One manifest
Runtimes and MCP servers declared together in a single ribosome.json,
instead of two separate, unsynced configs.
ribosome
One manifest declares the language runtimes and the MCP servers a project needs. ribosome resolves them together, deduplicates shared runtimes, and pins everything into one reproducible lockfile — before any workflow runs.
The problem
Section titled “The problem”Standing up a project that uses MCP servers today means doing two unrelated
jobs by hand: installing the language runtimes each server needs, and
configuring each server separately, with no shared version policy, no
deduplication when ten servers all want node@24, and nothing pinned.
Nothing ties the two together — you find out something’s missing when a run
fails halfway through.
What ribosome does
Section titled “What ribosome does”Resolved up front
A missing tool or an unresolvable server fails at validation time – before any workflow runs – with every failure reported at once, not one at a time.
Deduplicated and pinned
Ten servers needing node@24 produce one pool entry, one install. Every
resolution lands in a single reproducible lockfile.
How it works
Section titled “How it works”- Declare your runtimes and MCP servers together in one
ribosome.json. - Resolve — ribosome provisions each runtime, resolves each server against its registry, and deduplicates shared runtimes into a pool. If anything can’t be resolved, it reports every failure at once, before any workflow runs.
- Pin — the whole result lands in one
ribosome.lock.json: a deduplicated runtime pool plus one isolated environment view per server, reproducible on the next run.
npx @medullaflow/ribosome resolveReady to try it? Start with the Quickstart →, or read how ribosome works for the full design.