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/sema-lisp/sema/agents-mdgit clone --depth 1 https://github.com/sema-lisp/semaWrote 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/sema-lisp/sema/agents-md)<a href="https://agentmods.dev/instructions/sema-lisp/sema/agents-md"><img src="https://agentmods.dev/badge/instructions/sema-lisp/sema/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 | $0.07158 | $0.07158 |
| Opus 5 | $0.03579 | $0.03579 |
| Sonnet 5 | $0.01432 | $0.01432 |
| Haiku 4.5 | $0.00716 | $0.00716 |
Grade A, and why
sema 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 4d 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 — 218 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Sema (Lisp with LLM primitives, in Rust)
Canonical instructions for any coding agent working in this repo. CLAUDE.md just points here.
Shipping invariant — NO dev-checkout assumptions in shipped code (HARD RULE)
A shipped sema (crates.io / cargo install, a cargo-dist archive, the shell
installer, or Homebrew) runs on a machine that has none of: this git repo,
jake, npm/node/wasm-pack, gitignored-or-generated build artifacts, or the
maintainer's absolute paths. Any user-facing behavior that depends on those is a
shipped bug, not a dev convenience. This is non-negotiable. It already
happened once: sema web shipped telling users to run jake wasm.web-runtime
because its runtime assets were gitignored and a build.rs file-existence check
silently compiled the feature out. It must never recur. Enforced rules:
- Anything a shipped command needs at runtime MUST be a git-tracked package
input embedded at build time (
include_str!/include_bytes!/rust_embedover a tracked path).cargo packageships only git-tracked files, so neverinclude_*!a gitignored or build-generated path — the.cratewould lack it. If ajakerecipe generates the artifact, commit the generated output (don't.gitignoreit), and verify it embeds the whole directory (preferrust_embed/include_dirover a hand-maintained file list that can drift). build.rsmust NEVER change what compiles based on file existence, an env var, or the network.Path::new(...).exists()gating arustc-cfgis true in your tree and false from a packaged.crate→ a silently reduced binary. Use an unconditionalinclude_*!(a missing input then fails the build loudly) or an explicit, documented Cargo feature — never an implicit file-presence gate.- No runtime path may point into the source tree.
env!("CARGO_MANIFEST_DIR"),env!("OUT_DIR"), and repo-relative reads (examples/,website/,Jakefile, …) are fine intests/andbuild.rs(they run from the checkout) but forbidden in shipped runtime code. Resolve companion binaries viastd::env::current_exe(), not repo layout. - Never emit a user-facing message telling an installed user to run
jake,make,cargo build,npm, or otherwise assume they have the repo. If a feature can be absent, degrade with an action the end user can actually take. - Shelling out to a dev tool (
jake,cargo,npm,node,wasm-pack,vercel) from a shipped code path is a bug.gitis acceptable only where it is the feature's inherent prerequisite (sema pkg, git stdlib bindings).
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.
- 4d ago First seen · 218 lines · 7,158 tokens per session scan A 83809c08f880
sema AGENTS.md is an instructions file published in the GitHub repository sema-lisp/sema (41 stars, last pushed today), licensed MIT. It adds 7,158 tokens to every session, about $0.0358 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
AIL-old CLAUDE.md
Instructions for hyun06000/AIL-old, covering cast — 이 프로젝트를 만드는 이름들, core philosophy, rule 1 — 벤치마크가 유일한 이정표, rule 2 — 언어 기능 추가 필터 and rule 3 — 금지 목록 (hyun06000 명시 승인 필요).
fluxon CLAUDE.md
Instructions for fluxon-lang/fluxon, covering claude.md — how to work on fluxon (for ai agents), 0. what this project is, 1. language, 2. where things are (navigation) and 3. build, test, run.
vera CLAUDE.md
Claude Code instructions for aallan/vera, covering claude.md — project orientation for claude code, virtual environment, key commands, project layout and writing vera code.
vera AGENTS.md
AGENTS.md instructions for aallan/vera, covering agents.md — instructions for ai agents, for agents writing vera code, conformance programs as reference, workflow and commands.
autolith AGENTS.md
AGENTS.md instructions for lambda-symbolics/autolith, covering repository guidelines, purpose and sources of truth, upstream references, architectural guardrails and package policy.
squeez CLAUDE.md
Claude Code instructions for claudioemmanuel/squeez, covering claude.md, commands, hard constraints, architecture and key files.