agentmods.dev

agent-runtime-proof AGENTS.md

A set of project instructions for Agent Runtime Proof, a project that checks whether local coding-agent processes match stated expectations. It defines the project’s boundaries, evidence rules, and first-version scope.

In plain words
What is it for?
Use it when building or changing Agent Runtime Proof, especially for local runtime inspection, command-line and local MCP interfaces, platform support, and acceptance checks.
Why use it?
It keeps development work aligned with the project’s rules and prevents unsupported claims about what a running process or installed artifact proves.

Instructions file for CodexOpenCode

1 4d ago A scan 380 tokens original Apache-2.0
View the source file fantasyce/agent-runtime-proof
Install

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.

agentmods
npx agentmods add instructions/fantasyce/agent-runtime-proof/agents-md
Clone the repo
git clone --depth 1 https://github.com/fantasyce/agent-runtime-proof

Made for: Codex, OpenCode.

Per session 380 This file is loaded in full into every session.
When invoked 380 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00380 $0.00380
Opus 5 $0.00190 $0.00190
Sonnet 5 $0.00076 $0.00076
Haiku 4.5 $0.00038 $0.00038

Measured yesterday against content hash 3539848fe7d1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-runtime-proof 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 yesterday.

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.

AGENTS.md · 42 lines

What it actually says

Agent Runtime Proof Project Rules

Product Boundary

Agent Runtime Proof is an independent open-source Across ecosystem project. It must build and run without Across Agents Assistant, Across Context, Across Orchestrator, or Across Autopilot.

Do not place its implementation in any of those four repositories. Do not make it a fourth first-party managed AAA plugin. AAA may consume it only through the same generic local interface available to other Agent hosts.

Version 1 Scope

  • Local Agent runtimes only.
  • CLI and local stdio MCP are the public interfaces.
  • A launch Witness may proxy a local child process byte-for-byte.
  • macOS and Windows require real-machine acceptance; Linux core, CLI, and MCP acceptance may run in local Docker.
  • Runtime and host inspection is read-only. The tool must not kill processes, rewrite Agent configuration, delete host caches, upgrade plugins, or repair installations.
  • No HTTP/SSE server, remote Agent support, cloud control plane, telemetry, multi-tenancy, or remote attestation in v1.

Evidence Rules

  • Keep verdict and proof strength separate.
  • Never claim loaded-memory proof from an on-disk digest.
  • Bind process evidence to PID plus creation time to prevent PID reuse.
  • Treat permission denial, races, scan limits, and incomplete evidence as an explicit indeterminate result, not a pass.
  • Do not persist raw environment values, complete command lines, credentials, transcripts, file contents, or unredacted home paths.
  • Tests must use task-owned roots and leave no process or filesystem residue.

Authoritative Design

docs/architecture-development-acceptance.md defines the architecture, contracts, development phases, platform matrix, and GO/NO-GO release gates. Scope or proof-semantics changes require updating that document and its tests.

Changes

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.

  1. yesterday First seen · 42 lines · 380 tokens per session scan A 3539848fe7d1

Subscribe to this mod's changes

agent-runtime-proof AGENTS.md is an instructions file published in the GitHub repository fantasyce/agent-runtime-proof (1 stars, last pushed 4d ago), licensed Apache-2.0. It adds 380 tokens to every session, about $0.0019 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.

Securityai-agentsdeveloper-toolsmcpprovenanceruntime-securitysupply-chain-security
Related

Other instructions, from other repositories

matrixscroll AGENTS.md

Instructions for SSX360/matrixscroll, covering agents.md, commit attribution and building the docs.

SSX360/matrixscroll · 621 tokens

matrixscroll CLAUDE.md

Instructions for SSX360/matrixscroll, covering claude.md, voice, name real things, banned words and no em-dashes.

SSX360/matrixscroll · 928 tokens

dev-machine-guard AGENTS.md

Instructions for step-security/dev-machine-guard, covering dev machine guard — coding guidelines, 0. prime directives, 1. project shape & where code goes, 2. cross-platform code and 2.1 the os boundary is executor.executor.

step-security/dev-machine-guard · 11,328 tokens

aguara CLAUDE.md

Instructions for garagon/aguara, covering claude.md, project status, build & test commands, single package test and single test function.

garagon/aguara · 5,264 tokens

aguara AGENTS.md

Instructions for garagon/aguara, covering agents.md - aguara reference for ai agents, quick start, install, scan a directory and scan with ci defaults (fail on high+, no color).

garagon/aguara · 2,769 tokens

grove CLAUDE.md

Instructions for jmilinovich/grove, covering grove, architecture rules, running locally, code conventions and what not to do.

jmilinovich/grove · 804 tokens