Set up your agent (MCP)
Connect Claude Code, Codex, or any MCP client so your agent can read and write your maps.
Alkahest ships an MCP server in the CLI (alkahest mcp). Register it once and your agent can scan, publish, file and update issues, take notes, and — most importantly — ask you questions when it hits a decision it shouldn't make alone.
Claude Code
npm install -g @cr8rcho/alkahest
claude mcp add alkahest -s project -- alkahest mcp
Codex
Add to ~/.codex/config.toml:
[mcp_servers.alkahest]
command = "alkahest"
args = ["mcp"]
Any other MCP client works the same way — the server command is alkahest mcp.
Authentication
The MCP server needs a publish token for anything that touches the cloud (publishing, issues, notes). Either:
- run
alkahest login --token alk_…once (the MCP server picks up the stored credential), or - set
ALKAHEST_TOKENin the MCP server's environment (useful for per-project configs).
Create tokens at alkahest.app → Publish tokens. Local-only tools (scan, the map readers) work without a token.
What your agent can do
| Area | Tools |
|---|---|
| Code map | scan, publish, overview, get_screen, who_calls, set_summary, set_prd |
| Issues | issues, add_issue, update_issue, link_issues, map_issue, complete_issue |
| Decisions | ask_issue, reply_issue, issue_comments, resolve_issue_question |
| Notes | notes, add_note, get_note, update_note, link_notes, map_note |
| Comments | comments, add_comment, reply_comment, resolve_comment, comment_to_issue |
| Meta | list_projects, maps, create_map, history, check_version |
The issues tool marks each issue as actionable or not — done, blocked-by, and awaiting decision are all derived — so an agent can pick its next task without guessing. When it needs you, it calls ask_issue and the issue stops being actionable until you answer. See Issue maps & decisions.
One-prompt setup
Don't want to do any of this by hand? Paste the Quickstart prompt into your agent — it installs the CLI, registers MCP, scans, and opens the map for you.