What is James-RAG-Evol CLAUDE.md?

A session briefing for JAMES, a local-first system for auditable knowledge reasoning. It records the project's current version, direction, and recent work.

Use it before working on JAMES to understand its current roadmap, releases, audit benchmarks, and where to look for more detail.

It quickly gives a new coding-agent session the project context and important rules without relying on earlier chat history.

James-RAG-Evol CLAUDE.md — Instructions file for coding agents by Hashevolution: A session briefing for JAMES, a local-first system for…

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 instructions/hashevolution/james-rag-evol/claude-md

Clone the repo

git clone --depth 1 https://github.com/Hashevolution/James-RAG-Evol

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/hashevolution/james-rag-evol/claude-md.svg)](https://agentmods.dev/instructions/hashevolution/james-rag-evol/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/hashevolution/james-rag-evol/claude-md"><img src="https://agentmods.dev/badge/instructions/hashevolution/james-rag-evol/claude-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 9,410 This file is loaded in full into every session.

When invoked 9,410 The same file — it is already loaded in full.

Security scan A 1 finding. 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.

Model	Per session	Once invoked
Fable 5	$0.09410	$0.09410
Opus 5	$0.04705	$0.04705
Sonnet 5	$0.01882	$0.01882
Haiku 4.5	$0.00941	$0.00941

Measured today against content hash f6e5589505a7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

James-RAG-Evol CLAUDE.md scanned grade A with 1 finding 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 today.

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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

테스트는 operator 가 띄운 서버에 curl 로 붙거나 단발 foreground 실행으로

CLAUDE.md · 305 lines

How it starts

The opening of the file, as written. The whole thing — 305 lines — stays where its author put it; the contents beside it link to each section on GitHub.

SEKOS / JAMES — Session Briefing for Claude Code

Read this first. This file orients any new Claude Code session to the project's current direction in under 60 seconds.

What JAMES is

A local-first, auditable knowledge reasoning system. See docs/ARCHITECTURE.md for full design principles and non-goals.

Where we are right now

단일 진실원 (single source of truth): docs/handovers/v0.6.2-restart-roadmap-2026-09-03.md. 아래는 그 요약입니다. 충돌하면 로드맵 문서가 우선입니다.

