Token Savior -- v4.3
One MCP server. One profile. 97.9% on tsbench at -80% tokens. Structural code navigation, persistent memory, and Bash output compaction for AI coding agents.
-brightgreen){kind=icon}
mibayy.github.io/token-savior -- project site + benchmark landing github.com/Mibayy/tsbench -- benchmark source + fixtures
Benchmark -- 96 real coding tasks (Claude Opus 4.7, May 2026)
| Plain Claude Code | With Token Savior | |
|---|---|---|
| Score | 141 / 180 (78.3%) | 188 / 192 (97.9%) |
| Active tokens / task | 17 221 | 3 395 (-80%) |
| Wall time / task | 110.6 s | 18.9 s (-83%) |
Reproduces with the optimized profile (single env var). See BENCHMARK-SUMMARY.
What's new
v4.3.0 -- bench-driven coverage push (May 2026)
Real-world bench against 7 days of transcripts (1130 Bash outputs) drove this release. Cumulative savings now sit at ~20.4 K tokens/week (19.3% match rate, 68.9% mean compaction) vs ~12 K/week on v4.2.0.
- Fixed
pytestregex: now matchespython3 -m pytest,uv run pytest, venv-prefixed forms,poetry/hatch/pdm/rye run pytest. - 5 more git compactors:
fetch,checkout,branch,worktree list,stash list. - 4 more gh compactors:
gh repo view,gh pr view,gh issue view,gh pr diff. grep,find,catcompactors. Group hits by file, strip common prefix, head/tail truncation. 83-96% savings on the fixtures.- Compound command splitter:
cd /root/foo && git statusnow compacts by picking the last meaningful segment of&&/;chains. Bails on subshells, heredocs, pipes, loops, unterminated quotes.
