api.kalandra.tech/mcpTalk to this site from your AI assistant.
This site runs its own Model Context Protocol server. Connect it to Claude or any MCP-compatible assistant, and the assistant can act here on your behalf — submit a job offer, follow the conversation about it, browse the blog, and join the comments. Every action goes through the same public API the website itself uses.
Connect
https://api.kalandra.tech/mcpStreamable HTTP, stateless. Reading the blog needs no account; tools that act as you expect your access token in the Authorization header.
claude mcp add --transport http kalandra https://api.kalandra.tech/mcp --header "Authorization: Bearer <token>"Sign in and your access token gets filled into the command above — that's what lets the assistant act as your account.
You're signed in. The copy button above includes your current access token.
Access tokens are short-lived (about an hour) — when the assistant starts getting 401s, grab a fresh one here.
What it can do
Job offers
submit_job_offerAccountSend me a job offer — the same flow as the Hire Me form.
list_my_job_offersAccountYour submitted offers with their current status.
get_job_offer_commentsAccountThe conversation thread on one of your offers.
add_job_offer_commentAccountReply in that thread — I get notified by email.
Blog
list_blog_postsPublicPublished posts with summaries and tags. Signed in, each one shows whether you've already read it.
get_blog_post_commentsPublicThe public comment thread of any post.
post_blog_commentAccountComment on a post, or reply to someone, as you.
Your activity
get_my_commentsAccountEverything you've said here — blog and job-offer comments — with the replies they received.
Try asking
- “Submit a job offer to Pavel: senior .NET engineer at Acme, remote, salary 120–150k.”
- “What has Pavel written on his blog lately?”
- “Did anyone reply to my comments on kalandra.tech?”
How it works
- The tools are a second front door on the API: each call runs the same code the website's own requests do, as your account, with the same validation, rate limits, and email notifications.
- Blog posts are read from the site's RSS feed, so the assistant sees exactly what is published.
- Built in .NET with the official MCP C# SDK, served as a /mcp route on the API itself — no separate service.
The full build story is on the Project page.