request Type" Equal "<TypeName>" |
| requests" / "appeals" / "Privacy Act | (ALSO add the other clauses, e.g. |
| requests" (even with other filters) | State, when both are named) |
| Bare "all requests" with NO other filter| "Request Type" Equal "FOIA" |
| Any other filter given, no type named | (no Request Type clause — all types) |
| "open" / "still open" / "not closed" | "State" NotEqual "Closed" |
| WITHOUT "status" (request is in any | ("open" ≡ NOT Closed — i.e. any of |
| state other than Closed) | In Progress / On Hold / Perfected) |
| "in progress" WITHOUT "status" | "State" Equal "In Progress" |
| (or "... state") | |
| "closed" / "perfected" / "on hold" | "State" Equal "<word>" |
| WITHOUT "status" (or "... state") | |
| the user says the word "status": | "Status" Equal "<value>" |
| "<value> status" (interim sub-status) | |
| "overdue" / "past due" | "Due Date" LessThan <today ISO> |
| "due today" | "Due Date" Equal <today ISO> |
| "due in N days" | "Due Date" GreaterThanOrEqual today |
| | AND LessThanOrEqual today+N |
| "filed by X" / "requester X" | "Requester" Equal "<X name OR email>" |
| "assigned to X" / "owned by X" | "Assigned To" Equal "<X name OR email>" |
| "closed by X" | "Closed By" Equal "<X name OR email>" |
| "closed on X" / "close date X" | "Close Date" Equal <ISO date> |
| "closed after X" / "closed since X" | "Close Date" GreaterThanOrEqual <ISO> |
| "closed before X" | "Close Date" LessThan <ISO> |
| "close remark(s) contain X" / | "Close Remarks" Contains "<X>" |
| "closing note mentions X" | (label is PLURAL: "Close Remarks") |
| "delivered by mail" / "delivery via X" | "Delivery Mode" Equal "<X>" |
| "estimated to complete by X" / | "Estimated Completion Date" |
| "target completion date X" | LessThanOrEqual <ISO> |
| "in <C> component" / "<C> component" | "Component" Equal "<C>" |
The labels above ("Request Type", "State", "Status", "Due Date", "Requester",
"Assigned To", "Closed By", "Close Date", "Close Remarks", "Delivery Mode",
"Estimated Completion Date", "Component") are the standard FOIA labels — use them
EXACTLY as written; they must match the organization's search-profile alias (e.g.
"Close Remarks" is PLURAL). If a clause comes back dropped with reason
"unknown fieldname", the label you sent didn't match a searchable field — either the
search profile doesn't project it, OR the label is spelled differently than the
catalog's (check singular/plural and exact wording). If that droppedClauses entry
includes a "didYouMean" value, silently re-issue the SAME search ONCE with that exact
label instead; otherwise tell the user the filter couldn't be applied.
STATE vs STATUS — two distinct fields; route by the WORD the user uses:
- NO "status" word (or the user says "... state") → "State" (the request status). Values:
"In Progress", "Closed", "On Hold", "Perfected". "open" / "still open" means NOT Closed
→ "State" NotEqual "Closed" (it is NOT a synonym for "In Progress").
• "open requests" / "still open" → "State" NotEqual "Closed"
• "requests that are in progress" → "State" Equal "In Progress"
• "requests in progress state" → "State" Equal "In Progress"
- The user says the word "status" → "Status" (the finer interim / sub-status). Pass the
plain value; the server matches the interim status by exact name and searches EVERY
interim id sharing that name (OR'd) — do NOT guess or expand ids yourself.
• "requests in progress status" → "Status" Equal "In Progress"
• "FOIA requests in progress status"→ "Status" Equal "In Progress" AND "Request Type" Equal "FOIA"
- If the user says "<X> status" but no interim status named X exists, the search returns a
no-match warning; it does NOT fall back to "State".
In responses these map to the row fields `state` (request status) and `status` (interim).
DO NOT
══════
- Add a "Request Type" clause when "FOIA" only names the FOIA application, or when the
user gave another filter (Status, Date, etc.) but named no type — omit it so all types
return. (Exception: a bare "all requests" with NO other filter → "Request Type" Equal "FOIA".)
