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/codeyogi911/able/agents-mdgit clone --depth 1 https://github.com/codeyogi911/ableWhat 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 | $0.00768 | $0.00768 |
| Opus 5 | $0.00384 | $0.00384 |
| Sonnet 5 | $0.00154 | $0.00154 |
| Haiku 4.5 | $0.00077 | $0.00077 |
Grade A, and why
able AGENTS.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 — 35 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Able
Goal
Build a trustworthy, agent-first ERP for service businesses. Able Desk is the first application: a single-tenant Desk + CRM foundation on Cloudflare Workers where MCP is the normal operator interface, customers use accessible approved channels, and /ops is a small Cloudflare Access-protected recovery surface. Grow through deep business modules with small agent interfaces—not generic CRUD, a plug-in framework, or a universal object model.
Working context
Two different jobs happen in this repository. Decide which one you are doing before you change anything.
- Upstream work — improving Able itself. Everything you produce must be tenant-neutral: generic issues, neutral fixtures, no business named anywhere. This is the default when no private context is present.
- Deployment work — configuring, deploying, or extending Able for one specific business. If
AGENTS.local.mdexists, read it first: it holds that deployment's standing context. Tenant facts belong there, in the deployment platform, or in runtime workspace settings — never in tracked files, tests, fixtures, commit messages, or issues.
Prefer the lowest customization level that solves the problem: runtime workspace settings, then a module or adapter that respects the public seams, then core change. docs/adopt.md defines the boundary and the levels; when deployment work uncovers a generic defect, fix it upstream with a neutral test instead of patching locally.
Operating model
Act as the lead orchestrator. Start from the user outcome, inspect relevant evidence, define the completion bar, and own integration and final verification. For sizeable work with genuinely independent streams, delegate bounded research, implementation, or review tasks with the context, constraints, success criteria, and expected return each delegate needs. Run safe independent reads in parallel. Work directly when the task is simple, sequential, or tightly coupled; do not delegate merely to duplicate work.
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 · 35 lines · 768 tokens per session scan A 19fc97a61f3a
able AGENTS.md is an instructions file published in the GitHub repository codeyogi911/able (1 stars, last pushed yesterday), licensed Apache-2.0. It adds 768 tokens to every session, about $0.0038 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 2026-08-31.
Other instructions, from other repositories
brass-monkey GEMINI.md
Instructions for actinon-com/brass-monkey, covering brass-monkey (gemini cli extension for odoo), project mandate, critical protocols, 1. orientation (the "world map" rule) and 2. explanations first & buy-in.
orbiteus AGENTS.md
Instructions for orbiteus/orbiteus, covering agent / contributor policy, vendor-neutral copy and links, readme hero tagline (locked), stack and boring tech filter.
brass-monkey CLAUDE.md
Instructions for actinon-com/brass-monkey, covering brass-monkey — claude code working context, non-negotiable rules for this initiative, shared mandates (consistent with gemini.md), explanations first & buy-in and structural fixes over band-aids.
S.HiveDesk AGENTS.md
AGENTS.md instructions for Akshit1018/S.HiveDesk, covering app builder workspace, 0. two worlds (read this first), project instructions, where you are and /workspace/startup.sh (required — you maintain this).
localtry-mcp AGENTS.md
AGENTS.md instructions for philly88r/localtry-mcp, a project described as: Tenant-isolated MCP server for operating and customizing LocalTry CRM from ChatGPT and other AI clients.
flarestarter CLAUDE.md
Instructions for FlareStarter/flarestarter: This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.