MCP App Store
by KeenEthics
Hardline icon

Hardline MCP Server for Claude and ChatGPT

By Hardline AI Corp.

Overview

Hardline MCP connects your notes from your phone calls and site meetings to Claude. Use this MCP when you are looking to connect real time data from the jobsite into your tools and workflows. - Writing a daily log: Pull the decisions, commitments, and delivery updates from today's calls straight into the log, instead of backfilling from memory hours later. - Drafting an RFI or change order: Pull the field context behind it. What the architect approved, what the sub flagged, what changed on the call, all in the draft. - Settling a "you told me to" dispute: Find who committed to what and when, with the exact call it came from. The contemporaneous record, on demand. Hardline helps you connect valuable information and photos to the rest of your system.

How to connect

  1. Claude: Open the Hardline listing in the Claude connectors directory and choose Connect. Claude listing Hardline exposes 22 tools here.
  2. ChatGPT: Open the Hardline listing in the ChatGPT app directory and choose Connect. ChatGPT listing Hardline exposes 53 tools here.

Tools

archive_project

ChatGPT
Archive a Hardline project. This disables new project writes and Assistant runs.

associate_source_project

ChatGPT
Associate a source passage privately; never change source or task sharing.

associate_task_project

ChatGPT
Add one private project association while preserving existing links and audiences.

create_contact

ChatGPT
Create a contact for the signed-in user.

create_project

ChatGPT
Create a Hardline project and make the signed-in user its owner.

create_task

ChatGPT
Create a task, optionally linked to one call or artifact.

delete_contacts

ChatGPT
Soft-delete contacts by explicit contact IDs only.

delete_tasks

ChatGPT
Soft-delete tasks by explicit task IDs only.

draft_task_follow_up

ChatGPT
Prepare a private draft for an existing conversation. Never sends or replaces a draft.

get_artifact

ChatGPT
Retrieve one note or capture artifact with metadata, results, tasks, and media. For exact accepted saved Capture Notes or Action Output, use view="notes". This returns bounded document text, distinct from an audio transcript or generated summary. Continue with notes_offset=saved_notes.range.next_offset while has_more is true. Omit view to keep the legacy metadata/media response. Notes view does not fetch or sign media attachments.

get_artifact_image

ChatGPT
Return a primary or attached artifact image as MCP image content when small enough. Use this read-only tool when a photo's contents or relevance need visual inspection before including it in a draft or provider write. Pass the exact mediaId from capturePhotos or list_artifact_media with its owning artifact_id. Inspection does not upload or associate the image with a provider record; use the provider tool's contract for those separate operations. Omit media_id to return the primary image, or the only attached image. If an artifact has multiple attached images and no primary image, call list_artifact_media first and pass the selected media_id.

get_call

ChatGPT
Retrieve call metadata, participants, summary, recordings, tasks, and transcript overview.

get_call_transcript

ChatGPT
Retrieve a call transcript as timestamped segments or plain text. start_time and end_time are deprecated and ignored for call-level transcripts.

get_contact

ChatGPT
Retrieve one contact.

get_daily_activity

ChatGPT
Return calls, notes/capture artifacts, and tasks for a bounded day/window. Prefer this for daily logs, today's notes, field notes, or capture notes. Notes may use several artifact types, so include artifacts rather than filtering only to type="note".

get_document_pages

ChatGPT
Retrieve exact extracted page text from one uploaded document. Call this after search_artifacts returns document evidence when a question asks for an exhaustive list, a comparison, or details spanning adjacent pages. Request at most 20 page numbers per call. When pages is omitted, pages are returned from the beginning until max_chars is reached.

get_follow_up_candidates

ChatGPT
Return open tasks and recent call/artifact sources likely needing follow-up.

get_messaging_conversation

ChatGPT
Get one account-wide messaging conversation and its current participants.

get_messaging_mailbox

ChatGPT
Get the signed-in user's messaging entitlement, mailbox state, and send capabilities.

get_messaging_message

ChatGPT
Get one message with current aggregate and per-recipient delivery status.

get_messaging_timeline

ChatGPT
Read messages and system events from a conversation without marking it read.

get_project

ChatGPT
Retrieve one accessible Hardline project by its explicit project ID.

get_project_plan

ChatGPT
Read one native plan and tasks; optionally inspect actual full-sheet or cropped pixels. region is a full-sheet normalized rectangle (left/top/right/bottom). Expand tiny indexed text boxes to inspect room boundaries. The image manifest states its full-sheet frame. Reinspect current pixels before visual placement; pass inspectionReceipt to the link tool.

get_project_snapshot

ChatGPT
Return a bounded deterministic snapshot; this never generates or persists an AI brief.

get_project_task

ChatGPT
Read an exact native project task including extension revision and current capabilities. Use returned updatedAt/revision as expected_updated_at/expected_revision for a reviewed native task-plan link. Numeric task references and names must first be resolved to native IDs.

get_task

ChatGPT
Retrieve one task with linked call/artifact context.

get_task_catalog

ChatGPT
Discover valid task vocabulary and directory identities, one bounded section at a time.

