1. Conduid
  2. Agents
  3. Frink
MCP server · Agents

Frink

Desktop-first, open-source (MIT) application for running parallel agents, and multi-step custom workflows triggered by schedules, webhooks or events from tools like Linear and Sentry, across your already-used providers.

63Good

Scored 3 hours ago · breakdown

About Frink

Frink is an MCP server published by benord-labs in the Agents category: desktop-first, open-source (MIT) application for running parallel agents, and multi-step custom workflows triggered by schedules, webhooks or events from tools like Linear and Sentry, across your already-used providers.

The repository has 181 stars and 7 forks, with the last commit 10 days ago. Recent commit activity is the strongest signal that reported issues will get attention.

Install

From source
git clone https://github.com/benord-labs/frink

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Tools (15)

fetchAllProjectsList all projects available on this machine.

Parameters not published.

frink_batch_messageSend a message from the CEO agent to employee agents in a batch. Use for broadcasting corrections, follow-up instructions, or targeted guidance to a specific run. Broadcast: omit flowRunId. Targeted: provide flowRunId. Active runs queue the message for delivery on completion; terminal runs are marked delivered immediately.

Parameters not published.

frink_flows_add_stage_runsAdd runs to an existing stage incrementally (max 50 per call, max 20 calls per session). Use this after frink_flows_define_stages to add more runs to any stage (root or non-root) before calling frink_flows_start_batch. All stages remain pending until start_batch is called — runs added here are also pending. Per-run errors are reported without aborting the whole call. For epic-scale work (e.g. 144

Parameters not published.

frink_flows_define_stagesDefine staged execution for a batch, with optional DAG dependencies between stages. All stages start as pending — call frink_flows_start_batch after planning is complete to begin execution. This allows multi-call DAG assembly (up to 50 stages per call) for large epics without starting work prematurely. The server validates the flow graph in run mode before creating stages. Non-root stages fire aut

Parameters not published.

frink_flows_getGet the graph and config of an existing Frink Flow by ID.

Parameters not published.

frink_flows_get_batchInspect batch/run execution state for a flow. Three modes: (1) omit batchId — list the flow’s recent runs (status, batch_id, timing); (2) pass batchId — batch summary with completed/failed/active counts + per-run status (optional status filter). The counts and batch_total always cover every run in the batch, whatever limit and status are; total is the number of runs matching the status filter, and

Parameters not published.

frink_flows_get_runInspect a flow run: get status, per-node outcomes, errors, and durations. Call this after frink_flows_run to check results or debug failures. Pass wait: true to block until the run finishes (polls internally with backoff, default 120s timeout) — preferred over manual poll loops. Returns a compact summary; pass nodeRunId to drill into a specific node

Parameters not published.

frink_flows_listList existing Frink Flows. Optional projectId narrows to flows assigned to that project plus workspace-wide flows (no project). Omit projectId to list all flows for the user.

Parameters not published.

frink_flows_list_catalogList a reference catalog for building flows. kind selects which:

Parameters not published.

frink_flows_patchCreate a new flow or modify an existing one using incremental operations. To CREATE: pass name + operations (creates the flow and applies ops to an empty graph). To MODIFY: pass flowId + operations. The graph is saved after each call even with incomplete configs — stricter run-mode validation (URLs, webhooks, agent instructions, etc.) runs when you start the flow and when the server creates batch

Parameters not published.

Show all 15 tools
frink_flows_runStart or queue a flow run using the latest published graph version. A successful response status is

Parameters not published.

frink_flows_start_batchActivate a planned batch. Call this after frink_flows_define_stages (and any frink_flows_add_stage_runs calls). Root stages transition from pending to running and stage runs are submitted to Flow admission up to the per-stage dispatch limit; admitted runs may still wait in the global queue. A true started value means at least one root stage was activated, not that a run began executing; totalEnque

Parameters not published.

frink_navigation_contextRead the chat

Parameters not published.

requestSwitchProjectMove this chat to a different project. The user must approve. On success, the runtime performs an execution handoff and resumes in the new project context with the correct working directory.

Parameters not published.

searchProjectsSearch projects by keywords matched across name, description, and path (every word must appear; e.g.

Parameters not published.

Ask AI

✦

Ask AI about Frink

Powered by Claude

Ask about Frink — answers draw on its README.

Security checks

6 checks have not been run yet.

Questions

About Frink

How do I install Frink?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Frink safe to use with an AI agent?

Its trust score is 63 out of 100 (good). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Frink still maintained?

Yes — the last commit was 10 days ago. The repository has 181 stars and 7 forks.