Your agent re-reads the whole file. You pay for every line.
SOTA · Terminal-Bench800+codedb
Your agent re-reads the whole file.
You pay for every line.CodeDB Pro cuts that token bill \~97%.
Free `graff` is already SOTA on Terminal-Bench. CodeDB Pro keeps the exact same eval scores and gets there in a fraction of the tokens. Your agents just stop dumping whole files into context.
curl -fsSL https://codegraff.com/install.sh | bash
copy
Cut your token bill · $99/yrsee the math →
macOS arm64 + Linux x86\_64 · zero runtime deps
\~ codegraffstock file tools
tokens this task
18,400
5 tool calls · 8.4s wall-clock · Claude / Codex native
baseline
$graff -p "add error handling to handleAuth"
read src/auth.zig 1,240 lines 2,103 tok
read src/errors.zig 312 lines 487 tok
grep "handleAuth" whole-file context
edit src/auth.zig rewrites 169 KB
shell cargo test ok · 0.7s
same task, same result · hover to pause
Engineering teams shipping with Codegraff
2.5B
tokens saved · last 30d
145k
ops · last 7d
63µs
per op · codedb p50
47t / 14t
outline / lookup
counters from activated installs · speed numbers from bench/ on github · hyperfine, n=10, macOS arm64
The token ledger
The same task. A fraction of the tokens.
One real file, measured three ways. Default tools dump the raw text. codedb, the free engine, reads it structurally. CodeDB Pro lands the exact symbol and edits in kilobytes. The red bar is what you pay for today.
Understand a 2,300-line file
29× leaner than the raw dump
865 tok
with CodeDB Pro
codedb 6,305 tok25,324 toktoday
Search the whole codebase
codedb’s search engine ships inside Pro
built in
with CodeDB Pro
codedb 2,160 tok5,938 toktoday
Edit one function
Pro only · codedb reads, it doesn’t edit
1.9 KB
with CodeDB Pro
169 KBtoday
measured · codedb-pro 0.2.9 · same 2,318-line file · "built in" = Pro ships codedb's search engine
How CodeDB Pro works
It finds the right file. Then it barely reads it.
01
codedb picks the right file
Most agent failures aren't bad reasoning, they're the agent looking at the wrong file. codedb is graff's retrieval layer: a local index of symbols, outlines, and call paths the agent queries before it acts. SOTA on Terminal-Bench, free, open-source.
02
CodeDB Pro gets there in \~97% fewer tokens
Same eval scores, a fraction of the footprint. outline returns 47 tokens where a raw read pulls 2,103. symbol lookups land the exact function, never the whole file.
03
Then Pro edits, symbol-safe
The one thing the free engine can't do. Symbol-safe patches write 1.9 KB where default tools rewrite 169 KB. Immune to line drift, concurrent agents, and mid-session file growth.
SWE-bench Lite writeup: 4/4 recall, 3/4 top-1 →justrach/codedb →
What CodeDB Pro adds
Compact reads. Symbol edits. Batched ops.
Scope mode context, not noise
ripgrep can't do this
Every match returns its enclosing function. Not a line. Not a file. The exact block your agent needs to act on.
`zigrep -S 'validateInput' src/`
Structural read 47 tokens vs 2,103
outline first
outline gives the full symbol map of a file. Then symbol pulls exactly the function, never the whole file.
`zigread outline src/main.zig`
Symbol-safe patching no line drift
edit by name
Edit by function name, not line numbers. Survives upstream diffs, concurrent agents, and mid-session file growth.
`zigpatch --symbol parseToken src/auth.zig`
Batch ops 10 reads in 7ms
0.7ms per op
Persistent daemon. No fork/exec per call. Reads, searches, diffs, and edits run in parallel over one stdio pipe.
`codedb-pro batch [read, search, diff, edit]`
Benchmark · same query
6.4× faster
zigrep5.2ms
ripgrep33.1ms
hyperfine · 4.7 GB repo · macOS arm64 · n=10
Pricing
Per-seat math. No contact-sales wall.
$99/year is one seat. The ledger above is one real file: 25,324 tokens of raw dump down to 865, edits in kilobytes not hundreds. Every week you wait, that gap is tokens you already pay for, spent re-reading files.
14-day refund · no auto-renew lock-in · key by email, instantly
Individual
$99/year
1 seat · $99/seat · all tools, all editors
Get Pro · 1 seat
most teams
Team
$160/year
2 seats · $80/seat · save $38 vs 2 solo seats
Get Pro · 2 seats
Team Plus
$349/year
5 seats · $69.80/seat · save $146 vs 5 solo seats
Stripe checkout link by email · same-day
Engineering org · 10+ seats
Volume pricing, invoiced annually, MSA on request. One-day turnaround on quotes.
Refund within 14 days · License keys by email · No subscription auto-renew lock-in
Codegraff Gateway/pay-as-you-go · no API keys to manage
One login. Twelve models. No keys.
Run `graff provider login codegraff` and you get instant access to the frontier set. No API keys to manage, no provider accounts to create. Credits start at $10, top up from the dashboard.
Claude Opus 4.8
$5/Mtok in
Claude Sonnet 4.6
$3/Mtok in
DeepSeek V4 Pro
$0.52/Mtok in
GPT-5.5
$5/Mtok in
Grok Build
$1/Mtok in
Kimi K2.7 Code
$1.14/Mtok in
Kimi K2.7 Code Highspeed
$2.28/Mtok in
GLM-5.2
$1.68/Mtok in
MiniMax M3
$0.72/Mtok in
MiMo V2.5 Pro UltraSpeed
$1.305/Mtok in
MiMo V2.5 Pro
$0.435/Mtok in
MiMo V2.5
$0.14/Mtok in
Stop your agents from burning the cap.
CodeDB Pro. One install per engineer. Works with Claude Code, Codex, Cursor, Windsurf. Per-seat licenses, by email.
Get CodeDB Pro · $99/seatTalk to Rach
curl -fsSL https://codegraff.com/install.sh | bash codedb-pro activate CGPRO-XXXX-XXXX engineer@company.com codedb-pro mcp register --client claude-code
Runs locally · your code never leaves the machine · uninstall and your agent falls back to its built-in tools
Built by Rach Pradhan · hi@codegraff.com · Terms · Privacy