MCP App Store
by KeenEthics
Navarch icon

Navarch MCP Server for ChatGPT

By SagentLab LLC

Overview

Connect Navarch to inspect your projects, running and completed tasks, blockers, agent activity, delivery analytics, and project health. Queue tasks through the linked GitHub repository and provide guidance on blocked work. Task creation can publish task text to GitHub and schedule agent work; guidance can resume execution. Access follows the connected account’s current project permissions. Requires an existing Navarch account.

How to connect

  1. ChatGPT: Open the Navarch listing in the ChatGPT app directory and choose Connect. ChatGPT listing Navarch exposes 24 tools here.

Tools

add_task

ChatGPT
Create one or up to 5 tasks in a selected project when the user asks to queue work. Supply request or requests. Publishes task text to the project's linked GitHub repository (which may be public), creates triage tasks, and schedules planning that can lead to agent execution. Requires project member access. Matching open requests are deduplicated; a later retry after completion may create new work.

cancel_qa_run

ChatGPT
Cancel a run and its open shards. Returns the run and board links. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

cancel_task_guidance

ChatGPT
Cancel one live guidance message that the agent has not read. A pending or waiting message is canceled at once (result canceled). A message already handed to the running session returns cancel_requested; the session then confirms the cancel or reports it was too late. Members cancel their own messages; project owners cancel any.

dispatch_qa_run

ChatGPT
Dispatch selected cases with trigger mcp. Returns the run and board links for all shard tasks. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

get_agent_activity

ChatGPT
Get bounded current assignment and recent activity evidence for agents in a selected project.

get_profile

ChatGPT
Identify the connected Navarch account. Returns its stable opaque user ID and available display name; use it to distinguish connected accounts.

get_project_delivery_analytics

ChatGPT
Get bounded monthly delivery analytics with safe null/unavailable values for a selected project.

get_project_health

ChatGPT
Get bounded task, blocker, capability, dependency, and milestone health evidence for a selected project.

get_qa_run

ChatGPT
Get a run and board links for its shards. Set include_items to page through case results with limit and offset. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

get_recent_completed_tasks

ChatGPT
Get up to eight recently completed tasks in a selected project.

get_running_tasks

ChatGPT
Get up to 20 currently claimed, in-progress, reviewing, or blocked tasks in a selected project.

get_task_guidance_status

ChatGPT
Get the delivery state of one live guidance message from submit_task_guidance: pending, dispatched, delivered, fell_back (waits for the next session), consumed, canceled or stale, with the effective mode and any fallback reason. Does not return the message text.

get_tasks_needing_guidance

ChatGPT
List blocked needs-human tasks in a selected project. With include_running and live steering enabled for the project, also list claimed, in-progress and reviewing tasks. Each result includes a short-lived grounding_reference required by submit_task_guidance.

list_projects

ChatGPT
List every Navarch project the authenticated user can currently access.

list_qa_run_schedules

ChatGPT
List QA schedules, next boundaries and last outcomes. Page with limit and offset. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

list_qa_runs

ChatGPT
List QA runs with bounded offset pagination. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

list_test_cases

ChatGPT
List mirrored test cases, filters, mirror status and a next cursor. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

list_test_suites

ChatGPT
List saved test suites, 100 per page. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

request_test_case_authoring

ChatGPT
Request an agent-authored test-case PR; active requests collapse per feature area. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

request_test_case_automation

ChatGPT
Create or reuse one automation build task per selected feature. Manual cases cannot be automated. Supply feature_path or case_ids. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

rerun_failed_test_cases

ChatGPT
Dispatch failed and blocked cases from a finished run with trigger mcp. Returns the new run and board links. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

save_test_suite

ChatGPT
Create a test suite or update it by suite_id. Selection uses tags_expr or case_ids. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

set_qa_run_schedule

ChatGPT
Create a scheduled suite run, or update by schedule_id. Requires a project owner because it delegates future repository configuration reads. Cron uses an IANA timezone and runs at most every 15 minutes. Agent shards spend model budget; automated shards do not. Agent shards spend model budget; automated shards do not. Auto mode uses agent shards until the runner is enabled.

submit_task_guidance

ChatGPT
Save the user's guidance on one task. A blocked task is requeued so agent execution resumes. With live steering enabled, a running task receives the message in its current session without a restart; the result gives message_id, state, effective_mode_preview and fallback_reason. Requires project member access and the task's short-lived grounding_reference from get_tasks_needing_guidance. Stale or cross-account references are rejected. A reference identifies one delivery and prevents duplicate guidance on retries.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

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/navarch"><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/navarch)