- Invent or guess field labels — use only the standard labels in the trigger table.
- Use Between/In/NotIn — not supported in v1; use OR-chained Equal clauses.
- Sort by anything other than a supported sort key (see sortBy below) — unknown keys
silently fall back to "tracking_number" with warnings.adjustedSort set.
- Loop the same payload after an error envelope — recompose, don't retry as-is.
INPUT (stringified JSON)
════════════════════════
{
"where": [{ "fieldname":"<display label>","operator":"<op>","value":"<display value>","operand":"And"|"Or" }, ...],
"sortBy":"<sort key>","sortOrder":"asc"|"desc",
"currentPage": <int >= 1>
}
Operators: Equal, NotEqual, Contains, StartsWith, EndsWith,
LessThan, LessThanOrEqual, GreaterThan, GreaterThanOrEqual, IsNull, IsNotNull.
sortBy: one of tracking_number, received_date, perfected_date,
adjusted_due_date, closed_date, estimated_completion_date
(default: tracking_number). Anything else falls back to tracking_number.
closed_date is only meaningful when the population is restricted to closed
requests (State Equal "Closed"); rows without a close date sort last.
Limits: max 50 clauses per payload; max 256 chars per value.
Values: for a choice/lookup field, pass the option's display name as value. For people
fields (Requester / Assigned To / Closed By), a name or email also works — the
resolver matches "<name> (<email>)" rows by name alone, email alone, or full string.
Dates: ISO-8601 (yyyy-MM-dd). "today" resolves to the current server-UTC date.
Pagination: `currentPage` is optional (default 1). Page size is fixed by server
configuration and is NOT a caller parameter. "more" / "next page" →
re-call with currentPage + 1.
Empty payload ("" / "{}" / {"where":[]}) is rejected with NO_CLAUSES — there is no
field-listing mode; always send at least one clause.
OUTPUT
══════
{
items: [<row>, ...],
warnings: {
droppedClauses: [{ fieldname, operator, reason }],
unresolvedClauses: [{ fieldname, value, reason }],
ambiguousResolution: [{ fieldname, value, matchedCount, pickedValue }],
adjustedSort: <sort key | null>
},
pagination: { totalCount, currentPage, recordPerPage, totalPages, hasMore, outOfRange },
message: <string | null>
}
row = { tracking_number, request_type, requester_name, received_date, perfected_date,
adjusted_due_date, state, status, processing_track, component, days_remaining,
closed_by, closed_date, close_remark, deliver_mode, estimated_completion_date }
state = the request status (e.g. "In Progress", "Closed", "On Hold", "Perfected").
status = the interim status — a finer-grained sub-status within the current state; may be
null when no interim status is set.
closed_by = user (name / email) who closed the request; null while
the request is still open.
closed_date = ISO date the request moved to Closed state; null when
the request is still open.
close_remark = free-text closing remark captured at close time; null
when the closer left it blank.
deliver_mode = delivery-method label used to send the response to the
requester (e.g. "Email", "Portal", "Mail"). Null when
no delivery has been logged yet.
estimated_completion_date = ISO date the assigned officer projected as the
completion target for open work; null when no estimate
is set. This is a planning field — do NOT conflate with
adjusted_due_date (the statutory / extended deadline).
When listing/grouping/counting BY TYPE, group on request_type — never on the
tracking-number prefix (prefixes are not a type signal and misclassify rows).
(`fieldname` in warnings is the same display label you sent.)
Notes:
- warnings is present ONLY when at least one warning exists; when there are none the
whole warnings object is omitted from the response (do not expect an empty warnings block).
- pagination.recordPerPage in the OUTPUT reflects the server-configured page size.
- pagination.totalCount is rows on THIS page (not DB total); use hasMore as the
"is there more?" signal.
WARNING HANDLING (informational, NOT retry signals)
════════════════════════════════════════════════════
- SUCCESS-SILENT: warnings appear ONLY when something went wrong. When the warnings
object is absent, say NOTHING about filters or validation — do NOT announce that
nothing was dropped/unresolved. Never surface the words "clause", "dropped", or
"unresolved" to the user; when a warning does occur, describe it in plain terms.
