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/n0819/sonder_engine/agents-mdgit clone --depth 1 https://github.com/N0819/Sonder_EngineWhat 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 | $0.27758 | $0.27758 |
| Opus 5 | $0.13879 | $0.13879 |
| Sonnet 5 | $0.05552 | $0.05552 |
| Haiku 4.5 | $0.02776 | $0.02776 |
Grade A, and why
Sonder_Engine 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 — 696 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Editing Guide for Coding Agents
This file is the operational map for changing Sonder Engine safely. It is written for both human contributors and AI coding agents.
First-pass orientation
Before editing behavior:
- Read
Design.md§ Genre-agnostic substrate, world-specific law before adding or specializing a mechanic. - Read
docs/guides/PIPELINE.mdfor execution order and ownership boundaries. - Search
docs/CODE_MAP.mdfor the handler or function involved. - Read
docs/guides/DATABASE.mdbefore changing persistent state, archives, or restore paths. - Read the relevant schema in
llm/schemas.pybefore changing any model output. - Read the corresponding commit function before adding fields that should persist.
- Read
docs/guides/TESTING.md, find the nearest regression test, and run the narrow test first.
docs/CODE_MAP.md is generated; never hand-edit it. Regenerate and verify it
after moving or adding functions:
make map
make structure
AGENTS.md, Design.md and everything in docs/guides/ are the maintained
guidance set — current implementation authority, changed in the same commit as
the behaviour. docs/design/, docs/experiments/ and docs/archive/ are
context rather than authority: a design note argues for one subsystem, an
experiment record reports one unrepeatable run, and an archived document has
been superseded. Check any of their claims against source and the maintained
guides before acting on them. docs/README.md is the index
and says which is which.
Edit routing
Charter identity changes route through world/charter_identity.py,
world/charter_model.py and world/charter_runtime.py. The Charter/body dict
key is permanent identity. A generated personal name is materialized once;
profile edits affect only unnamed future bodies. A pool sizes the law's
phonology, never the population: the allocator spreads each
spread_components element breadth-first (family, because recurrence inside
one generated population implies kinship nobody authored, plus any component
address_components says the law calls people by alone) — authored pool
values first, then the extension lane (extension_profile: the profile's own
syllable parts, else parts derived from the pool's own words by
derived_name_parts), with reuse only once the whole reachable space is
spent. Spreading is preference; name_is_reserved stays the only refusal
beside components_repeat (a name may not double its own component). No
pool means no derivation — the no-implicit-default doctrine is unchanged, and
both minting paths (materialize_body_names and
story/naming.minted_presence_name) share the extension.
Rank/post titles are aliases,
and transcript color derives from the permanent identity so title changes and
promotion do not repaint the person. Name learning must still prove an exact
delivered line and physical co-location; index the roster once per beat rather
than scanning a large institution once per quote.
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 · 696 lines · 27,758 tokens per session scan A d0845e0f34bc
Sonder_Engine AGENTS.md is an instructions file published in the GitHub repository N0819/Sonder_Engine (57 stars, last pushed 4d ago), licensed MIT. It adds 27,758 tokens to every session, about $0.1388 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
claudemon CLAUDE.md
Instructions for zamarrowski/claudemon, covering ai agent guidelines, project shape, general guidelines, general rules and guards & defensive code.
unity-code-style-guide AGENTS.md
Instructions for krogh-jacobsen/unity-code-style-guide, covering agents.md — unity 6 c, project setup — edit this block, never do these — they corrupt the project, deprecated in unity 6 and if you read nothing else.
unity-cli-skill CLAUDE.md
Instructions for niqibiao/unity-cli-skill: Project instructions are maintained in and imported from @AGENTS.md.
Nyamu AGENTS.md
Instructions for polyblank66/Nyamu, covering about mcp, nyamu mcp workflow guidelines, file operation workflows, compilation tools and error handling.
sts2-llm copilot-instructions.md
Instructions for ttxttx1111/sts2-llm, covering copilot instructions for this repository, source of truth for customizations, runtime assumptions, authoring rules for this repo and gameplay-specific expectations.
game-and-watch-retro-go-sd CLAUDE.md
Claude Code instructions for sylverb/game-and-watch-retro-go-sd, covering claude.md, what this project is, build / flash workflow, architecture and three storage tiers, one elf.