skills CLAUDE.md

skills CLAUDE.md is an instructions file for coding agents from mixedbread-ai/skills. It costs 1,225 tokens per session, scanned A, original, Apache-2.0.

A set of repository instructions for Mixedbread skills, which support search, retrieval-augmented generation (RAG), and document parsing. It explains the folder structure, conventions, and how to add and test skills.

In plain words
What is it for?
Use it when building or maintaining Mixedbread skills, search or parsing integrations, repository documentation, and skill tests.
Why use it?
It helps contributors understand how the repository is organized and what each new skill must contain, reducing inconsistent changes and setup mistakes.

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/mixedbread-ai/skills/claude-md
Clone the repo
git clone --depth 1 https://github.com/mixedbread-ai/skills

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 skills CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/mixedbread-ai/skills/claude-md.svg)](https://agentmods.dev/instructions/mixedbread-ai/skills/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/mixedbread-ai/skills/claude-md"><img src="https://agentmods.dev/badge/instructions/mixedbread-ai/skills/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,225 This file is loaded in full into every session.
When invoked 1,225 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 $0.01225 $0.01225
Opus 5 $0.00613 $0.00613
Sonnet 5 $0.00245 $0.00245
Haiku 4.5 $0.00122 $0.00122

Measured 5d ago against content hash 7d97d2c74852, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

skills 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 · 72 lines

How it starts

The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Mixedbread Skills

Agent skills for building search, RAG, and document parsing with Mixedbread.

Repository Structure

  • skills/ — Each subdirectory is a self-contained skill with a SKILL.md
    • skills/mxbai-cli/ — CLI tool usage
    • skills/mixedbread-search/ — Stores API & SDKs
    • skills/mixedbread-parsing/ — Parsing API & OCR
    • skills/mixedbread-search-agent/ — Toast-1 completions API, tool contracts & Stores wiring
    • skills/mixedbread-search-agent-harness/ — Custom search-model harness design
  • .claude-plugin/ — Claude Code plugin configuration
  • .cursor-plugin/ — Cursor plugin configuration
  • .mcp.json — MCP server configuration
  • gemini-extension.json — Gemini CLI extension configuration
  • agents/AGENTS.md — Fallback instructions for Codex/OpenAI agents
  • SKILL_TREE.md — Navigable index of all skills
  • benchmark.sh — Blind A/B comparison of agent output with and without the skills

Conventions

  • Each skill's SKILL.md uses YAML frontmatter with name and description fields
  • The description field determines when the agent activates the skill — write it as a trigger condition
  • Open with docs links and close with a Troubleshooting table. Everything between is up to the skill — name sections after what they actually cover rather than forcing a fixed template
  • Do not group guidance by severity. Where a reference table already carries the constraint, let it; where the constraint is a mistake to avoid, state it as a flat "Don't" list
  • Prefer tables and code over prose. State the fact and the fix; cut the explanation of why unless the agent gets it wrong without it
  • Keep SKILL.md under 500 lines. Past that, move detail into references/ and say in the pointer when to read each file
  • Do not duplicate broad public API reference docs. Bundle focused behavioral references when a skill depends on non-obvious product or harness contracts, and link public docs for the general surface.
  • Keep the search-model guidance backend-agnostic and isolate the Mixedbread wiring in its own reference. The model brings no retrieval of its own, so a skill that only shows Stores-backed tools teaches the wrong contract.
  • Skills install as separate plugins, so a relative path into a sibling skill does not resolve. When two skills need the same reference — tool-contracts.md is the current case — duplicate the file in full rather than summarizing it in one of them, and update both copies together.
  • The served model id is toast-1.
  • SDK-facing examples normally cover both Python and TypeScript. Stores and parsing use the first-party SDKs (mixedbread, @mixedbread/sdk); the search agent uses the official OpenAI clients against Mixedbread's base URL. Harness-internal guidance may stay Python.
  • The two OpenAI clients carry Mixedbread's protocol extensions differently — Python requires extra_body, Node takes them inline with a type cast — so write each language reference from a verified example rather than translating the other.
  • API base URL: https://api.mixedbread.com/
  • API key env var: MXBAI_API_KEY
  • CLI tool: mxbai (installed via npm install -g @mixedbread/cli)

Read the full file on GitHub · 72 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 · 72 lines · 1,225 tokens per session scan A 7d97d2c74852

Subscribe to this mod's changes

skills CLAUDE.md is an instructions file published in the GitHub repository mixedbread-ai/skills (13 stars, last pushed 13d ago), licensed Apache-2.0. It adds 1,225 tokens to every session, about $0.0061 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.