- warnings.droppedClauses non-empty → MUST tell the user which filter was dropped
and why (e.g. "I couldn't filter by X because ..."). Don't silently hide it.
- warnings.unresolvedClauses non-empty → the user's value didn't match any organization
lookup row. Ask them to clarify the spelling.
- warnings.ambiguousResolution non-empty → multiple lookup rows share the same name;
the first was picked. Surface the ambiguity to the user.
- warnings.adjustedSort set → an unknown sortBy was coerced to the default; surface it.
ERROR ENVELOPES (each terminal — recompose, don't loop)
═══════════════════════════════════════════════════════
- INVALID_JSON → repair the payload's JSON syntax.
- NO_CLAUSES → empty where[]. Add ≥1 clause. For a bare "all requests" with
no other filter, use "Request Type" Equal "FOIA". If the user
gave a real filter (Status, Date, etc.), send that instead and
add no Request Type clause.
- PAYLOAD_TOO_LARGE → where.length > 50. Consolidate (Contains instead of many Equals).
- RESERVED_FIELD → "values":[] is reserved for v1.1; use OR-chained Equal in v1.
- INVALID_PAGINATION → currentPage < 1.
- unauthorized_context → the session has no valid tenant context. Tell user to reconnect.
- all_clauses_dropped → every clause failed validation. Re-read warnings.droppedClauses;
if reason is "unknown fieldname", that field isn't filterable in
this organization — tell the user. Do NOT keep guessing labels.
- all_clauses_unresolved → every choice-field clause missed lookup; ask user to clarify.
- casepoint_api_timeout → backend slow. Narrow the search (more selective clauses).
PRE-SUBMIT CHECKLIST — re-read your composed `where` BEFORE returning
═════════════════════════════════════════════════════════════════════
1. Request Type clause? Add "Request Type" Equal "<TypeName>" if (a) a type describes the
requests ("FOIA requests", "appeals", "Privacy Act requests") — KEEP it even when the
prompt also names a State/Date/etc.; add both clauses (e.g. "in-progress FOIA
requests" → State AND Request Type), or (b) it is a bare "all requests" with NO other
filter → Equal "FOIA". Do NOT add it if the prompt gave a filter and named no type, or
"FOIA" only names the application — then all types return.
2. Did the prompt mention status? Route by the word:
- WITHOUT the word "status" (or said as "... state") → a "State" clause (NOT "Status").
"open" / "still open" → "State" NotEqual "Closed" (NOT the literal "Open", and NOT
"In Progress"); "in progress" → "State" Equal "In Progress".
- WITH the word "status" ("<value> status") → a "Status" clause (interim). Send the
single value as-is; the server OR-expands it across all matching interim ids.
3. Did the prompt name a person ("filed by X" / "assigned to X" / "closed by X")?
→ correct person-field clause present? Use label "Requester" / "Assigned To" /
"Closed By" respectively.
