> ## Documentation Index
> Fetch the complete documentation index at: https://closedloop.sh/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Tools Reference

> Reference for every MCP tool in ClosedLoop AI: query product insights, Context, customers, Themes, and competitive intelligence from your AI dev agent.

## Read and write tools

Every tool has a readable title and tells your AI assistant whether it only reads your ClosedLoop AI data or makes a change. All tools only read, except these four:

| Tool | What it changes |
| - | - |
| `send_closedloop_feedback` | Saves your feedback about ClosedLoop AI for the ClosedLoop AI team. |
| `create_watch` | Saves a watch that notifies you (Slack, Teams, email or a webhook) when it fires. |
| `create_proof_permission_request` | Logs a test permission request. It never emails the customer. |
| `update_proof_permission_status` | Changes the status of a test permission request. |

## Overview & Discovery

### get\_overview

Get a high-level summary of your product intelligence landscape. Use this as a starting point.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `time_range` | `7d` `30d` `90d` `all` | `30d` | Time window |

Returns: total insights, deal blockers, churn risks, top categories.

### get\_facets

Get aggregated counts by dimension. Use to understand distributions before drilling in.

| Parameter | Type | Required | Description |
| - | - | - | - |
| `dimensions` | array | yes | Which dimensions: `category`, `severity`, `feature_area`, `emotion`, `country`, `industry`, `source_type`, `account_status` |
| `date_from` | string | no | Start date YYYY-MM-DD |
| `date_to` | string | no | End date YYYY-MM-DD |

Returns: top 20 values with counts per dimension. `source_type` separates calls, support, and email. `account_status` counts insights by the status their customer's account has in your CRM.

### get\_trends

Get time-bucketed counts for trend analysis.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `query` | string | none | Topic to trend. Leave empty for overall volume |
| `category` | string | none | Filter by category |
| `feature_area` | string | none | Filter by feature area |
| `granularity` | `daily` `weekly` `monthly` | `weekly` | Time bucket size |
| `date_from` | string | 90 days ago | Start date |
| `date_to` | string | today | End date |

Returns: time-bucketed counts with unique customers and average frustration per period.

***

## Customer Intelligence

### search\_customers

Find customers by name, industry, country, or engagement metrics. Every result includes revenue context.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `query` | string | none | Customer name, domain, or keyword |
| `industry` | string | none | Filter by industry |
| `country` | string | none | Filter by country |
| `has_active_deals` | boolean | none | Only customers with open deals |
| `has_deal_blockers` | boolean | none | Only customers with deal blockers |
| `has_churn_risks` | boolean | none | Only customers with churn risks |
| `min_insights` | integer | none | Only customers with N+ insights |
| `sort` | `insight_count` `arr` `deal_value` `frustration` `recency` `churn` `name` | `insight_count` | Sort order. `arr` ranks by what each customer pays now; `churn` ranks churned, then at-risk customers first |
| `limit` | integer | 10 | Max results (1-50) |
| `offset` | integer | 0 | Skip first N results |

Returns: customer profiles with CRM data, deal values, and intelligence stats. Customers with equal sort values come in customer ID order, so pages never repeat or skip a customer. Customers with nothing to sort by come last, in customer ID order, after negative values too: no product insights for `insight_count` and `recency`, no current ARR and no product insights for `arr`, no won deal value for `deal_value`, no frustration score for `frustration`, and no churn state, stated churn or product insights for `churn`.

### get\_customer

Get full profile for a specific customer.

| Parameter | Type | Required | Description |
| - | - | - | - |
| `id` | string | yes | Customer ID from search\_customers. Pass the full ID, never a shortened one |

Returns: CRM data, deals, contacts, intelligence summary, strategic context, and the account family the customer is counted under, with each of its accounts' CRM record id.

