TaskMaster MCP Bootstrapper
Set up the TaskMaster MCP server and wire it into BMAD's Skills + Hooks workflow. Use this skill whenever a project needs TaskMaster's task engine instead of the lightweight Markdown-only flow.
When to Invoke
Activate when the user:
- Wants Claude to "create", "install", or "bootstrap" the TaskMaster MCP server.
- Requests a managed TaskMaster environment that BMAD Skills can call.
- Needs guidance configuring TaskMaster's API keys, persistence, or hook automation.
- Wants to update/upgrade an existing TaskMaster deployment to a new version.
Skip this skill if the user only needs Markdown-based task tracking (use the Skills-only bundle) or is troubleshooting TaskMaster logic (hand off to TaskMaster docs/support).
Responsibilities
You are accountable for:
- Environment Readiness – confirm Node.js/npm/git availability and workspace permissions.
- Server Provisioning – fetch TaskMaster MCP via
npxor git clone, install dependencies, and prepare launch scripts. - Configuration – collect API keys, choose tool subsets, set persistence paths, and produce
mcp.json/.enventries. - Verification – run health checks, list exposed tools, and confirm TaskMaster can read/write the shared
tasks.mdfiles. - Hook Alignment – outline how Hooks (load on session start, save on updates) integrate with the MCP endpoints.
- Handover – document the final state, command snippets, and recovery actions for operators.
Use WORKFLOW.md for the ordered procedure and CHECKLIST.md as the go/no-go gate before declaring success. Load REFERENCE.md when you need detailed env var guidance, editor-specific MCP config snippets, or troubleshooting steps.
Guardrails
- Do not embed private API keys in responses; instruct the operator to supply them interactively.
- Keep configuration templates minimal—placeholder tokens should make it obvious where secrets go.
- Prefer deterministic scripts (see
scripts/bootstrap_taskmaster.sh) for repetitive shell work; describe how to run them rather than pasting large command blobs repeatedly. - Call out manual steps the user must complete outside Claude (e.g., storing secrets, restarting an editor).
- If the environment lacks required tools or permissions, stop and escalate instead of improvising unsupported installs.
Quickstart Snapshot
- Review
WORKFLOW.md→ confirms prerequisites, chooses install mode (globalnpxvs managed clone). - Gather secrets per
REFERENCE.md→ decide which providers to enable and capture required tokens. - Execute provisioning commands → either run the bootstrap script or craft manual shell steps tailored to the user's OS.
- Produce editor MCP config +
.envscaffolding → highlight key paths and placeholders. - Validate with
npx task-master-ai --helpandtask-master-ai status→ confirm TaskMaster reports enabled tools. - Align BMAD Hooks → restate how session-load/save hooks call the MCP-managed Markdown files.
Once the checklist passes, hand back a concise operations summary: version deployed, commands to start/stop, file locations, and follow-up tasks (e.g., schedule backups).
Hand-off Artifacts
Every successful run should deliver:
- Installation log or summary (commands executed, directories created).
- Final MCP configuration block with placeholders swapped for user-provided secrets.
- Instructions for starting/stopping the server and verifying health.
- Recommendations for connecting BMAD Hooks, archiving
tasks.md, and future upgrades.
Document assumptions and any deviations from the standard workflow so future sessions can resume confidently.