If any of 1–3 are in the prompt but missing from `where`, your payload is wrong.Casepoint GovCasepointPluginrequiredOperations
- Create dropcreate · Save an item (text, note, link, code, quote, image reference) to the signed-in user's private CherryDrop inbox. Use only when the user explicitly asks to save or drop something into CherryDrop.CherryDropCherryDropPluginrequiredAI
- List contextslist · List the named contexts the user saved in CherryDrop, newest first, including which agent wrote last. Use only when the user explicitly asks which CherryDrop contexts exist.CherryDropCherryDropPluginrequiredAI
- List recent dropslist · List the most recent drops the user saved in CherryDrop. Use only when the user explicitly asks what is in their CherryDrop inbox or what they saved recently.CherryDropCherryDropPluginrequiredAI
- Load contextload · Load a named context the user saved in CherryDrop — the stored conversation state plus any drops attached to it — so this agent can continue where another agent left off. Use only when the user explicitly asks to load or continue a CherryDrop context.CherryDropCherryDropPluginrequiredAI
- Save contextsave · Save or update a named context (a summary of the current conversation state) in the signed-in user's CherryDrop account, so another AI agent can load it later. Use only when the user explicitly asks to save the context or hand it over to another agent.CherryDropCherryDropPluginrequiredAI
- Search dropssearch · Search the drops the user saved in CherryDrop by text content, saved page metadata, or URL. Use only when the user explicitly asks to find something in their CherryDrop inbox.CherryDropCherryDropPluginrequiredAI
- Find churches in a cityfind · Find Christian worship churches in a named city, ranked by directory quality. Filter by worship style, denomination, or language.GospelChannel Church GuideGospelChannelPluginnoneConsumer & Lifestyle
- Find churches near a placefind · Find Christian worship churches near the user, ranked by distance. Uses only the coarse location metadata ChatGPT shares, never raw location input fields. If the metadata includes only a city, it searches that city. Filter by worship style, denomination, or language.GospelChannel Church GuideGospelChannelPluginnoneConsumer & Lifestyle
- Get a church profileget · Get the full profile for one church by its slug (the last path segment of its gospelchannel.com/church/<slug> URL): worship styles, denomination, language, summary, top worship songs, contact, and recorded service times if available.GospelChannel Church GuideGospelChannelPluginnoneConsumer & Lifestyle
- Daily health summaryget · Retrieve all available personal health metrics and the self-reported Daily Check-in
for one recent local calendar day, including available seven-day personal baselines
for measured metrics. Prefer this tool for questions
such as "how was yesterday?" or "how did I sleep last night?". Use
get_health_data instead for a multi-day trend.
date must be within the most recent 14 local calendar days, including today.
A metric group or an individual optional field is absent when the corresponding
reading is unavailable; absence never means zero.
baseline_observed_days states how many of the seven preceding local dates
contributed to that metric's personal baseline. Heart-rate, activity, and sleep
stage values are rounded to whole practical units. Sleep hours retain two decimal
places and include display labels such as "7h 42m" for natural-language replies.
sleep_efficiency_percent is the percentage of the recorded sleep session
classified as asleep; its seven-day baseline follows the same definition.
Sleep is associated with the local date on which the user woke up; its start
time may be on the preceding calendar date.
A daily_check_in contains four subjective 1-5 ratings: physical_condition, energy,
mood, and calmness. The endpoints are very poor to great, exhausted to high, very low
to great, and very tense to very calm, respectively; 3 is the middle response. It is
not sensor-measured, does not have a combined score in this response, and is not a
medical assessment.
Wearable measurements are not a medical diagnosis.Stress Platform: Health CoachStress PlatformPluginrequiredHealth & Wellness
- Platform setup statusget · Check whether the user is set up to access their personal Stress Platform
health metrics.
Call this tool before the first personal-health request in a conversation, or
whenever the user needs help with account setup, data connection, billing, or access.
Returns whether the user has a Stress Platform profile, the sign-in method
associated with their account (`google`, `apple`, or `email`), subscription
plan and status, subscription source and expiry date, and whether a
health-data source is connected.
Important: the user must use the same sign-in method they used when joining
Stress Platform and connecting their health-data source. Using a different
sign-in method, even with the same email address, may open a different account.
Personal health metrics require, in order: a Stress Platform profile,
effective Premium access, and a connected health-data source. Follow `requirements`, which
contains only the user's next setup action; do not skip ahead or infer
access from the provider billing status. `can_read_metrics` is the final
access decision and accounts for supported trials and billing grace periods.
If no profile exists, guide the user to join Stress Platform at
stress-platform.com using the sign-in method shown by this tool. If Premium
access is required, explain neutrally that personal metrics are unavailable under
the current account entitlement. Do not initiate checkout, facilitate a purchase,
or promote an upgrade. If no health-data source is
connected after the account and Premium steps are complete, guide them to
download Stress Platform iOS app from the App Store. They can complete a subjective
Daily Check-in, and may also connect Apple Health to start syncing Apple Watch data.
A past-due or cancelled billing status may explain limited access even during a
grace period; report the status without initiating renewal or payment actions.Stress Platform: Health CoachStress PlatformPluginrequiredHealth & Wellness
- Recent health dataget · Retrieve the authenticated user's recent daily health measurements, activity,
and self-reported Daily Check-ins.
