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/tetracoralla/deterministic-dependency-engine/agents-mdgit clone --depth 1 https://github.com/tetracoralla/deterministic-dependency-engineWrote 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/tetracoralla/deterministic-dependency-engine/agents-md)<a href="https://agentmods.dev/instructions/tetracoralla/deterministic-dependency-engine/agents-md"><img src="https://agentmods.dev/badge/instructions/tetracoralla/deterministic-dependency-engine/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 | $0.00600 | $0.00600 |
| Opus 5 | $0.00300 | $0.00300 |
| Sonnet 5 | $0.00120 | $0.00120 |
| Haiku 4.5 | $0.00060 | $0.00060 |
Grade A, and why
deterministic-dependency-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 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 — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deterministic Dependency Engine repository contract
Before any review, read docs/REVIEW_CONTRACT.md. A plain owner request to
review, audit, 审核, or 复核 invokes that contract end to end; treat it as the
minimum scope and report tools-dev workspace escalations without asking the
owner for another checklist.
This repository owns a deterministic dependency-reasoning product. Product
identity lives in docs/PRODUCT_IDENTITY.md; product meaning
lives in docs/PRODUCT_MODEL.md; current executable Zod models and tests are
the runtime authority.
Invariants
- The engine proves derivations over declared relations, never the truth of those relations.
- Public dependency edges are always named
dependentandprerequisite. Do not expose ambiguousfrom/toinput fields. - Graph structure, current satisfied facts, and the query remain separate.
- Equal candidates use code-point ordering. Input array order must not change a canonical result.
- V1 is a directed dependency reasoning engine. Do not add alternatives, conflicts, version ranges, provider selection, SAT/SMT, task execution, graph discovery, a graph database, or a visual graph editor without a new owner requirement.
- The authorized sphere viewer is a human-only, read-only projection. Keep camera state, coordinates, styles, focus, and rendering details out of the public graph, result, receipt, CLI, HTTP, MCP, and Skill contracts.
- Keep the transparent sphere as the dominant human surface. Source and Analysis are progressively disclosed exact-data layers, not competing default panels; the sphere must never imply an inferred or persisted spatial truth.
- Keep canvas chrome as bounded edge controls and overlays, not full-width bars or layout columns that resize the spatial graph.
- Library, CLI, MCP, HTTP, and UI call the same core operations. Adapters do not reimplement graph semantics.
- MCP accepts inline typed data only. It never resolves Agent-controlled file paths against ambient process state.
- The complete serialized result must respect the selected response limit. Requests, node counts, relation counts, query counts, and wall time are cumulatively bounded.
- A receipt identifies a calculation and its hashes. It is never called proof and never asserts that declared business relations are true.
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 · 54 lines · 600 tokens per session scan A 2c881170d850
deterministic-dependency-engine AGENTS.md is an instructions file published in the GitHub repository tetracoralla/deterministic-dependency-engine (0 stars, last pushed 4d ago), licensed Apache-2.0. It adds 600 tokens to every session, about $0.0030 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
lattice AGENTS.md
Instructions for plausibleventures/lattice, covering lattice — how to work in this repo, the eleven non-negotiables, layout, commands and house style.
fullbleed-official AGENTS.md
Instructions for fullbleed-engine/fullbleed-official, covering fullbleed repository guidance, product boundary, agent contracts, validation and release discipline.
lattice CLAUDE.md
Instructions for plausibleventures/lattice: See AGENTS.md — it is the constitution for this repo and applies to every agent and every human equally. Start with .lattice/kit.json for the machine-readable index of what exists, and .lattice/tasks.json for what to do next.
QGIS-Claude-Skill-Package CLAUDE.md
Instructions for Impertio-Studio/QGIS-Claude-Skill-Package, covering qgis skill package — claude.md, standing orders — read this first, identity, privacy protocol (p-000a) and workspace setup protocol (p-000b).
reelier AGENTS.md
Instructions for seldonframe/reelier, covering reelier — what it actually does, 1. one-line identity, 4. skill grammar (what a step can express), 5. substrate reality — this is not git-only and 6. reelier cloud — what it adds over oss.
cli AGENTS.md
Instructions for vibgrate/cli, covering agents.md, build, test, and check, prefer vg over grepping, determinism rules (non-negotiable) and public-only rule.