task-assigner

Suggest and assign the best-matched developer for a work.tasks row, via the deterministic-scoring + optional AI re-rank Edge Function already used by the dfl-task-assigner frontend.

Endpoint	https://engineering.mcp.devfellowship.com/mcp
Tools	1
Backing data	work.tasks (read + owner_id write), read of work.epics/work.projects/business_units for matching context.

Connect

Add https://engineering.mcp.devfellowship.com/mcp to your client (see Getting Started) with your Authorization: Bearer <token> header.

One package, multiple tool groups

dfl-mcp-engineering is a single umbrella package covering diagrams, spec-builder, task-assigner and documents tool groups — one MCP server instead of one per domain (decision from Tainan, 2026-07-09). This page documents only the task-assigner group; other groups get their own page here.

Task Assigner

Tool	Description
assign_developer	Suggests and assigns the best-matched developer for a work.tasks row. Calls the dfl-ai-task-assigner-matching-task Edge Function (deterministic scoring + optional AI re-rank), always writes the top-scored candidate to work.tasks.owner_id, and returns the full ranked suggestion list.

How matching works

assign_developer resolves the task’s context server-side — work.tasks.epic_id → work.epics.project_id → work.projects (name + business_unit_id) → business_units.name — then POSTs { task, context } to the dfl-ai-task-assigner-matching-task Edge Function using the calling user’s own JWT (same “golden rule” as every other tool in this package: never service_role). The Edge Function scores candidates deterministically (technology overlap, completed task count, average rating) and optionally re-ranks with an LLM, returning { developerId, deterministicScore, completedTasks }[] sorted by score.

The tool always assigns the top-scored candidate — there is no “suggest only” mode. If you need a different developer, call assign_developer to see the ranked candidates, then correct work.tasks.owner_id manually if the auto-pick isn’t right for this task.

No candidates ≠ error

When the Edge Function returns an empty candidate list (e.g. no completed tasks with owner_id set yet for that project’s epics), the tool returns {"status": "no_candidates"} and does not write owner_id — this is not treated as an error.