namakan-mcp-projects
by cfollette18
README.md
# namakan-mcp-pm
Internal MCP server: one project/work-management surface. Namakan agents call `pm_find_task` whether the backend is Jira, Asana, Monday, ClickUp, Smartsheet, Microsoft Project, or Trello.
`delivery-manager` uses the same tools to project engagement status into the client's board. The client PMO does not configure MCP.
## How agents connect
```yaml
mcp_servers:
namakan-mcp-pm:
command: uvx
args:
- --from
- git+https://github.com/cfollette18/namakan-mcp-pm.git
- namakan-mcp-pm
- serve
env:
NAMAKAN_PM_BACKEND: mock
trust: full
```
Bootstrap writes that into every Hermes profile with `using-namakan-mcp`.
## Architecture

## Full AI workflow

1. `president-coo` assigns delivery tracking to `delivery-manager`.
2. Agent calls `pm_run_workflow` (list → find → blocked status update).
3. Writes stay off until `ciso` approval. Dry-run shows the would-be call.
Demo without an agent process:
```bash
namakan-mcp-pm workflow
```
Expected: four JSON steps. Last step `"error": "writes_disabled"`.
## Tools
| Tool | Access | Arguments |
|---|---|---|
| `pm_find_task` | read | `query` |
| `pm_list_tasks` | read | `project_id?` |
| `pm_list_projects` | read | — |
| `pm_update_status` | write | `task_id`, `status` |
| `pm_add_comment` | write | `task_id`, `body` |
| `pm_run_workflow` | read + blocked write | `use_case?`, `query?` |
Mock statuses: `todo`, `doing`, `blocked`, `done`. Map vendor states in the adapter, not in the tool name.
## Potential use cases

| Use case | Which agent | Why it matters |
|---|---|---|
| Engagement board | `delivery-manager` | Mirror Phase 1/2 status into the client's Jira without learning Jira's API. |
| Where is the audit? | `delivery-manager` | Mock today, Asana after go-live. Same tool. |
| Close the task | `ciso` gates the write | Blocked until writes are enabled. |
`use_case`: `engagement-board` (default), `where-is-the-audit`, `close-the-task`.
## Tests
```bash
pip install -e ".[dev]"
pytest
```
## License
MIT. Copyright (c) 2026 Namakan AI Engineering.
This server cannot be deployed
Maintenance
ActivityMaintained
ResponsivenessNo issues