hacknplan-mcp
Enables Claude to interact with HacknPlan project management via the MCP protocol. Provides tools for managing projects, work items, boards, milestones, time logs, design models, and cross-project portfolio views.
README
@mtuska/hacknplan-mcp
A Model Context Protocol (MCP) server for the HacknPlan project-management API, built for use with Claude. It exposes HacknPlan projects, boards, work items, sub-tasks, dependencies, time logs, the design model, and metrics as MCP tools — plus a cross-project portfolio dashboard and a deadline schedule view that HacknPlan has no native equivalent for.
Pure TypeScript/Node — runs over npx, no Python or other runtime required.
Quick start
npx @mtuska/hacknplan-mcp install
The installer will:
- Ask where to register the server with Claude — global (all your
projects), project (this repo, shared via
.mcp.json), or local (this repo, private to you). - Prompt for your HacknPlan API key (input is hidden) and store it in the chosen Claude MCP config.
- Install the bundled
hacknplanClaude skill into the matching skills directory (global →~/.claude/skills, project/local →./.claude/skills).
Then ask Claude to run the hacknplan_whoami tool to confirm the key works.
(Restart Claude / reload the window first if it was already running.)
The bundled skill (auto-updating)
install drops a hacknplan skill that teaches Claude how to drive these tools
(id-resolution, feature breakdown, destructive-confirm rules, the portfolio /
schedule / standup views). It's part of the npm package, so on every later
server launch — which Claude runs via npx -y @mtuska/hacknplan-mcp (always the
latest) — the server re-syncs the skill if the package version changed,
keeping it current with no action from you.
The refresh only touches a skill copy this package installed (tracked by a
.installed-by.json marker), so your own edits to other skills are never
affected. Pass --skip-skill to register the MCP server without the skill.
Updating from a version before the skill existed (≤ 0.2.0): the auto-sync is update-only — it refreshes a skill it previously installed, but does not create one. If you installed the MCP server before the skill shipped (v0.3.0), re-run
installonce to add it:npx -y @mtuska/hacknplan-mcp install --globalAfter that, every later upgrade auto-updates it on the next server launch.
Getting an API key
HacknPlan → your avatar → Settings → API → Create. Tick the scopes you need.
Install options
npx @mtuska/hacknplan-mcp install [options]
-g, --global Register for all projects (private, in ~/.claude.json). Default.
-p, --project Register for this repo only, shared via ./.mcp.json.
--local Register for this repo only, private to you.
--api-key KEY Provide the key non-interactively (else prompted / $HACKNPLAN_API_KEY).
--name NAME Server name to register (default: "hacknplan").
--skip-skill Register the MCP server only; don't install the Claude skill.
-y, --yes Non-interactive; use defaults and fail if the key is missing.
Non-interactive example (e.g. CI / dotfiles):
HACKNPLAN_API_KEY=hp_xxx npx @mtuska/hacknplan-mcp install --global -y
Project scope & secrets:
--projectwrites the key into./.mcp.json, which is typically committed. For shared repos prefer--global, or add.mcp.jsonto.gitignore.
The installer uses the official claude mcp add-json command when the Claude
Code CLI is available, and otherwise writes the MCP config file directly.
Manual configuration
If you'd rather not use the installer, add this to your Claude MCP config
(~/.claude.json for global, or a project .mcp.json):
{
"mcpServers": {
"hacknplan": {
"command": "npx",
"args": ["-y", "@mtuska/hacknplan-mcp"],
"env": { "HACKNPLAN_API_KEY": "hp_your_key_here" }
}
}
}
Environment variables
| Variable | Required | Purpose |
|---|---|---|
HACKNPLAN_API_KEY |
yes | Authorization: ApiKey <key> for the HacknPlan API. |
HACKNPLAN_GROUPS |
no | JSON map of group label → project names, for the portfolio grouping. |
HACKNPLAN_GROUPS='{"Products":["Website","Mobile App"],"Ops":["Infra"]}'
Tools
75 tools across the HacknPlan surface. Highlights:
- Read/introspect —
hacknplan_whoami,list_projects,get_project,list_work_items(rich server-side filters),get_work_item,list_stages/categories/tags/boards/milestones/importance_levels. - Search & focus —
find_work_items(free-text + facet search to resolve a task by name into its id) andmy_work(what's assigned to you, per-project or across all). - Work items —
create_work_item(with story→taskparent_id, assignees, dependencies, checklists),update_work_item(re-stage / re-prioritize / recategorize / re-estimate / reassign / move board·milestone),plan_feature(create a user story + child tasks + checklists in one call),delete_work_item,add_comment, sub-tasks, tags, user assignment. - Planning —
create_project/update_project/delete_project, stages, categories, tags, importance levels, boards (create_board/update_board/close_board/reopen_board), milestones (create_milestone/update_milestone/delete_milestone), dependencies. - Lifecycle — non-destructive archive/release via
close_project/reopen_projectandclose_milestone/reopen_milestone. - Activity & metrics —
recent_activity(the "what changed this week" / standup feed),log_work,list_work_logs,get_project_metrics,get_milestone_metrics,get_board_metrics(sprint burndown). - Design model —
*_design_element/*_design_element_type, plusget_design_element_metricsand design-element comments (list/add_design_element_comment) for the feature/knowledge tree. - Portfolio (cross-project) —
portfolio_overview(completion %, urgent / blocked / due-soon / overdue across all projects) andschedule_overview(a deadline countdown bucketed by horizon).
Destructive tools (delete_*, remove_dependency) require confirm=true.
Development
npm install
npm run build # tsc -> dist/
npm run typecheck
npm run dev # run the CLI from source via tsx
node dist/cli.js serve # run the MCP server over stdio
Publishing
Releases are tag-driven via
.github/workflows/release.yml. Pushing a
vX.Y.Z tag validates that it matches package.json, publishes to npm, and
creates the GitHub release:
npm version patch # bumps package.json (CLI + server read it at runtime)
git push --follow-tags
Publishing uses npm Trusted Publishing (OIDC) with provenance — no
NPM_TOKEN secret. One-time maintainer setup:
- Publish
v0.1.0once manually (npm publish --access public) to claim the@mtuska/hacknplan-mcpname. - On npmjs.com → the package → Trusted Publishers, add this repo
(
mtuska/hacknplan-mcp) and the workflow filename (release.yml).
After that, every tag-driven release publishes with provenance and no
long-lived token. Users can verify with npm audit signatures.
License
MIT © Montana Tuska
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.