coco-search CLAUDE.md

coco-search CLAUDE.md is an instructions file for coding agents from VioletCranberry/coco-search. It costs 8,169 tokens per session, scanned A, original, MIT.

Project instructions for CocoSearch, a local code-search tool that indexes source code and lets agents search it through a command line, server, or interactive prompt.

In plain words
What is it for?
Use them when developing CocoSearch, especially for searching code, finding symbols such as functions or classes, and tracing dependencies.
Why use it?
They tell the agent which search method to use and provide the project’s setup requirements, reducing inconsistent or inefficient code searches.

Instructions file

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/violetcranberry/coco-search/claude-md
Clone the repo
git clone --depth 1 https://github.com/VioletCranberry/coco-search

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.

agentmods badge for coco-search CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/violetcranberry/coco-search/claude-md.svg)](https://agentmods.dev/instructions/violetcranberry/coco-search/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/violetcranberry/coco-search/claude-md"><img src="https://agentmods.dev/badge/instructions/violetcranberry/coco-search/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 8,169 This file is loaded in full into every session.
When invoked 8,169 The same file — it is already loaded in full.
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 5.1 $0.08169 $0.08169
Opus 5 $0.04084 $0.04084
Sonnet 5 $0.01634 $0.01634
Haiku 4.5 $0.00817 $0.00817

Measured 5d ago against content hash c239020797d4, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

coco-search CLAUDE.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 5d 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.md · 280 lines

How it starts

The opening of the file, as written. The whole thing — 280 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Project Instructions

This file provides guidance to AI coding agents (Claude Code, OpenCode, and others) when working with code in this repository. It is the single source of truth — AGENTS.md is a symlink to this file.

Project Overview

CocoSearch is a local-first hybrid semantic code search tool powered by CocoIndex and Tree-sitter. It indexes codebases into PostgreSQL with pgvector embeddings and provides search through CLI, MCP server, or interactive REPL. Local by default with Ollama; optional remote embedding providers (OpenAI, OpenRouter) available for teams that prefer managed infrastructure. Requires Python >=3.11.

Tool Routing (MANDATORY)

When CocoSearch MCP tools are available, ALWAYS use them instead of Grep, Glob, or Task/Explore agents for code search and exploration. These rules are mandatory, not advisory. Violations degrade search quality and create unnecessary permission prompts.

Task Use this NOT this
Code search / "how does X work?" search_code Grep, Glob, Task (Explore)
Symbol lookup / "find function Y" search_code with symbol_name/symbol_type Grep for def/class patterns
Dependency tracing / "what imports X?" get_file_dependencies / get_file_impact Grep for import statements
Batch dependency analysis (multiple files) get_batch_dependencies / get_batch_impact Per-file get_file_dependencies calls
Search debugging / "why no results?" analyze_query Manual pipeline investigation

Fall back to Grep/Glob ONLY for:

  • Exact literal string matches (e.g., a specific error message or config value)
  • File path pattern matching (e.g., "find all *.test.ts files")
  • Editing operations that need line numbers from a known file

Development Setup

# Prerequisites: Docker, uv (Python package manager)
# One-command setup (starts infra, pulls model, installs deps, indexes codebase):
./dev-setup.sh

# Or manually:
docker compose --profile ollama up -d    # PostgreSQL 17 + Ollama
uv sync                                 # Install dependencies
uv run cocosearch index .               # Index the codebase

Read the full file on GitHub · 280 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.

  1. 5d ago First seen · 280 lines · 8,169 tokens per session scan A c239020797d4

Subscribe to this mod's changes

coco-search CLAUDE.md is an instructions file published in the GitHub repository VioletCranberry/coco-search (37 stars, last pushed today), licensed MIT. It adds 8,169 tokens to every session, about $0.0408 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.