Call this tool after get_platform_status confirms that personal health data is
available, when a question about the user's own sleep, recovery, fatigue,
heart rate, activity, exercise, reported state, stress patterns, or recent trends
would benefit from personal data. Do not use it for general health education that
does not require the user's data.
Each requested metric group contains available daily values and coverage. Measured
metric groups also contain summary statistics.
Heart-rate variability is RMSSD in milliseconds; resting heart rate is in beats
per minute. Sleep durations are in hours and sleep stages are in minutes.
sleep_efficiency_percent is the percentage of the recorded sleep session
classified as asleep: it uses time in bed when the source provides it,
otherwise recorded asleep and awake stages. Activity fields identify their own units.
DAILY_CHECK_IN is subjective reported state rather than a sensor measurement. Its
physical_condition, energy, mood, and calmness ratings are required integers on a
1-5 scale, where 1 is the most negative response, 3 is the middle response, and 5 is
the most positive response. Specifically: physical_condition ranges from very poor to
great, energy from exhausted to high, mood from very low to great, and calmness from
very tense to very calm. Keep the four dimensions separate; do not invent an aggregate
score or treat them as physiological recovery.
A metric group's days list contains only local calendar dates with available
data, ordered oldest to newest. coverage.observed_days shows how many of the
coverage.requested_days have a usable value. Missing dates have an unknown
cause (for example no check-in, device non-wear, unavailable permissions,
delayed syncing, or another source issue) and are never represented as zero.
A requested metric group is absent when it has no usable values in the
selected period. Statistics are calculated only from the returned available days.
Values are rounded to whole practical units for conversation, except sleep hours,
which retain two decimal places.
The current day may be partial. For sleep, date is the local wake-up/reported date, so
sleep_start_time may be on the preceding calendar date. Sleep
statistics can include typical bedtime and wake time. Wearable measurements
are not a medical diagnosis.Stress Platform: Health CoachStress PlatformPluginrequiredHealth & Wellness
- Recovery statusget · Retrieve the user's physiological recovery score and its HRV, resting-heart-rate,
and sleep components compared with their personal 14-day baseline. Use when the user asks
about recovery, fatigue, or training readiness.
components_used identifies the inputs that contributed to the score and
components_missing identifies the inputs that could not. quality is FULL when all
three inputs contributed and PARTIAL otherwise. A component may contain its
baseline and unit but omit value and change_percent when today's reading is not yet
available; that component was not included in the score. A missing component means
no usable data for it. HRV and resting-heart-rate values are rounded to whole units;
sleep hours retain two decimal places. change_percent is calculated before display
rounding, so do not recompute it from displayed values.
The result supports a cautious, personalized discussion of activity intensity;
it is not a medical diagnosis. If data is insufficient, the response explicitly
reports that instead of presenting a score. An optional daily_check_in contains the
user's four subjective 1-5 ratings as separate context. It never contributes to score,
quality, components_used, or components_missing, and may be present even when
physiological recovery data is insufficient. Its absence is normal and never makes
recovery status INSUFFICIENT_DATA or quality PARTIAL.Stress Platform: Health CoachStress PlatformPluginrequiredHealth & Wellness
- Today's personalized health planget · Retrieve the authenticated user's personalized wellness recommendations for today from their
personalized seven-day Stress Platform health plan. Use it when the user asks
what to focus on today, discusses their health plan, wants help following their
recommended sleep schedule or activity targets, or asks about progress against
the plan.
Plan targets are already personalized from the user's real historical patterns,
model insights, selected focus area, and the effort level they were willing to
make for this week. They may intentionally be incremental steps toward the user's
longer-term optimal zone. Treat these as the user's active personalized targets for the current week: do
not replace them with generic population goals, increase them toward an assumed
ideal, or criticize them for being lower than a universal recommendation.
ACTIVE returns only today's phase, daily tip, recommended sleep schedule, and
actionable targets. Tips are recommendations for this exact plan day; numbers that
describe a usual or average pattern are historical context, not today's live
measurement. sleep_hours, bedtime, and wake_up_time are recommended schedule
values.
To assess live adherence, also call get_daily_summary for the same local date and
compare only corresponding available measurements. Today's readings may be partial;
absence is unknown, not zero. Do not overwrite the plan's recommendation with the
observed value.
