Nerve Kanban Bwm
WarnAudited by ClawScan on May 10, 2026.
Overview
The skill matches its Kanban-management purpose, but it can change/delete board data, change workflow policy, and start background agent runs without documented approval limits.
Review before installing. This appears to be a straightforward Nerve Kanban API reference, but allow it only if you trust the Nerve server and are comfortable with the agent changing board data and starting subagent runs. Confirm destructive actions, configuration changes, and task execution manually.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
An agent using the skill could delete tasks, approve or reject work, start execution, or change board settings if it decides those calls are relevant.
The skill exposes destructive, workflow-changing, agent-spawning, and configuration-changing API operations, but the instructions do not add confirmation or scoping guidance for these high-impact actions.
| Delete task | DELETE | `/api/kanban/tasks/:id` | ... | Execute (spawn agent) | POST | `/api/kanban/tasks/:id/execute` | ... | Update config | PUT | `/api/kanban/config` |
Require explicit user confirmation for delete, approve/reject, execute, and configuration updates; prefer read-only actions or proposals unless the user clearly asks for a mutation.
Starting a Kanban task can create autonomous background work outside the immediate chat, potentially consuming resources or acting on task instructions without further review.
The execute endpoint starts a separate agent session and background polling behavior, but the artifacts do not define the subagent's permissions, lifetime, resource limits, or required user approval.
**Side effects:** Spawns a gateway subagent session with label `kb-<id>`. Background poller watches for completion and auto-transitions to `review`.
Only execute tasks after explicit operator approval, and ensure the Nerve server provides clear stop/abort controls, permission limits, logging, and task-scoped execution.
A bad or unintended configuration change could weaken review requirements or auto-approve future proposals across the board.
Board configuration includes review/proposal policy controls that can affect future tasks and agent workflows, and SKILL.md documents `PUT /api/kanban/config` without guardrails.
allowDoneDragBypass: boolean; ... proposalPolicy: 'confirm' | 'auto'; ... defaultModel?: string; defaultThinking?: ThinkingLevel;
Restrict configuration updates to trusted operator/admin actions, require confirmation, and keep an audit trail with an easy rollback path.
If the Nerve server is exposed or weakly protected, run-completion data or task status changes could be accepted from an unintended source.
The skill documents a webhook-style completion path and agent-run flow, but the supplied artifacts do not describe how webhook origin, authentication, or data boundaries are enforced.
| Complete run (webhook) | POST | `/api/kanban/tasks/:id/complete` |
Use this only with a trusted, access-controlled Nerve server; require authentication or signed callbacks for webhook-style completion endpoints.