HackerReposA FIELD GUIDE TO OPEN SOURCE
ProSponsor the shelf♧ 0
← Back to top all-time
Coral-Bricks-AI
reef
Python2026–2026
Coral-Bricks-AI / reef

reef

A harness framework for building domain-specific agents, the worked finance instance that beats every frontier model on public benchmarks, and the autonomous loop coordinator that drives 100-experiment optimization runs unattended.

2Distinct people
2Mentions
0Explicit recommenders
56GitHub stars
REPOSITORY AT A GLANCE

What to know before exploring

What it is

A harness framework for building domain-specific agents, the worked finance instance that beats every frontier model on public benchmarks, and the autonomous loop coordinator that drives 100-experiment optimization runs unattended.

GitHub

Where it is used

No documented product use has been verified yet.

What people are saying

No recommendation phrase was found in the indexed comments.

No classified warning is attached to this page. Read the discussions before treating silence as approval.

Also in these discussions

No other repository is repeatedly linked in the same discussions.

This week on Hacker News

Nov 7–Nov 14 UTC · Direct repository links and separately labeled related project stories. Indexed through 11/14/2026, 5:04:57 PM UTC.

No direct repository links were found in the indexed HN discussions over the past seven days. Older discussions are available below.

Mentions over time

HN mentions by hour2 dated mentions

Observed mentions: 2026-05-27 17:25:53 UTC to 2026-05-28 10:17:38 UTC.

012026-05-27 17:00:00 UTC to 2026-05-27 18:00:00 UTC (exclusive): 1 mentions; 1 submissions, 0 commentsMay 27, 17:002026-05-27 18:00:00 UTC to 2026-05-27 19:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-27 19:00:00 UTC to 2026-05-27 20:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-27 20:00:00 UTC to 2026-05-27 21:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-27 21:00:00 UTC to 2026-05-27 22:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-27 22:00:00 UTC to 2026-05-27 23:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-27 23:00:00 UTC to 2026-05-28 00:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 00:00:00 UTC to 2026-05-28 01:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 01:00:00 UTC to 2026-05-28 02:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 commentsMay 28, 01:002026-05-28 02:00:00 UTC to 2026-05-28 03:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 03:00:00 UTC to 2026-05-28 04:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 04:00:00 UTC to 2026-05-28 05:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 05:00:00 UTC to 2026-05-28 06:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 06:00:00 UTC to 2026-05-28 07:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 07:00:00 UTC to 2026-05-28 08:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 08:00:00 UTC to 2026-05-28 09:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 09:00:00 UTC to 2026-05-28 10:00:00 UTC (exclusive): 0 mentions; 0 submissions, 0 comments2026-05-28 10:00:00 UTC to 2026-05-28 11:00:00 UTC (exclusive): 1 mentions; 0 submissions, 1 commentsMay 28, 10:00

Counts come from the repository’s dated source evidence. Index cutoff: 2026-11-02 08:16:59 UTC.

View mention counts and date ranges
UTC ranges include their start and exclude their end. Select a period to filter the source evidence.
PeriodStart (UTC)End (exclusive, UTC)MentionsSubmissionsComments
May 27, 17:002026-05-27 17:00:00 UTC2026-05-27 18:00:00 UTC110
May 27, 18:002026-05-27 18:00:00 UTC2026-05-27 19:00:00 UTC000
May 27, 19:002026-05-27 19:00:00 UTC2026-05-27 20:00:00 UTC000
May 27, 20:002026-05-27 20:00:00 UTC2026-05-27 21:00:00 UTC000
May 27, 21:002026-05-27 21:00:00 UTC2026-05-27 22:00:00 UTC000
May 27, 22:002026-05-27 22:00:00 UTC2026-05-27 23:00:00 UTC000
May 27, 23:002026-05-27 23:00:00 UTC2026-05-28 00:00:00 UTC000
May 28, 00:002026-05-28 00:00:00 UTC2026-05-28 01:00:00 UTC000
May 28, 01:002026-05-28 01:00:00 UTC2026-05-28 02:00:00 UTC000
May 28, 02:002026-05-28 02:00:00 UTC2026-05-28 03:00:00 UTC000
May 28, 03:002026-05-28 03:00:00 UTC2026-05-28 04:00:00 UTC000
May 28, 04:002026-05-28 04:00:00 UTC2026-05-28 05:00:00 UTC000
May 28, 05:002026-05-28 05:00:00 UTC2026-05-28 06:00:00 UTC000
May 28, 06:002026-05-28 06:00:00 UTC2026-05-28 07:00:00 UTC000
May 28, 07:002026-05-28 07:00:00 UTC2026-05-28 08:00:00 UTC000
May 28, 08:002026-05-28 08:00:00 UTC2026-05-28 09:00:00 UTC000
May 28, 09:002026-05-28 09:00:00 UTC2026-05-28 10:00:00 UTC000
May 28, 10:002026-05-28 10:00:00 UTC2026-05-28 11:00:00 UTC101

What Hacker News says

2 sources
TL;DR = ~$22,720 total compute @ Opus 4.7 if no caching = $113,418 (5.9x more) - this is just one month on one server... I have 3 more servers like this where I work all time! // Generated wit Claude Ran this on my own ~/.claude/projects/ (933 sessions, 93,842 model calls, mix of main thread + spawned subagents). Numbers came out very close to yours in shape, different in scale. cost.py (Opus 4.7 list rates, main thread only): cache reads (re-reading context) 21.69B tok $10,843 56% cache writes (1h) 678M tok $6,781 35% output (incl. reasoning) 63.5M tok $1,589 8% fresh uncached input 1.6M tok $8 0% TOTAL 22.43B tok $19,221 if no caching: $113,418 (5.9x more) input:output ratio: 353:1 cache hit rate: 97.0% token_time_br
Read the rest
eakdown.py (179M unique tokens, 166h wall clock): reasoning (hidden thinking) 29% of tokens, 102h (61%) of time bash 1.4% 23h (14%) writing tool calls 4% 14h (8%) summaries 2.5% 9h (5%) reading/searching/web 1.6% 7h (4%) subagents 0.2% 6h (4%) editing 0.1% 5h (3%) pasted attachments 25.3% - typed prompt 34.4% - system+tools 1.4% - reread_breakdown.py (per-activity share of billed input): reasoning 59.5% (~$11.4k of the bill is re-sending old hidden thinking back to the model) attachments 22.6% tool calls 7.8% bash 3.0% reading/web 2.4% my prompt 1.6% summaries 1.5% system+tools 0.8% subagents 0.4% main_vs_sidecar.py: main sidecar combined sessions/agents 449 484 933 assistant calls 63,820 30,022 93,842 cache hit 97.0% 94.4% 96.6% turns/agent 142 (median 20, max 11,058 in one session) 62 (median 44) reasoning % of out 82% 51% 77% cost @ Opus 4.7 $19,225 $3,495 $22,720 sidecar = 32% of calls but only 15% of cost. Subagents are doing their job (cheap, focused, short context). Same shape as yours: re-read dominates, reasoning is the biggest re-read line, caching is the only thing keeping it sane. The one that surprised me was a single 11,058-turn main session - some autonomous loop I forgot to kill. Going to grep for that. Repo: github.com/Coral-Bricks-AI/coral-ai/tree/main/claude-code-token-xray
In Show HN: Claude Code's $200 plan is a 17× subsidy on the raw API

MAKE YOURSELF AT HOME

A few shortcuts

/
Search the collection
?
Open this guide
Esc
Close dialogs and navigation

Your saved repositories and interests stay in this browser.