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/nativesoil/handover/agents-mdgit clone --depth 1 https://github.com/nativesoil/handoverWrote 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/nativesoil/handover/agents-md)<a href="https://agentmods.dev/instructions/nativesoil/handover/agents-md"><img src="https://agentmods.dev/badge/instructions/nativesoil/handover/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 | $0.05788 | $0.05788 |
| Opus 5 | $0.02894 | $0.02894 |
| Sonnet 5 | $0.01158 | $0.01158 |
| Haiku 4.5 | $0.00579 | $0.00579 |
Grade C, and why
handover AGENTS.md scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf "$SOIL_HOME" How it starts
The opening of the file, as written. The whole thing — 523 lines — stays where its author put it; the contents beside it link to each section on GitHub.
For the agent reading this
This is the entry point, and it is written to be enough on its own for the
two things an agent does here: install Soil Handover for somebody, or work on
this repository. Rationale and depth live in docs/ and spec/. What you
have to DO is here.
What this project is
Soil Handover is an open JSON format for a project's working state, so work saved in one AI tool can be continued in another. A handover carries the decisions with their reasons, the constraints and why they bind, the approaches already rejected, the task in flight and the next step, in a document a person can read and any conformant implementation can parse.
What it solves is a boundary. A session fills up, compacts or ends. The work moves to another agent, another tool, another model provider, or to a colleague. A transcript preserves what was said, but not which decisions are current, which approaches were already abandoned, or what should happen next. A summary is smaller, and compression usually takes the reasons with it. "Use X" survives. "Use X because Y failed under load, and do not reopen Z" often does not, and a decision without its reason is a decision waiting to be relitigated.
This repository holds the whole thing: the specification (1.0.0), five
independent implementations of it (TypeScript, Python, Go, JVM and .NET), a
soil command line tool, a local MCP server over stdio, a self-hostable
server preview, and a conformance suite that every implementation is held to.
Apache 2.0, copyright Mastodont AB.
The local round trip needs no account, no network and no telemetry.
Handovers are plain JSON files under ~/.soil (or $SOIL_HOME), one file
per handover, readable with cat. Nothing in this repository calls a hosted
service, needs one, or degrades without one.
Nothing here is published to a package registry yet, so every install today builds from a checkout.
The boundary that is easy to get wrong
A memory system helps an agent remember inside one system. Soil Handover moves the current working state between systems. If you catch yourself explaining this as "so your assistant remembers things", you have described something else.
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 · 523 lines · 5,788 tokens per session scan C 6dec5d2c5ba6
handover AGENTS.md is an instructions file published in the GitHub repository nativesoil/handover (19 stars, last pushed 6d ago), licensed Apache-2.0. It adds 5,788 tokens to every session, about $0.0289 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
ratel AGENTS.md
AGENTS.md instructions for ratel-ai/ratel, covering agents.md — working in the ratel repo, build & test, rust, ts and python (from src/sdk/python/; needs uv).
Context-Engine GEMINI.md
Gemini CLI instructions for Context-Engine-AI/Context-Engine, covering ai agent rules for context-engine mcp tools, stop — read this first, glossary, tool selection decision tree and requirements.
laravel-auditor AGENTS.md
Instructions for MrPunyapal/laravel-auditor, covering laravel auditor, package conventions, quick commands and local skills.
json-difference AGENTS.md
Instructions for lukascivil/json-difference, covering general guidelines for working with nx, scaffolding & generators and when to use nxdocs.
elephant-agent copilot-instructions.md
Copilot instructions for agentic-in/elephant-agent, covering github copilot instructions, source of truth order, review priorities, harness-specific checks and validation expectations.
reapfield AGENTS.md
AGENTS.md instructions for PedroHenriqueNS/reapfield, covering agents.md, the short version, two rules that override anything else you infer and commits.