cb-proof

A command that compares two versions of code-brain context search and reports ranking, repeatability, storage, latency, and context-size results.

In plain words
What is it for?
Running an A/B-style proof of code-brain context-search effects and reporting the measured checks, including an optional expected path.
Why use it?
It checks whether the newer search actually improves results without adding data or making responses too slow.

Command for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/ezbuilder/code-brain/cb-proof
Clone the repo
git clone --depth 1 https://github.com/ezBuilder/code-brain

Made for: Claude Code.

Per session 30 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 366 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00030 $0.00366
Opus 5 $0.00015 $0.00183
Sonnet 5 $0.00006 $0.00073
Haiku 4.5 $0.00003 $0.00037

Measured 2d ago against content hash 716c71dbe506, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cb-proof scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/commands/cb-proof.md · 28 lines

What it actually says

$ARGUMENTS가 비었으면 .ai/bin/ai context prove --json, 있으면 .ai/bin/ai context prove "$ARGUMENTS" --json 실행. exit 코드 보존.

평문만 출력한다.

코드브레인 효과 검증: {ok ? "통과" : "실패"}
- 쿼리: {query} ({query_selection})
- 효과: {effect.status}
- graph/PPR: {v2.graph_ranking_applied ? `적용 · ${v2.ranked_node_count}노드` : "정책 활성 · 적용 가능한 call edge 없음"}
- 결정론: legacy signature {legacy.signature_count}개 / v2 signature {v2.signature_count}개 / receipt {v2.receipt_count}개
- 무증가: {durability.unchanged ? "통과" : "실패"}
- 지연 p95: legacy {legacy.p95_ms}ms / v2 {v2.p95_ms}ms
- context: {v2.context_bytes}/{v2.max_context_bytes}B

expected.path가 있으면 다음 줄을 추가한다.

- 정답: legacy rank {legacy.path_rank} span {legacy.span_overlap} / v2 rank {v2.path_rank} span {v2.span_overlap}

실패한 checks는 마지막에 - 실패: {check 이름} 형식으로 모두 출력한다. JSON에 없는 값은 추측하지 않는다.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 28 lines · 30 tokens per session scan A 716c71dbe506

Subscribe to this mod's changes

cb-proof is a command published in the GitHub repository ezBuilder/code-brain (11 stars, last pushed 2d ago), licensed Apache-2.0. It adds 30 tokens to every session and 366 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.