You build with AI. Now prove it.
Your best work is invisible.
You spend hours coding with AI, but your best work disappears. Promptbook changes that.
Agent refactored sidebars and updated dark theme styling
frontend
AUTONOMOUS
First session on frontend
First build tracked for this project
844.2ktokens
0prompts
3:33:03time
Claude Codeclaude-sonnet-4-6[1m]$6.32
Rewrote sidebar components with dark theme updates and new tab management features.
Shipped E4 recall depth feature and F4 handoff
memd
FULL SEND
Top 3% session this week
3315 sessions tracked this week across all builders
1.7Mtokens
2prompts
33:40time
587lines
Claude Codeclaude-opus-4-7[1m]Markdown$32.40
Completed E4 recall depth feature with tests (557 passing, no regressions). Prepared F4 handoff packet documenting next phase prerequisites.
One-shot debugging on secret-broker
secret-broker
ONE-SHOT
Top 14% session this week
3333 sessions tracked this week across all builders
128.2ktokens
1prompts
41:44time
1lines
Codexgpt-5.4-mini$0.17
41m session running commands and testing. 101K tokens across 1 prompts using gpt-5.4-mini.
Built agent working memory with reflection
persiscal-agents
FULL SEND
8.2x your typical session
1.4M tokens vs typical 164.0K
1.4Mtokens
10prompts
6:44:03time
890lines
Claude Codeclaude-opus-4-7TypeScript$22.22
Shipped agent working memory with automatic reflection and conversation compaction.
Agent workflow updates for personal website
personal-website
CLEAN SWEEP
Top 13% session this week
3368 sessions tracked this week across all builders
157.2ktokens
3prompts
7:41time
1,005lines
Codexgpt-5.4$0.86
A short, tool-heavy session updated local agent and memd helper files for the personal website, with most work done through shell automation and a single patch edit.
X
@xexyz42· 6hOne-shot debugging on machine
machine
ONE-SHOT
First session with gpt-5.5
First build using this model
156.9ktokens
1prompts
34:55time
3lines
Codexgpt-5.5Python$0.82
34m session running commands and testing. 156K tokens across 1 prompts using gpt-5.5.
Fixed config parsing by restructuring hooks configuration
josue
ONE-SHOT
Top 15% session this week
3365 sessions tracked this week across all builders
109.1ktokens
1prompts
4:14time
103lines
Claude Codeclaude-opus-4-7[1m]$2.33
Restructured hooks configuration to resolve config parsing errors in codex 0.125.
Started dogfood clock for recall measurement
memd
Top 11% session this week
3364 sessions tracked this week across all builders
229.5ktokens
5prompts
31:32time
262lines
Claude Codeclaude-opus-4-7[1m]Markdown$5.31
Configured the environment, compiled the measurement infrastructure, verified ledger output, and started the tracking clock.
Updated Codex configuration settings with testing
josue
Top 1% build time this week
3376 sessions tracked this week across all builders
40.6ktokens
2prompts
11:17time
Claude Codeclaude-opus-4-7[1m]$0.63
Modified configuration file and tested the changes.
Fixed rules version selection on event creation form
Field Day Commander
Top 22% session this week
3384 sessions tracked this week across all builders
74.5ktokens
2prompts
18:18time
23lines
Claude Codeclaude-opus-4-7PHP$1.33
Fixed event creation form showing year as default rules version when no event type was selected.
Shipped memory continuity compiler with full test coverage
memd
ONE-SHOT
6-week shipping streak
42 consecutive days
2.0Mtokens
1prompts
1:59:39time
862lines
Claude Codeclaude-opus-4-7[1m]JSON$30.78
Landed D4 phase with seven feature commits implementing memory continuity, test fixtures, and 939 passing tests workspace-wide.
Refactored actor system to eliminate runtime object lookups
dirigible2D
FULL SEND
72.4x your typical session
1.6M tokens vs typical 22.2K
1.6Mtokens
19prompts
1:35:02time
84lines
Claude Codeclaude-opus-4-7[1m]C#$32.28
Replaced unsafe runtime object lookups with dependency injection across spawning systems, verified zero compilation errors.