최신 공식 릴리스: v0.4.4 — DOI 10.5281/zenodo.20652679. v0.5 는 main 에서 closed (2026-06-13) 이지만 DOI 미발행.
사이클 상태: v0.5 closed / v0.6 정식 미진입. v0.5 → v0.6 게이트 = Dim F (외부 고객 6 개월 이상 파일럿) 미통과. 2-fork 계약이 유효 — Fork A LOI 체결 → Track D 버티컬 팩 / Fork B 6 개월 무LOI → 전략 재평가. 판정 시점 ≈ 2026-12-13. 둘 다 operator 결정이며, 결정 전에도 아래 Phase 1–5 는 전부 진행 가능합니다.
실제로 진행된 것: v0.6 / v0.6.1 제품 하드닝 스트림 — PR #886–#1078 (약 190 PR, 2026-06-13 → 2026-06-26, 릴리스 태그 없음). 주요 산출물: 운영 하드닝 P1–P4 (신뢰 프록시 / HTTPS 가이드 / 테넌트 미들웨어 / 온보딩·롤백·추론 시각화·용어집) · 템플릿(양식) 엔진 · 에이전트 트랙 (core/agent_tools + tool-use 루프 + run_shell 기본 OFF) · LLM 라우팅 통합 + 모드별 측정 라우팅 · 채팅 UX 전면 개편 · UI 8→5 페이지 통합 + 비주얼 회귀 하네스 · CSP 인라인 스타일 이관 · 이미지-OCR/비전 체인 수리 · heartbeat 스트리밍 · detailed 답변 스타일.
🔴 lifecycle live-consistency arc (#1018–#1027, #1033–#1034): 프로브(#1020)가 라이브 그래프 탐색이 lifecycle status 를 무시함을 측정으로 증명 (캐스케이드가 라이브 계층에서 사실상 무효였음) → relation_is_live() 게이트를 탐색 / 그래프 스코어 / T1 유효 윈도우 / 3D 스냅샷 전 표면에 적용, 시간여행 경로 격리 검증, 백로그 재측정으로 회귀 없음 확인 (#1028–#1032). 이것이 core/graph traversal 0-라인 streak 의 유일한, 측정 근거로 승인된 예외입니다. 이전 핸드오버의 "0 라인 streak 유지" 문장을 그대로 복사하지 마세요.
🟠 main CI 는 아직 빨간불 — 단 규모가 크게 줄었습니다 (2026-09-03 재측정): .github/workflows/test.yml (pytest) 은 최신 실행(#1080, 08-28)까지 failure 이지만, #1080 이 원인별로 정리해 CI 실패 5 건 가 됐습니다. CI 로그 실측 (df55e21): 5 failed, 4368 passed, 6 skipped — ① test_lrb_v021_cross_model S2 재현 (기대 0.7125 vs 실행 0.6875, README 는 0.688 로 문서화 — 발표 벤치 수치에 직결, 상수를 실행값에 맞춰 고치지 말 것), ②③④ test_measurement_critical_surfaces FixtureLock (gitignore 된 픽스처를 요구 → CI 에서 통과 불가, 구조적), ⑤ test_mobile_responsive (mobile.css !important 29 > 25). ruff / bandit 게이트는 초록입니다. → 재개 첫 작업 = 로드맵 Phase 2 (CI 그린 복구). 그 전에 새 기능 금지. ⚠️ 이전 핸드오버의 "60 failures / rule #5 2건 위반"은 낡은 수치입니다.
유지보수 4 PR (#1077 #1078 08-19 / #1079 08-26 / #1080 08-28): v0.3.3 DOI 계보 정정, ruff F-class 해소, Ali 엔지니어링 4건 ①②③ 발송 + uuid7 production 결함 수리 (start_trace() 가 Python 3.14 전용 uuid.uuid7() 호출 → 지원 인터프리터 전부에서 /query/ 엣지 다운), 아랍어 파이프라인 능력 감사, 그리고 CI 원인별 정리 (_IncludedRouter 래퍼가 137 라우트를 숨긴 문제 + rule #5 2건).
유휴 구간: 마지막 기능 세션 2026-06-26. 이후 착륙한 것은 전부 문서·인용·CI 유지보수입니다.
전략 프레임: 여전히 mother platform. v1.0 이전 버티컬 분화 금지 (rule #1). v0.5 후보 도메인 = enterprise internal knowledge ontology (horizontal) — 이는 선택 기준이지 구현 승인이 아닙니다.
브랜드 규약: SEKOS = 제품 브랜드 (UI / 대외 / 마케팅), JAMES = 엔진 코드명 (소스, JAMES_* env, --sut james, RAB/LRB, Zenodo DOI). PR #934 에서 확정 — 재현성 때문에 코드/논문 쪽은 JAMES 유지.

Read the full file on GitHub · 305 lines

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.

today Changed · +33 lines · +498 tokens per session f6e5589505a7
4d ago First seen · 272 lines · 8,912 tokens per session scan A d30a2df681c7

Subscribe to this mod's changes

James-RAG-Evol CLAUDE.md is an instructions file published in the GitHub repository Hashevolution/James-RAG-Evol (21 stars, last pushed yesterday), licensed MIT. It adds 9,410 tokens to every session, about $0.0471 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Data and AI ai-agent fastapi graph-rag knowledge-graph local-ai local-llm

James-RAG-Evol CLAUDE.md

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Makes network callslowCapability

How it starts

SEKOS / JAMES — Session Briefing for Claude Code

What JAMES is

Where we are right now

What this file has done since we first saw it

Other instructions, from other repositories

neurostack CLAUDE.md

dotty-stackchan CLAUDE.md

robosystems CLAUDE.md

Rapid-MLX AGENTS.md

ILLIP AGENTS.md

WorldOfTaxonomy CLAUDE.md