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.
npx agentmods add instructions/codefromkarl/contextatlas/agents-mdgit clone --depth 1 https://github.com/codefromkarl/ContextAtlasWhat 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 | $0.01144 | $0.01144 |
| Opus 5 | $0.00572 | $0.00572 |
| Sonnet 5 | $0.00229 | $0.00229 |
| Haiku 4.5 | $0.00114 | $0.00114 |
Grade A, and why
ContextAtlas AGENTS.md 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.
How it starts
The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
This file provides guidance to AI coding agents when working with code in this repository.
Project Overview
ContextAtlas 是一个为 AI 代码助手设计的语义检索引擎，采用混合搜索（向量 + 词法）、智能上下文扩展和 Token 感知打包策略。
Development Commands
# Build
pnpm build # 编译 TypeScript (tsup)
# Development
pnpm dev # Watch 模式开发
# Run locally
pnpm start # 运行编译后的 CLI
node dist/index.js # 直接运行
# CLI usage
contextatlas init # 初始化配置文件 (~/.contextatlas/.env)
contextatlas index [path] # 索引代码库
contextatlas search # 本地搜索
contextatlas mcp # 启动 MCP 服务端
Architecture
Core Pipeline
索引: Crawler → Processor → SemanticSplitter → Indexer → VectorStore/SQLite
搜索: Query → Vector+FTS Recall → RRF Fusion → Rerank → GraphExpander → ContextPacker
Architecture Layers
CLI commands → application layer → memory/search domain → storage
MCP adapters → (Zod schema only) → (business logic) → (SQLite/LanceDB)
MCP tools (src/mcp/tools/) are thin adapters: Zod schema validation + protocol formatting.
Business logic lives in src/application/ (memory, retrieval) and src/memory/ (domain).
CLI commands (src/cli/commands/) call application layer directly, bypassing MCP adapters.
Key Modules
| Module | Location | Responsibility |
|---|---|---|
| Application Layer | src/application/memory/ |
Memory 工具业务逻辑，CLI 和 MCP adapter 统一入口 |
| Retrieval Application | src/application/retrieval/ |
检索用例编排，结果卡片渲染 |
| SearchService | src/search/SearchService.ts |
搜索总编排层，串联 recall、rerank、cutoff、expand、pack |
| HybridRecallEngine | src/search/HybridRecallEngine.ts |
混合召回执行层，负责向量/词法召回、FTS 降级与 RRF 融合 |
| RerankPolicy | src/search/RerankPolicy.ts |
rerank 池选择与 Smart TopK cutoff 策略 |
| SnippetExtractor | src/search/SnippetExtractor.ts |
rerank 文本构造与命中片段截取 |
| MemoryStore | src/memory/MemoryStore.ts |
记忆 facade，协调 feature/project-meta/decision/long-term 四类子存储 |
| MemoryStoreBootstrap | src/memory/MemoryStoreBootstrap.ts |
项目注册、只读/可写初始化与兼容导入引导 |
| ProjectMetaStore | src/memory/ProjectMetaStore.ts |
checkpoint、profile、catalog、global memory 持久化 |
| FeatureMemoryRepository | src/memory/FeatureMemoryRepository.ts |
feature memory 的 SQLite CRUD |
| FeatureMemoryCatalogCoordinator | src/memory/FeatureMemoryCatalogCoordinator.ts |
feature memory 与 MemoryRouter catalog 的同步 |
| DecisionStore | src/memory/DecisionStore.ts |
架构决策记录的持久化与映射 |
| LongTermMemoryService | src/memory/LongTermMemoryService.ts |
长期记忆追加、检索、合并、状态与清理 |
| GraphExpander | src/search/GraphExpander.ts |
三阶段上下文扩展 (E1 邻居/E2 面包屑/E3 导入) |
| ContextPacker | src/search/ContextPacker.ts |
段落合并和 Token 预算控制 |
| SemanticSplitter | src/chunking/SemanticSplitter.ts |
AST 语义分片器 (Tree-sitter) |
| VectorStore | src/vectorStore/index.ts |
LanceDB 适配层 |
| Database | src/db/index.ts |
SQLite + FTS5 元数据和全文索引 |
| MCP Server | src/mcp/server.ts |
Model Context Protocol 服务端实现 |
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.
- 2d ago First seen · 90 lines · 1,144 tokens per session scan A b17ca16867d1
ContextAtlas AGENTS.md is an instructions file published in the GitHub repository codefromkarl/ContextAtlas (29 stars, last pushed 3mo ago), licensed MIT. It adds 1,144 tokens to every session, about $0.0057 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.
Other instructions, from other repositories
leantoken AGENTS.md
Instructions for morluto/leantoken, covering repository guidance, architecture, development, change-specific validation and contributions.
open-codebase-index AGENTS.md
Instructions for Helweg/open-codebase-index, covering agents.md - ai agent guidelines for open-codebase-index, build, test, and lint, run a single test, equivalent low-level command and architecture and file structure.
code-brain CLAUDE.md
Instructions for ezBuilder/code-brain, covering code brain agent contract, response, branches, loop and planning.
codebase-index AGENTS.md
Instructions for denfry/codebase-index, covering codebase-index, versioning policy and fork and pull request workflow.
Repo-Interrogator AGENTS.md
AGENTS.md instructions for taggedzi/Repo-Interrogator, covering agents.md, agent instructions, python environment, 0) decision hierarchy (important) and 0.5) architectural decision records (adrs).
Repo-Interrogator CLAUDE.md
Claude Code instructions for taggedzi/Repo-Interrogator, covering claude.md, project overview, environment setup, developer install and commands.