SCHEDULED means the plan starts on start_date. NO_ACTIVE_PLAN means there is no
current active week; guide the user to manage_plan_url to start a plan. If focus_area
is absent, the user has not configured one: do not infer a default area or tags, and
explain that plan setup on the portal will establish it. If focus_area is present,
it is the user's stored selection. The portal will handle whether this is a first
plan or a continuation after an earlier or expired plan, including any required
data analysis. WEEK_FINISHED means an active seven-day plan has passed its final
day; guide the user to manage_plan_url to review results and explicitly start the
next week. Never imply that the plan automatically renews or continues.
The plan supports general wellness behavior and is not a medical diagnosis.Stress Platform: Health CoachStress PlatformPluginrequiredHealth & Wellness
- Weekly health comparisonget · Compare two consecutive seven-day local-date periods: the most recent seven days
including today, and the seven days immediately before them. Use for questions
about recent progress or whether the last week differs from the previous week.
The response includes the exact inclusive date range for each period. The current
day may be partial. A metric is absent when neither period has data. If only one
period has data, a measured metric returns its available value and unit while
its missing value and percent change are absent; absence never means zero.
recent_observed_days and previous_observed_days state how many dates with usable
data contributed to each average; a missing date has an unknown cause and is not
zero. Values are rounded to whole practical units except sleep hours, which retain
two decimal places. change_percent is calculated by the platform before display
rounding, so do not recompute it from the displayed values.
sleep.sleep_efficiency_percent, when available, compares the percentage of each
recorded sleep session classified as asleep.
daily_check_in compares each subjective 1-5 dimension separately. change_points is
the absolute difference between the recent and previous averages on that scale; it is
not a percentage change. Use the reported observed-day counts before describing a trend.
Use get_daily_summary for one day and get_health_data for a detailed trend.Stress Platform: Health CoachStress PlatformPluginrequiredHealth & Wellness
- consultar_disponibilidadconsultar · Úsala cuando el usuario pregunte qué seguros puede comparar con IAseguro o pida una línea que no esté en la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil). Devuelve las líneas que se pueden comparar ahora mismo en España. No necesita datos.IAseguroIAseguroPluginnoneFinance
- cotizar_cochecotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de coche más barato en España ("cuánto pagaría por el seguro de mi coche", "seguro a todo riesgo más barato", "compárame el seguro del coche"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: fecha de nacimiento del conductor, código postal, años de carnet, modalidad (terceros, terceros ampliado, todo riesgo con o sin franquicia) y marca y año del vehículo. No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_decesoscotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de decesos más barato en España ("cuánto cuesta un seguro de decesos", "seguro de entierro más barato", "compárame seguros de decesos"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: fecha de nacimiento, código postal y tipo de prima (nivelada o natural). No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_hogarcotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de hogar más barato en España ("cuánto cuesta el seguro de mi piso", "seguro de hogar barato para inquilino", "compárame seguros de casa"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: tipo de vivienda (piso o casa), si es propietario o inquilino, superficie en m², código postal y nivel (esencial, completo o premium). No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_mascotacotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de mascotas más barato en España ("cuánto cuesta el seguro de mi perro", "seguro para gato barato", "compárame seguros de mascotas"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: especie (perro o gato), edad de la mascota y nivel (responsabilidad civil, completo o premium). No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_motocotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de moto más barato en España ("cuánto cuesta el seguro de mi moto de 125", "seguro de moto a terceros más barato", "compárame seguros de moto"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: modalidad, fecha de nacimiento, años de carnet, código postal y cilindrada. No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_patinetecotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de patinete eléctrico más barato en España ("cuánto cuesta el seguro del patinete", "seguro obligatorio de patinete barato"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: el nivel (responsabilidad civil o completo); el valor del patinete es opcional. No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_rccotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de responsabilidad civil familiar o individual más barato en España ("cuánto cuesta un seguro de responsabilidad civil", "seguro de RC familiar barato"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: el alcance (individual o familiar) y el capital asegurado. No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_viajecotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de viaje más barato en España ("cuánto cuesta un seguro de viaje a Japón", "seguro de viaje anual barato", "compárame seguros para mi viaje"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: zona de destino, tipo de viaje (puntual o anual), nivel (básico, completo o premium) y, si es puntual, los días. No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- cotizar_vidacotizar · Úsala cuando el usuario quiera saber cuánto cuesta, comparar precios o encontrar el seguro de vida más barato en España ("cuánto cuesta un seguro de vida de 100.000 €", "seguro de vida para la hipoteca", "compárame seguros de vida"). Devuelve una comparativa de precios orientativos de varias aseguradoras ordenada de más barata a más cara, con datos propios de IAseguro (no hace falta buscar en la web). Antes de llamarla reúne: fecha de nacimiento, código postal y capital a asegurar. No la uses para riesgos fuera de España ni para contratar: para eso enlaza a la web de IAseguro.IAseguroIAseguroPluginnoneFinance
- detalle_cochedetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de coche en España, en general o de una aseguradora concreta ("qué cubre el todo riesgo de AXA", "qué incluye un seguro a terceros ampliado"). Devuelve, desde la base de conocimiento de IAseguro, las modalidades, garantías (asistencia en carretera, lunas, coche de sustitución) y límites y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_decesosdetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de decesos en España, en general o de una aseguradora concreta ("qué incluye el seguro de decesos de Santalucía", "qué cubre un seguro de decesos"). Devuelve, desde la base de conocimiento de IAseguro, los servicios incluidos (sepelio, trámites, traslado), tipos de prima y asistencias añadidas y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_hogardetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de hogar en España, en general o de una aseguradora concreta ("qué cubre el seguro de hogar de Allianz", "cubre el seguro de casa los daños por agua"). Devuelve, desde la base de conocimiento de IAseguro, los niveles, garantías (daños por agua, robo, asistencia 24 h, responsabilidad civil) y límites y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_mascotadetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de mascotas en España, en general o de una aseguradora concreta ("qué cubre el seguro de mascotas de MAPFRE", "cubre el seguro del perro al veterinario"). Devuelve, desde la base de conocimiento de IAseguro, las especies y niveles, límite veterinario, si cubre perros PPP y edad máxima de alta y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_motodetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de moto en España, en general o de una aseguradora concreta ("qué cubre el seguro de moto de Mutua Madrileña", "qué incluye un seguro de moto a terceros"). Devuelve, desde la base de conocimiento de IAseguro, las modalidades, asistencia y cobertura del equipamiento del motorista y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_patinetedetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de patinete eléctrico en España, en general o de una aseguradora concreta ("qué cubre el seguro de patinete", "el seguro del patinete cubre el robo"). Devuelve, desde la base de conocimiento de IAseguro, los niveles, el capital de responsabilidad civil y si cubre robo y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_rcdetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de responsabilidad civil en España, en general o de una aseguradora concreta ("qué cubre un seguro de responsabilidad civil familiar", "cubre la RC los daños de mi hijo"). Devuelve, desde la base de conocimiento de IAseguro, el alcance, el capital máximo y qué incluye y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_viajedetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de viaje en España, en general o de una aseguradora concreta ("qué cubre el seguro de viaje de IATI", "cubre el seguro de viaje la cancelación"). Devuelve, desde la base de conocimiento de IAseguro, las zonas, niveles, límite de gastos médicos, cancelación, deportes y preexistencias y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- detalle_vidadetalle · Úsala cuando el usuario pregunte qué cubre, qué incluye o qué garantías tiene un seguro de vida en España, en general o de una aseguradora concreta ("qué cubre el seguro de vida de MAPFRE", "qué garantías tiene un seguro de vida"). Devuelve, desde la base de conocimiento de IAseguro, las coberturas, garantías opcionales (invalidez, enfermedades graves, doble capital por accidente) y edades asegurables y un precio "desde" orientativo, sin necesidad de datos personales ni de buscar en la web. Llámala siempre antes de responder de memoria. No la uses para líneas fuera de la lista (vida, coche, hogar, decesos, viaje, mascota, moto, patinete o responsabilidad civil).IAseguroIAseguroPluginnoneFinance
