mcp-context-cost

memory — context cost

2,378 tokens across 9 tools — light (1–5K). Measured 2026-08-17 under methodology v1.0.

   
server (self-reported) memory-server v0.6.3
status measured
tokenizer tiktoken / o200k_base
launch command npx -y @modelcontextprotocol/server-memory@2026.7.4
isolation host process (no container)
env vars supplied none
canonical SHA-256 d028274f76dc9aa2e622ae02a17ce313aa65d7b5935254ca53889d4094238abb
category official-reference
source https://github.com/modelcontextprotocol/servers

Where the tokens are

tool tokens share description schema
search_nodes 323 13.6% 11 51
open_nodes 322 13.5% 10 51
create_entities 294 12.4% 8 111
create_relations 294 12.4% 17 106
read_graph 291 12.2% 5 24
add_observations 249 10.5% 10 97
delete_relations 225 9.5% 7 115
delete_observations 212 8.9% 9 98
delete_entities 166 7.0% 11 53

Each tool is tokenized on its own, so the parts do not sum exactly to the whole: the array adds its own brackets and commas, and the tokenizer merges tokens across object boundaries. The badge number is always the count of the whole array, never a sum of parts.

Over time

date tokens tools change
2026-08-16 2,378 9
2026-08-17 2,378 9 no change

Full series: results/history.csv.

Re-derive it

npx -y mcp-context-cost verify results/memory/measurement.json

That re-tokenizes the published capture and checks the count and the hash. If it disagrees with the badge, the badge is wrong — open an issue and it gets corrected.

Badge JSON · All servers · Leaderboard · Methodology