relaycore AGENTS.md

A project instruction file for RelayCore, a system used to store durable project rules, decisions, lessons, and other working memory.

In plain words
What is it for?
Use it to load project context at the start of work, follow required workflows, and save lasting rules and decisions to RelayCore.
Why use it?
It prevents important project knowledge from being lost between sessions or kept only in an agent's temporary memory.

Instructions file for CodexOpenCode

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/totooss/relaycore/agents-md
Clone the repo
git clone --depth 1 https://github.com/totooss/relaycore

Made for: Codex, OpenCode.

Per session 844 This file is loaded in full into every session.
When invoked 844 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 $0.00844 $0.00844
Opus 5 $0.00422 $0.00422
Sonnet 5 $0.00169 $0.00169
Haiku 4.5 $0.00084 $0.00084

Measured yesterday against content hash a9c516684867, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

relaycore 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 · 64 lines

How it starts

The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.

RelayCore Project Instructions

This project uses RelayCore as the only durable project memory layer.

RULES.md is the versioned editorial source for durable methodology and runtime rules. Sync it into RelayCore with relaycore sync-rules --rules-file RULES.md whenever the file changes.

Memory Policy

  • Treat native model memory as optional, unstable, and non-authoritative.
  • Do not rely on ChatGPT, Codex, Claude, or any other runtime's built-in memory for project continuity.
  • Do not store durable project decisions only in native agent memory.
  • Durable memory for this project means any reusable decision, rule, lesson, rejected option, artifact reference, or workflow state that should survive the current session.
  • If durable memory is not written to RelayCore, treat it as not stored.

Required Workflow

  1. At task start, call memory_auto_prepare.
  2. If memory_auto_prepare is unavailable, fall back to memory_begin_task and memory_context.
  3. Before planning or implementation, do a short preflight that names the relevant loaded rules, decisions, lessons, and rejected paths that apply to the task.
  4. When a durable fact, decision, rule, or lesson appears, write it with memory_add or memory_propose.
  5. When a decision supersedes or rejects another option, preserve the rejected path and reason through RelayCore memory flows.
  6. Before ending the task, call memory_commit_task.

Required RelayCore Endpoints

  • MCP: http://127.0.0.1:9090/mcp
  • Mission Control: http://127.0.0.1:8080/mission-control
  • Memory Viewer: http://127.0.0.1:8080/mission-control/memories

Multi-Runtime Topology

  • Treat RelayCore MCP as the shared backend for cross-session and cross-runtime collaboration.
  • Multiple local runtimes such as Codex, Claude Code, or Hermes can share one relaycore mcp-http process on the same machine.
  • Mission Control is optional for agent tool calls; it is primarily the operator UI and manual override surface.
  • Use the same session_id when different runtimes should collaborate on the same task.
  • Use distinct agent_id values for different runtimes or instances so heartbeats, events, and command handling remain attributable.
  • If runtimes are on different machines or isolated containers, they must all reach the same shared RelayCore service and canonical database.

Read the full file on GitHub · 64 lines

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 · 64 lines · 844 tokens per session scan A a9c516684867

Subscribe to this mod's changes

relaycore AGENTS.md is an instructions file published in the GitHub repository totooss/relaycore (1 stars, last pushed 1mo ago), licensed MIT. It adds 844 tokens to every session, about $0.0042 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.