ainative-opencode-memory-mcp AGENTS.md

Instructions for giving OpenCode persistent memory between work sessions. OpenCode is a coding agent that normally starts each session without knowledge of earlier decisions, conventions, or fixes.

In plain words
What is it for?
Use it to reload relevant project context at the start of a task, save durable decisions or warnings, and keep useful knowledge across sessions.
Why use it?
It reduces repeated explanations and helps prevent the agent from making the same mistakes in later sessions. It also describes what information should be saved and recalled.

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/ainative-studio/ainative-opencode-memory-mcp/agents-md
Clone the repo
git clone --depth 1 https://github.com/AINative-Studio/ainative-opencode-memory-mcp

Made for: Codex, OpenCode.

Per session 617 This file is loaded in full into every session.
When invoked 617 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 $0.00617 $0.00617
Opus 5 $0.00309 $0.00309
Sonnet 5 $0.00123 $0.00123
Haiku 4.5 $0.00062 $0.00062

Measured 2d ago against content hash ce1d10745860, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ainative-opencode-memory-mcp 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.

AGENTS.md · 49 lines

How it starts

The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Onboarding — AINative Memory (ZeroMemory for OpenCode)

You have persistent, cross-session memory through the ainative-memory MCP server. OpenCode is stateless by default; these tools let you carry decisions, conventions, and gotchas from one session into the next.

Zero setup — you already have memory

You do not need to procure an API key to start. On first run this server auto-provisions a working trial key for you (via AINative instant-db) and caches it. If AINATIVE_API_KEY is already set, it uses that instead. Either way, the memory tools just work — call them.

The trial gives 1,000 memory records + 10,000 vectors. When you store something worth keeping, the tool returns a claim link — surface it to the human so they can turn the trial into a permanent (free-trial Hobbyist $5/mo) account and never lose the memory.

The five tools

Tool When to call it
opencode_recall_context(task) At the start of every task. Reloads the most relevant prior context so you don't repeat past mistakes.
opencode_store_memory(content, tags?) The moment you learn something durable: a decision, a convention, a fix, a "don't do X".
opencode_search_memory(query, limit?) When you need a specific past fact mid-task.
opencode_memory_stats() To check how much you've remembered.
opencode_clear_memory(tag?) To prune stale memory (by tag, or all).

Operating rhythm (do this every session)

  1. Start: call opencode_recall_context with a short description of the task. Treat what comes back as ground truth about this project.
  2. During: whenever you make a non-obvious decision or discover a gotcha, opencode_store_memory it in one sentence. Tag it (["decision"], ["gotcha"], ["convention"]) so it's easy to prune later.
  3. Don't re-derive things already in memory. Recall first, work second.

What to remember (and what not to)

  • Remember: architecture decisions, naming conventions, non-obvious constraints, bugs and their fixes, "we chose A over B because…", API quirks.
  • Skip: anything trivially re-derivable from the code, secrets, or one-off scratch state.

Read the full file on GitHub · 49 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. 2d ago First seen · 49 lines · 617 tokens per session scan A ce1d10745860

Subscribe to this mod's changes

ainative-opencode-memory-mcp AGENTS.md is an instructions file published in the GitHub repository AINative-Studio/ainative-opencode-memory-mcp (0 stars, last pushed 1mo ago), licensed MIT. It adds 617 tokens to every session, about $0.0031 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.