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/nirholas/xactions/claude-mdgit clone --depth 1 https://github.com/nirholas/XActionsWhat 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 | $0.00505 | $0.00505 |
| Opus 5 | $0.00253 | $0.00253 |
| Sonnet 5 | $0.00101 | $0.00101 |
| Haiku 4.5 | $0.00051 | $0.00051 |
Grade A, and why
XActions 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 yesterday.
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 — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.
XActions: instructions for Claude
Read AGENTS.md. It is the single source of truth for how to work in this repository, and it is kept current. This file adds only what is specific to Claude; it deliberately does not restate AGENTS.md, because the last time four files described the repository they drifted apart and an agent worked a whole session from the stale one.
AGENTS.md answers, in this order: which lane to use (shell out to the xactions CLI, or
load the MCP server), the CLI lane and its output contract, the MCP lane, a table from
common request to the exact file that answers it, the skills, the directory map and the
three runtime contexts, and the mistakes that have actually cost people time here.
Claude Code specifics
Install the skills so they are available in every session:
xactions skills install --all --global # writes ~/.claude/skills/<id>/
xactions skills install --all # or ./.claude/skills/ in this project
xactions skills list # what is installed, and from where
Add the MCP server:
claude mcp add xactions -- npx -y xactions-mcp
Load only the tools the session needs. The full list is 152 tools and about 60 KB of JSON before you have read a single tweet:
claude mcp add xactions -- npx -y xactions-mcp --tools read,analytics
Hold every write for human approval before you let an agent near a real account:
XACTIONS_MCP_REQUIRE_APPROVAL=1 npx xactions-mcp
Writes then become drafts, released with xactions drafts approve <id> or the
x_approve_draft tool. Independently of that gate, every write is charged against a
rolling 24 hour per-account cap that survives a restart.
Claude Desktop
.mcpb is a bundle the user drags onto Settings > Extensions. It carries the server and
its dependencies and prompts for the session cookie and the tool groups at install time,
so nothing is typed into a config file. Built by node scripts/build-mcpb.mjs and
attached to each release.
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.
- yesterday First seen · 54 lines · 505 tokens per session scan A 4e56073da1fe
XActions CLAUDE.md is an instructions file published in the GitHub repository nirholas/XActions (492 stars, last pushed 4d ago), licensed Apache-2.0. It adds 505 tokens to every session, about $0.0025 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
langroid CLAUDE.md
Instructions for langroid/langroid, covering claude.md, commands, development, testing and linting and type checking.
arxiv-mcp-server CLAUDE.md
Instructions for blazickjp/arxiv-mcp-server, covering claude.md, project, setup and validation, architecture and tool groups.
helicone AGENTS.md
Instructions for Helicone/helicone, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
langwatch CLAUDE.md
Instructions for langwatch/langwatch, covering langwatch, before you implement anything, development environment, running with no container runtime and local dev by hostname — thuishaven / portless (recommended).
AstrBot AGENTS.md
AGENTS.md instructions for AstrBotDevs/AstrBot, covering setup commands, core, dashboard(webui), pre-commit setup and dev environment tips.
AstrBot copilot-instructions.md
Copilot instructions for AstrBotDevs/AstrBot, covering astrbot development instructions, working effectively, bootstrap and install dependencies, running the application and dashboard build (vue.js/node.js).