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/edgee-ai/edgee/agents-mdgit clone --depth 1 https://github.com/edgee-ai/edgeeWhat 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 | $0.03206 | $0.03206 |
| Opus 5 | $0.01603 | $0.01603 |
| Sonnet 5 | $0.00641 | $0.00641 |
| Haiku 4.5 | $0.00321 | $0.00321 |
Grade A, and why
edgee 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 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.
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 — 238 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Context for coding agents working in this repository. CLAUDE.md is a symlink to this file.
What Edgee is
Edgee is an Agent Gateway, built in Rust by Edgee Cloud SAS (Paris) with a US subsidiary, Edgee Corporation (Delaware).
Edgee sits between AI agents and LLM providers as a transparent gateway. It intercepts, routes, compresses, meters, and secures every LLM request, with no code change on the agent side. It works with all major coding agents (Claude Code, Codex, Cursor, GitHub Copilot, OpenCode, CodeBuddy, Crush) and all provider plans, including consumer subscriptions — not just API credits.
Who it's for: engineering organisations (typically 150+ developers) whose coding-agent bills are growing fast and whose leadership lacks visibility and control over token spend. Buyers are VPs of Engineering, CTOs, and platform teams.
Core value: cut token costs by up to 70% while giving engineering leaders full control and visibility over agent usage.
Product pillars, in order of strategic importance
- Smart routing and budget management — the moat. The routing engine can redirect requests to open-weight models mid-task without quality loss. The upcoming Strategies system lets customers define routing policies scoped to a person, a squad, or the whole organisation (e.g. a $100 weekly Claude budget: at 75% consumption Opus requests reroute to Kimi K3, at 85% Sonnet goes to GLM 5.2, at 100% everything routes to DeepSeek).
- Team observability — cost and usage per developer, repo, PR, model, and environment.
- Token compression — input trimming and output brevity, semantically lossless for coding tasks. A strong acquisition hook, but expected to become a commodity.
Positioning guardrails (apply to all user-facing copy)
- The category is Agent Gateway. Keep it. Lead with token cost reduction through routing and control — that's what converts and what wins POCs. Compression is a supporting feature, never the headline.
- Never write "Agent Engineering Platform" (or equivalents) in product copy. It reads as a platform for building agents — LangChain/CrewAI territory — whereas Edgee governs agents. The category label only broadens once the desktop app, Strategies, and skills/MCP management have shipped and been adopted.
- There is an expansion narrative — enter through cost reduction, become the control plane through which every AI agent in the company operates (routing, budgets, observability, skills, MCP, governance), the way Datadog expanded from infrastructure monitoring. It is reserved for the Series A deck, the vision page, and founder talks. It does not belong in the README, the CLI, the docs, or any copy you write here.
- Roadmap items may be named as roadmap items (see README) without implying a category change.
- Against OpenRouter and LiteLLM (the competitors seen in POCs): OpenRouter is a gateway for apps that consume LLMs; Edgee is specialised for agents, wrapping the agent itself with zero code change. Differentiators: agent-specialised, token compression, observability per developer/repo/PR rather than per API key, native 500-developer onboarding (seats, budgets, per-squad policies), and budget-driven Strategies with mid-task rerouting.
- Compliance facts that may be stated: SOC 2 and GDPR compliant, on-premise deployment available, BYOK supported.
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 · 238 lines · 3,206 tokens per session scan A 647a6287e7ae
edgee AGENTS.md is an instructions file published in the GitHub repository edgee-ai/edgee (127 stars, last pushed 5d ago), licensed Apache-2.0. It adds 3,206 tokens to every session, about $0.0160 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
agentic-playwright skill-creator.instructions.md
Instructions for idavidov13/agentic-playwright, covering skill creator, communicating with the user, creating a skill, capture intent and interview and research.
AgentEval copilot-instructions.md
Instructions for AgentEvalHQ/AgentEval, covering agenteval - ai coding agent instructions, architecture overview, environment setup, optional: secondary models for comparison and build & test commands.
framework AGENTS.md
Instructions for ai-driven-dev/framework, covering agents.md, behavior, communication, action and memory management.
AgentEval tracing.instructions.md
Guidelines for implementing trace recording and replay.
paiml-mcp-agent-toolkit AGENTS.md
Instructions for paiml/paiml-mcp-agent-toolkit, covering antigravity global rules for pmat, agent guidelines and contract-first, and how it is enforced.
maestro-orchestrate GEMINI.md
Instructions for josstei/maestro-orchestrate, covering maestro techlead orchestrator, startup checks, gemini cli integration constraints, context budget and settings reference.