Code relationship graph and temporal intelligence via abyss CLI. Provides caller tracing, impact analysis, hotspot detection, and file-level context gathering. Agent automatically runs abyss before modifying code to check impact. Works with any agent that has shell access.
npx skills add https://github.com/telagod/code-abyss --skill indexing-code
> 不是搜索工具,是代码的调用图 + 时间智能。通过 shell 调用 abyss CLI。
进入项目时,检查并初始化索引:
command -v abyss >/dev/null && [ ! -f .code-abyss/index.db ] && abyss index
abyss context即将修改任何代码文件时,先获取该文件的完整上下文(不需要知道函数名):
abyss context <要修改的文件路径> --json
返回:该文件所有函数的外部调用方、依赖的类型和函数、热点评分、耦合文件。
根据返回结果:
abyss impact <func> --json 深入分析| 场景 | 命令 |
|------|------|
| 初识项目架构 | abyss map --json |
| 追查 bug 来源 | abyss history <file> --symbol <func> --json |
| 搜索代码(比 grep 好) | abyss search "关键词" --json |
--json 输出结构化 JSON,agent 直接解析。不加 --json 输出人类可读文本。
{
"symbols_with_external_callers": [
{ "symbol": "SetError",
"external_callers": [
{ "file": "handler.go", "line": 42, "caller": "HandleRequest",
"confidence": 0.95, "is_test": false }
],
"possible_callers": []
}
],
"dependencies": [{ "name": "Account", "file": "types.go", "kind": "type_ref" }],
"hotspot": { "score": 5200, "changes_30d": 12, "complexity": 433 },
"coupled_files": [{ "file": "gateway.go", "co_changes": 13, "coupling": "65%" }]
}
external_callers:confidence ≥ 0.7 的可信调用方,按解析档位标注(1.0 同文件 / 0.95 同包 / 0.9 import 限定 / 0.8 全局唯一)possible_callers:confidence < 0.7 的歧义匹配——参考线索,不是事实,勿据此改调用方{
"direct_callers": 17,
"transitive_callers": 521,
"uncovered_paths": ["handler.go:DoSomething"],
"risk_score": 8.5,
"risk_factors": ["high blast radius", "319 paths without test coverage"]
}
abyss index # 建索引(~5s)
abyss context <file> [--json] # 文件完整上下文(改代码前用这个)
abyss callers <symbol> [--json] # 谁调了这个函数(默认隐藏 confidence < 0.7)
abyss callers <symbol> --min-confidence 0 # 连歧义匹配一起看
abyss impact <symbol> [--json] # 改了会影响什么(低置信边被排除时会在 risk_factors 标注)
abyss hook pre-edit # agent hook:stdin 读 tool JSON,增量刷新索引后输出警告
abyss hook post-edit # agent hook:编辑后增量刷新索引
abyss search "query" [--json] # 搜索代码
abyss map [--json] # 项目热点+耦合
abyss history <file> [--symbol X] # 变更历史
abyss stats # 索引统计
Hook 注入按 host 分两条路径(hybrid 切割架构,2026-06-25 锁定):
claude / codex / gemini — 由 abyss CLI 的 attach 子命令负责(production 主入口,abyss v0.5.20+):
abyss attach claude # → ~/.claude/settings.json
abyss attach codex # → ~/.codex/config.toml(Codex 0.125+ 数组表)
abyss attach gemini # → ~/.gemini/settings.json(SessionStart/BeforeTool/AfterTool)
abyss attach all # 三平台一次完成
幂等,重跑覆盖旧 shape,不污染其它键。
二进制查找顺序:PATH → ~/.code-abyss/bin/abyss。两处都没有时 hook 静默停用,后装 abyss 即生效,无需重装。
> v4.8.x → v4.9 deprecation 期变更
> - --with-hooks 对 claude/codex/gemini 仍写入 hook,但 install.js 打印 warning 引导改用 abyss attach <host>;v5.0 移除该路径(openclaw/pi/hermes 的 --with-hooks 永久保留并改造为 spawn install-hooks.sh)
> - --with-abyss 已移除(Agent OS v5.1)。请用 curl -fsSL https://raw.githubusercontent.com/telagod/abyss/main/install.sh | bash;claude/codex/gemini graph hooks 用 abyss attach <host>
> - --with-mcp 注册 abyss MCP 进入 deprecation,引导用户改用 abyss mcp 客户端自配;v5.0 移除
openclaw / pi / hermes — 由 code-abyss npm 包负责(abyss CLI 不接管这三平台,见 abyss src/attach/mod.rs 注释):
npx code-abyss --target openclaw --with-hooks # 自动 spawn install-hooks.sh
npx code-abyss --target pi --with-hooks
npx code-abyss --target hermes --with-hooks
或脱离安装器直接跑脚本:
bash skills/indexing-code/hooks/common/install-hooks.sh auto # 自动检测平台
自动检测平台并注入对应 hook 配置(幂等,JSON 合并用 node):
| 平台 | Hook 事件 | 配置位置 |
|------|----------|---------|
| Claude Code | PreToolUse(Edit\|Write) | .claude/settings.json |
| Codex CLI | PreToolUse(Bash\|shell\|apply_patch\|Edit\|Write) | ~/.codex/config.toml |
| Gemini CLI | BeforeTool(write_file\|replace) | ~/.gemini/settings.json |
| Pi Agent | tool_call(edit_file\|write_file) | ~/.pi/agent/settings.json |
| Hermes | pre_tool_call | ~/.hermes/config.yaml 或 plugin |
| OpenClaw | before_tool_call | plugin api.on() |
统一入口是 abyss hook pre-edit(shell 脚本只是带存在性守卫的薄壳):自动识别各平台 stdin JSON 形状、增量刷新索引(警告反映文件当前状态而非旧索引)、输出生产调用方 / 歧义引用 / 热点警告。编辑后可挂 abyss hook post-edit 保持索引新鲜。agent 无需手动调用。
Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).
Automatically creates user-facing changelogs from git commits by analyzing commit history, categorizing changes, and transforming technical commits into clear, customer-friendly release notes. Turns hours of manual changelog writing into minutes of automated generation.
Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR, or cleanup
Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).
React Native and Expo best practices for building performant mobile apps. Use when building React Native components, optimizing list performance, implementing animations, or working with native modules. Triggers on tasks involving React Native, Expo, mobile performance, or native platform APIs.
React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.
Next.js best practices - file conventions, RSC boundaries, data patterns, async APIs, metadata, error handling, route handlers, image/font optimization, bundling
Use when starting feature work that needs isolation from current workspace or before executing implementation plans - creates isolated git worktrees with smart directory selection and safety verification
Take telagod/indexing-code from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.
The instructions reference npx.
Without those the skill loads but fails at the first command.