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/elle-lisp/elle/agents-mdgit clone --depth 1 https://github.com/elle-lisp/elleWrote 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.
[](https://agentmods.dev/instructions/elle-lisp/elle/agents-md)<a href="https://agentmods.dev/instructions/elle-lisp/elle/agents-md"><img src="https://agentmods.dev/badge/instructions/elle-lisp/elle/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.03753 | $0.03753 |
| Opus 5 | $0.01877 | $0.01877 |
| Sonnet 5 | $0.00751 | $0.00751 |
| Haiku 4.5 | $0.00375 | $0.00375 |
Grade A, and why
elle 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 3d 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 — 317 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Elle
Elle is a Lisp. Source text becomes bytecode; bytecode runs on a VM.
This is not a toy. The implementation targets correctness, performance, and clarity — in that order. We compile through multiple IRs, we have proper lexical scoping with closure capture analysis, and a unified signal/capability system: signals flow up (from callee to caller), capabilities flow down (from parent fiber to child). See signals for the full design.
You are an LLM. You will make mistakes. The test suite will catch them. Run the tests. Read the error messages. They are designed to be helpful.
origin/main is always green. Every commit on main passes every test —
Elle scripts, Rust tests, examples, documentation. This is enforced by CI
and a merge queue. If a test fails on your branch, your branch caused it.
"Pre-existing defect" is not a valid explanation when main is green. Fix
every failure before merging — no skip lists, no expected failures, no
excuses. See CONTRIBUTING.md for the full policy.
Contents
- Architecture
- Signals and capabilities
- Products
- Directories
- Testing
- Invariants
- Intentional oddities
- Conventions
- Maintaining documentation
- Where to start
Architecture
Source → Reader → Syntax → Expander → Syntax → Analyzer → HIR → Lowerer → LIR → Emitter → Bytecode → VM
This is the only compilation pipeline. Source locations flow through the entire
pipeline: Syntax spans → HIR spans → LIR SpannedInstr → LocationMap in
bytecode. Error messages include file:line:col information.
Key modules
Pipeline (compilation order):
reader— Lexing and parsing toSyntaxsyntax— Syntax types, macro expansionhir— Binding resolution, capture analysis, signal inference, linting, symbol extraction, docstring extractionlir— SSA form with virtual registers, basic blocks,SpannedInstrfor source trackingcompiler— Bytecode instruction definitions, debug formattingpipeline— Compilation entry points (seesrc/pipeline/AGENTS.md)
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.
- 3d ago First seen · 317 lines · 3,753 tokens per session scan A 74ce2b23eb65
elle AGENTS.md is an instructions file published in the GitHub repository elle-lisp/elle (24 stars, last pushed 3d ago), licensed MIT. It adds 3,753 tokens to every session, about $0.0188 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
z-engine AGENTS.md
Instructions for lisachenko/z-engine, covering working on z-engine, the one rule that is non-negotiable: version matching, branch model, generated engine definitions — never hand-edit and regenerating without docker (native mode).
effection AGENTS.md
Instructions for thefrontside/effection, covering agents.md — effection agent contract, core invariants (do not violate), operations vs promises, structured concurrency is scope-owned and effects do not escape scopes.
autolith AGENTS.md
Instructions for lambda-symbolics/autolith, covering repository guidelines, purpose and sources of truth, upstream references, architectural guardrails and package policy.
sema AGENTS.md
Instructions for sema-lisp/sema, covering agents.md — sema (lisp with llm primitives, in rust), build & test, architecture (cargo workspace), key design patterns and writing style — plain technical english (mandatory).
native-php-matrix CLAUDE.md
Instructions for lisachenko/native-php-matrix, covering working on native-php-matrix, running tests, quality gates (all enforced in ci), anatomy of a .phpt test and repository map.
Lisa CLAUDE.md
Instructions for youngde811/Lisa, covering lisa — lisp-based intelligent software agents, build & load, choosing a belief system, project structure and bridge endpoints (port 8090).