learn_miriel

ChatGPT
Teach Miriel new content for the signed-in user's context.

list_artifact_media

ChatGPT
List ready media attachments for an artifact with short-lived signed URLs. Discover photos before preparing a photo-bearing draft or publication. Attached-artifact capturePhotos is a bounded snapshot of direct, non-excluded photos: call this tool when hasMore is true or readiness may have changed, including during a voice session. This tool lists current READY direct attachments, not pending/failed uploads or linked-source photos. Keep unavailable counts from the context explicit; absence here does not prove upload success. Use get_artifact_image to inspect a photo when its relevance or contents are uncertain. To upload media to an integration, pass artifact_id and mediaId to its upload tool instead of copying viewUrl; a copied signed URL fails if any character changes. Use the exact capturePhotos.mediaId for upload-tool media_id and get_artifact_image; capturePhotos.attachmentId (publicId in this tool's result) is the UUID for document attachment values and typed photo fields in managed tools. Keep each filename in draft photo references without calling write tools.

list_messaging_conversations

ChatGPT
List account-wide SMS/MMS conversations without changing unread state.

list_projects

ChatGPT
List accessible Hardline projects with IDs, names, counts, permissions, and activity. Use query to resolve a project name or address to an explicit Hardline project ID before filtering or mutating other resources. External provider project IDs are not accepted.

list_task_activity

ChatGPT
Read ordered task activity after a sequence, retaining actor and child identities.

list_task_comments

ChatGPT
Read task discussion separately from field and source activity, using the returned continuation.

list_tasks

ChatGPT
List canonical tasks with filters and pagination. Use view=all for obligation matching, including reviewed and closed work. Reuse nextCursor only with the same query.

query_miriel

ChatGPT
Ask Miriel using the signed-in user's learned Hardline context.

restore_project

ChatGPT
Restore an archived Hardline project and re-enable authorized project writes.

retry_messaging_message

ChatGPT
Retry only the failed recipients of a terminally failed outbound message.

review_task

ChatGPT
Explicitly review pending work or mark a duplicate. This does not approve unrelated field changes.

search_artifacts

ChatGPT
Search the signed-in user's artifacts, including notes and capture sessions, documents, call-derived artifacts, photos, and videos. A user's "notes" may be note, text-message, audio-recording, or voice-note artifacts, so do not filter to type="note" unless the user explicitly asks for typed notes only. Search by semantic meaning, keyword, and metadata. Use type="document" for uploaded-document questions. Those results include Percept evidence, exact locally extracted page evidence, stable numbered references, and a document catalog. Use artifact_ids with type="document" to restrict retrieval to specific documents. Empty queries use keyword/recency search even when semantic mode is requested. Hybrid ranks punctuation-normalized exact titles and phrases first, then semantic evidence, required-token keyword matches, and bounded spelling fallbacks.

search_calls

ChatGPT
Search calls by semantic meaning, metadata, phone number, transcript, and tasks. Empty queries use keyword/recency search even when semantic mode is requested. Hybrid ranks semantic distance matches first, then appends literal keyword matches by recency.

search_contacts

ChatGPT
Search contacts by name, phone, email, company, source, and favorite status.

search_messaging

ChatGPT
Search account-wide conversations by participant/title and message text.

search_project_plans

ChatGPT
Search native Hardline plans by sheet metadata and indexed room, label and note text. Returned evidence includes exact native plan IDs and full-sheet normalized text boxes. Investigate content matches and inspect the relevant crop before locating a room or pin. Unindexed plans still match metadata; index status distinguishes missing coverage from no hit.

search_tasks

ChatGPT
Search the signed-in user's extracted and manually created tasks.

search_workspace_context

ChatGPT
Search all account activity across notes/capture artifacts, documents, calls, tasks, and learned context. Use this for broad questions that could span multiple evidence sources. Results remain scoped to the signed-in account. Use source-specific tools to retrieve exact document pages, media, or complete call transcripts from the strongest matches.

send_messaging_message

ChatGPT
Send one SMS/MMS after host approval; returns queued, not delivered, status. Provide exactly one destination: conversation_id or recipient_phone_numbers. Resolve people with contact or messaging search first, never guess a number. At least text or one attachment is required. Existing messaging media and visible Hardline artifact media are supported.

update_contact

ChatGPT
Update one contact by explicit contact ID.

update_project

ChatGPT
Update current Hardline project metadata after approval; no resources are relinked.

update_tasks

ChatGPT
Bulk update tasks by explicit task IDs only.

verify_task

ChatGPT
Explicit human verification of current material evidence, separately from completion.

create_contact

Claude

create_task

Claude

delete_contacts

Claude

delete_tasks

Claude

get_artifact

Claude

get_artifact_image

Claude

get_call

Claude

get_call_transcript

Claude

get_contact

Claude

get_daily_activity

Claude

get_follow_up_candidates

Claude

get_task

Claude

learn_miriel

Claude

list_artifact_media

Claude

query_miriel

Claude

search_artifacts

Claude

search_calls

Claude

search_contacts

Claude

search_tasks

Claude

search_workspace_context

Claude

update_contact

Claude

update_tasks

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/hardline"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/hardline)