***

## Product Intelligence

### search\_insights

Find product intelligence by keyword or meaning. Returns 300-character content previews and a customers\_mentioned summary.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `query` | string | none | What to search for. Works with natural language |
| `customer_id` | string | none | Filter to one customer. Pass the full ID, never a shortened one |
| `category` | `bug` `feature request` `improvement` `integration issue` `missing functionality` `ux/ui issue` `documentation` `performance issue` `security issue` `reporting` | none | Filter by category |
| `severity` | `Critical` `High` `Medium` `Low` | none | Filter by severity |
| `is_deal_blocker` | boolean | none | Only deal-blocking insights |
| `is_churn_risk` | boolean | none | Only churn-risk insights |
| `feature_area` | string | none | Filter by product area |
| `emotion` | string | none | Filter by emotion (frustrated, disappointed, confused, hopeful, etc.) |
| `source_type` | `calls` `support` `email` | none | Filter by source. Email results include parsed forwarded-email threads and HubSpot emails behind matching product insights |
| `country` | string | none | Filter by customer country |
| `industry` | string | none | Filter by customer industry |
| `account_status` | array of strings | none | Filter by the status the customer's account has in your CRM, in your CRM's own words (any case), for example `["Prospect", "Working"]`. Insights whose account has none of them are left out |
| `product_ids` | array of strings | none | Insights filed under any of these products: up to 50 full IDs from your product structure |
| `area_ids` | array of strings | none | Insights filed under any of these areas: up to 50 full IDs from your product structure |
| `feature_ids` | array of strings | none | Insights filed on any of these features: up to 50 full IDs from your product structure |
| `date_from` | string | none | Start date YYYY-MM-DD |
| `date_to` | string | none | End date YYYY-MM-DD |
| `sort` | `relevance` `severity` `frustration` `recency` | `relevance` | Sort order |
| `limit` | integer | 20 | Max results (1-100) |
| `offset` | integer | 0 | Skip first N results |

Returns: matching insights with content previews, plus a `customers_mentioned` summary of the customers behind the insights on that page. With a `query`, the matches form one ranked list of at most 300 insights: `offset` and `sort` page through it and `total_matches` is its size. `semantic_matching_available: false` means only keyword matches came back, so a topic with no match may still have been raised in other words; pages only follow on from each other while that value stays the same.

### get\_insight

Get full detail for a specific insight.

| Parameter | Type | Required | Description |
| - | - | - | - |
| `id` | string | yes | Insight ID from search\_insights. Pass the full ID, never a shortened one |

Returns: full verbatim content, pain point, workaround, use case, business outcomes, kano category, and reporter details. The response includes a `conversation` pointer (`id` and `source`) to the call, ticket, or other source the insight came from, or `null` when no openable source is stored; pass both to `get_conversation` to read the full source. For an email insight, the response also includes every exact supporting span in `evidence`, ordered as used by the insight. Each span identifies its source part; Email Capture spans also identify their message.

***

## Conversation Evidence

### list\_conversations

List calls, support tickets, and email threads for a customer. Prefer `customer_id`; name lookup is a fallback.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `customer_id` | string | none | Customer ID from `search_customers`. Pass the full ID, never a shortened one |
| `customer_name` | string | none | Customer or account name when no ID is available |
| `source_type` | `calls` `support` `email` `all` | `all` | Filter by conversation source |
| `limit` | integer | 20 | Max results (1-50) |
| `customer_scope` | `family` `entity` | `family` | Expand to the customer family or use only the selected CRM entity |

Returns: conversation metadata without transcript content. Each Email Capture thread appears once, with the matched account, participants, ordered delivery IDs, message count, and source-completeness warnings. Each HubSpot email behind a matching product insight appears as one conversation with its source, date, participants and insight count.

### get\_conversation

Get the source content for one conversation returned by `list_conversations` or by the `conversation` pointer of `get_insight` or `get_signal`.

| Parameter | Type | Required | Description |
| - | - | - | - |
| `id` | string | yes | Conversation ID from `list_conversations` or from a `conversation` pointer. This is the ClosedLoop AI ID, not the ID the source system shows in its own links. For Email Capture, a thread ID or one of its delivery IDs; for HubSpot email, the email conversation ID |
| `source` | `gong` `fireflies` `granola` `plaud` `readai` `google_meet` `noota` `intercom` `freshdesk` `ada` `typeform` `slack` `mcp` `canny` `uservoice` `email` `hubspot_email` | yes | Source system, as given by `list_conversations` or the `conversation` pointer |

Google Meet, Noota, Typeform, Slack, Canny, UserVoice, submitted-feedback and HubSpot email conversations can be opened only when they are the source of at least one insight or context record.

For email, returns the complete parsed thread in chronological order as structured messages with sender, recipients, subject, body, and extracted attachments. It also returns a compatible `transcript`, the matched account, participants, ordered delivery IDs, and `source_complete`.

