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/iblai/api/claude-mdgit clone --depth 1 https://github.com/iblai/apiWhat 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 | $0.02797 | $0.02797 |
| Opus 5 | $0.01399 | $0.01399 |
| Sonnet 5 | $0.00559 | $0.00559 |
| Haiku 4.5 | $0.00280 | $0.00280 |
Grade B, and why
api CLAUDE.md scanned grade B with 2 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.
Reads MCP configurationmediumAgent snooping
mcp.json carries server URLs and auth tokens; reading it lets a mod discover and abuse other integrations.
**Skills vs MCP server — REST vs runtime chat.** Skills do everything reachable over REST (configure agents, datasets, memory, users, roles, notifications, discovery/search, recommendations, user profile + analytics, rep Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
5. `## Example` — one realistic `curl`. How it starts
The opening of the file, as written. The whole thing — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Repository Purpose
iblai/api is a toolkit for operating the ibl.ai platform from an AI agent. It ships two things:
- Skills (
skills/) — installable vianpx skills add iblai/api. Each/iblai-api-*skill maps one agent-configuration or platform-admin operation to its exactapi.iblai.appREST endpoints (method, URL, body). They drive the platform directly — no UI. - MCP server (
mcp/) — a single hosted Model Context Protocol server,iblai-api-agent-chat, for the one runtime capability that isn't a REST call: holding a live conversation with a deployed agent (streamed responses, tool use, RAG).
This is the API-driven complement to iblai/vibe, which provides UI components for building ibl.ai apps.
Naming. Skills are iblai-api-*; scope is encoded by prefix. iblai-api-agent-* acts on one agent; iblai-api-profile* on the signed-in user. Org-wide is the default scope, so those skills are bare — iblai-api-management, iblai-api-rbac, iblai-api-crm, iblai-api-token, iblai-api-integration, iblai-api-notification, iblai-api-invite — and iblai-api-org itself is the org-settings skill. Cross-cutting skills are bare too: iblai-api-search, iblai-api-analytics, iblai-api-course-create, iblai-api-login. (These iblai-api-* names overlap some of vibe's; that's accepted.)
Skills describe APIs, not UIs. Each skill documents endpoints (method, URL, body) and when to use them. Do not reference user interfaces (menus, tabs, buttons, pages, "Edit Agent → …") in skill prose — the UI was only a means to discover endpoints; the value is the API.
Skills vs MCP server — REST vs runtime chat. Skills do everything reachable over REST (configure agents, datasets, memory, users, roles, notifications, discovery/search, recommendations, user profile + analytics, reporting). The MCP server is only for live agent chat — and even wiring it up is a skill (/iblai-api-agent-chat writes the .mcp.json / claude mcp add config from the .env token + a chosen agent's unique_id), so the whole offering installs via npx skills add iblai/api. Rule: if a skill covers it, there is no server for it. That's why only iblai-api-agent-chat remains — the former analytics, agent-create, search, and user servers were removed once skills covered analytics (/iblai-api-analytics), agent creation (/iblai-api-agent-create), discovery + recommendations (/iblai-api-search), and user profile + analytics (/iblai-api-profile).
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 · 96 lines · 2,797 tokens per session scan B 0a8f5e680869
api CLAUDE.md is an instructions file published in the GitHub repository iblai/api (15 stars, last pushed 3d ago), licensed MIT. It adds 2,797 tokens to every session, about $0.0140 per session on Opus 5. A static security scan graded it B with 2 findings (reads mcp configuration, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
ios-agent-skill AGENTS.md
Instructions for Nagarjuna2997/ios-agent-skill, covering ios agent skill — claude ai expert ios/swift developer, when to load this skill, loading the right document, how these docs are structured and how you operate: delegation, loops, and verification.
bluetemberg CLAUDE.md
Instructions for prototypdigital/bluetemberg, a project described as: AI coding standards shipped like a dependency. Version-locked, integrity-verified, signed, and routed by role and stack version. Emits AGENTS.md, Cursor rules, and more.
Personetta CLAUDE.md
Instructions for EdwardAF-IT/Personetta, covering repository instructions, commit attribution and quality gates.
ultra-max-token-burner AGENTS.md
Instructions for floytra-dev/ultra-max-token-burner: When the user asks to expand, operationalize, professionalize, govern, validate, or make a prompt more manager-visible, use the Ultra Max Token Burner workflow in SKILL.md.
skraft-plugin AGENTS.md
AGENTS.md instructions for SebastienDegodez/skraft-plugin, covering graphify, skraft-framework documentation, repository structure, test placement rules and vally evaluation rules.
sdd-starter CLAUDE.md
Claude Code instructions for KazukiShinomiya/sdd-starter, covering claude.md, 共有文脈（全エージェント共通）, このプロジェクトの核（変更時に守る） and claude 固有の運用指針.