- Add Files to Directoryaddfilestodirectory · Add already-uploaded files to a directory, so they can be indexed. Takes file IDs from getUploadUrl or uploadFileByUrl. Files are added one by one, so the response reports which succeeded and which failed — retry only the failed ones. If the directory already backs an index, call syncIndex afterwards to pull the new files in.LlamaParseLlamaIndexPluginrequiredAI
- Classify Fileclassifyfile · Classify a file (based on specific categories) providing its file ID. Use with file IDs obtained with the getUploadUrl/uploadFileByUrl tool or that the user providedLlamaParseLlamaIndexPluginrequiredAI
- Create Directorycreatedirectory · Create a directory (folder) to hold source documents. A directory is what an index is built over: upload files, add them to a directory with addFilesToDirectory, then call createIndex on it.LlamaParseLlamaIndexPluginrequiredAI
- Create Extraction Config From Schemacreateextractionconfigfromschema · Create an extraction configuration from a JSON Schema you already have, and return its configuration id for use with `extractFile`. Supply either a `templateId` from `searchSchemaTemplates` (the server loads the schema for you) or an explicit `dataSchema` — a schema you wrote, or a template schema you edited. Unlike `generateExtractionConfig`, this needs no sample file and involves no LLM step.LlamaParseLlamaIndexPluginrequiredAI
- Create Indexcreateindex · Create an index over a directory, making its documents searchable with retrieveFromIndex and the other index tools. The directory must already contain the files you want indexed — upload them and call addFilesToDirectory first. Indexing runs in the background: the returned index is not queryable until getIndexStatus reports it ready.LlamaParseLlamaIndexPluginrequiredAI
- Estimate File Complexityestimatefilecomplexity · Estimate the parsing complexity of a PDF file (providing its file ID) using LiteParse. Returns a JSON object mapping each page with the LlamaParse tier it should be parsed with (or if you should use LiteParse), based on the parsing complexity and the need for OCR. Use in combination with parseFile and parseWithLiteParse. The tool needs a file ID obtained with the getUploadUrl/uploadFileByUrl tool or provided by the user. Only works with PDF files.LlamaParseLlamaIndexPluginrequiredAI
- Extract Structured Dataextractfile · Extract structured data from a file based on the configuration created with the `generateExtractionConfig` tool. Returns the extracted structured data.LlamaParseLlamaIndexPluginrequiredAI
- Extract Structured Data (Turbo)extractfileturbo · Extract structured data from a file in real time using the Turbo tier: one call, schema supplied inline, structured JSON back in seconds — no saved configuration step. Supply either a `templateId` from `searchSchemaTemplates` or an explicit `dataSchema` JSON Schema, plus a `fileId` from the upload tools. Turbo constraints: accepts only PDF, JPG and PNG; returns one object per document; produces no parse output; bills 35 credits/page. Prefer this tool whenever response time is user-facing. For per_page or per_table_row extraction, other file types, or a reusable saved configuration, use `createExtractionConfigFromSchema` + `extractFile` instead.LlamaParseLlamaIndexPluginrequiredAI
- Find Files in Indexfindfilesinindex · Search files within an index. Optionally provide the file name to filter for or a substring that should be contained in the file nameLlamaParseLlamaIndexPluginrequiredAI
- Generate Extraction Configgenerateextractionconfig · Generate the configuration to extract structured data from a specific file using the Extract service from the LlamaParse Platform. Provide a prompt describing what the schema of the extracted data, the ID of the file to extract and, optionally, a project ID.LlamaParseLlamaIndexPluginrequiredAI
- Get Index Statusgetindexstatus · Check whether an index has finished building. Indexing is asynchronous, so an index created or synced moments ago will not return results yet. Poll this until status is 'ready'; 'failed' means the build did not complete. Querying an index that is not ready looks identical to an index with no matching documents.LlamaParseLlamaIndexPluginrequiredAI
- Get Schema Templategetschematemplate · Retrieve the full JSON Schema for one starter extraction template, by id, as returned by `searchSchemaTemplates`. Pass the schema to `createExtractionConfigFromSchema` — edit it first if the user needs extra or fewer fields.LlamaParseLlamaIndexPluginrequiredAI
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.