For `hubspot_email`, returns the stored source used for analysis, with its subject, sender, participants, transcript and attributed body parts. Product and strategic evidence quotations refer to this source even if HubSpot later changes the logged email. Original MIME headers and attachments are not part of the stored HubSpot source.

For `email`, this is the complete **parsed** source, not raw MIME or attachment bytes. Check `source_complete` and `warning_codes`: `false` means parsing, extraction, or response limits omitted some source content, and the warnings explain why.

***

## Sales Intelligence

### search\_signals

Search strategic intelligence: satisfaction, churn, competitor mentions, buying behavior.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `query` | string | none | What to search for |
| `type` | string | none | Context type: `satisfaction`, `churn_reason`, `competitor_mention`, `decision_criteria`, `purchase_hesitation`, `win_reason`, `onboarding_friction`, and more |
| `customer_id` | string | none | Filter to one customer. Pass the full ID, never a shortened one |
| `date_from` | string | none | Start date YYYY-MM-DD |
| `competitor_name` | string | none | Filter by competitor name. Use get\_competitors to see valid names |
| `date_to` | string | none | End date YYYY-MM-DD |
| `sort` | `relevance` `recency` | `relevance` | Sort order |
| `limit` | integer | 20 | Max results (1-100) |
| `offset` | integer | 0 | Skip first N results |

Returns: strategic context records with content previews, churn urgency, satisfaction level, competitor names. With a `query`, the matches form one ranked list of at most 300 records that `offset` and `sort` page through, `total_matches` is its size, and `semantic_matching_available: false` means only keyword matches came back.

### get\_competitors

List your configured competitors with mention counts from customer conversations. Use this to see which competitors come up most, then drill in with `search_signals(competitor_name="...")`.

No parameters required.

Returns: competitor list with names, products, mention counts, and whether they're a direct competitor.

### get\_signal

Get full detail for a specific context record.

| Parameter | Type | Required | Description |
| - | - | - | - |
| `id` | string | yes | Context record ID from search\_signals. Pass the full ID, never a shortened one |

Returns: full content, context, source URL, and competitor details. The response includes a `conversation` pointer (`id` and `source`) to the call, ticket, or other source the record came from, or `null` when no openable source is stored; pass both to `get_conversation` to read the full source. For email, it also returns every exact supporting span in ordered `evidence`.

### search\_opportunities

Find the biggest Themes ranked by business impact. In tool names, `opportunity` means Theme.

| Parameter | Type | Default | Description |
| - | - | - | - |
| `query` | string | none | Search by topic |
| `status` | `open` `active` `new` `reviewing` `planned` `in_progress` `parked` `shipped` `declined` `all` | `open` | Filter by status. `open` is every status except shipped, declined and parked |
| `velocity_status` | `accelerating` `active` `cooling` `dormant` | none | Filter by growth trend |
| `sort` | `ric_score` `signal_count` `customer_count` `deal_blocker_count` `recency` | `ric_score` | Sort order |
| `min_signals` | integer | 10 | Minimum insight count |
| `has_deal_blockers` | boolean | none | Only Themes with deal blockers |
| `product_ids` | array of strings | none | Themes with an open need in any of these products: up to 50 full IDs from your product structure |
| `area_ids` | array of strings | none | Themes with an open need in any of these areas: up to 50 full IDs from your product structure |
| `feature_ids` | array of strings | none | Themes with an open need on any of these features: up to 50 full IDs from your product structure |
| `date_from` | string | none | Themes created on or after this date, YYYY-MM-DD |
| `date_to` | string | none | Themes created on or before this date (inclusive), YYYY-MM-DD |
| `limit` | integer | 10 | Max results (1-50) |
| `offset` | integer | 0 | Skip first N results |

Returns: Themes with RIC scores (reach/impact/confidence), customer counts, severity breakdown, and velocity status. A Theme can match several products, areas or features, and its counts always cover the whole Theme. With a status other than `open`, a Theme with no open needs is found by its closed needs. `declined` returns what a person declined; Themes and needs retired by processing are never returned.

### get\_opportunity

Get full detail for a specific Theme, including evidence and revenue impact. In tool names, `opportunity` means Theme.

| Parameter | Type | Required | Description |
| - | - | - | - |
| `id` | string | yes | Theme ID from `search_opportunities`. Pass the full ID, never a shortened one |

Returns: full evidence (top 50 quotes), customer list with IDs and deal values, `total_won_deal_value`, and Features associated with parent Themes.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.