agent-stdlib AGENTS.md

A guide for adapting the agent-stdlib add-on pack to different coding-agent systems, such as Cursor, Cline, or a custom agent.

In plain words
What is it for?
It is for mounting MCP servers, running Python scripts, loading skills as written rules, and adapting hooks, commands, and agents.
Why use it?
It explains which parts can be reused directly and which parts need new wiring for another agent system.

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/hoja-solutions/agent-stdlib/agents-md
Clone the repo
git clone --depth 1 https://github.com/Hoja-Solutions/agent-stdlib

Made for: Codex, OpenCode.

Per session 969 This file is loaded in full into every session.
When invoked 969 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 $0.00969 $0.00969
Opus 5 $0.00485 $0.00485
Sonnet 5 $0.00194 $0.00194
Haiku 4.5 $0.00097 $0.00097

Measured yesterday against content hash 2510a8ac604b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agent-stdlib 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 · 85 lines

How it starts

The opening of the file, as written. The whole thing — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Using agent-stdlib in any harness

This pack ships as a Claude Code plugin, but most of it is not Claude-specific. This file maps each part to its portable form for harnesses like OpenCode, Cursor, Cline, or a custom agent, on any tool-using model.

What ports cleanly

  • MCP servers (mcp-servers/think, tool-gateway, code-execution): MCP is an open protocol, so any MCP client can mount them. See "Mount the MCP servers".
  • Scripts (skills/*/scripts/*.py, scripts/locks.py): plain Python, run anywhere.
  • scripts/autonomy_loop.sh: set AGENT_RUNNER to your headless agent CLI (it defaults to claude -p), e.g. AGENT_RUNNER="opencode run".

What needs re-wrapping

  • Skills (skills/<name>/SKILL.md): the content is harness-neutral; the auto-loading and triggering is a Claude Code feature. Load the file as a rule or paste its body into your system prompt. See "Use the skills as plain content".
  • hooks (hooks/): the classification logic in both the action gate and the injection screen is reusable; the hooks.json wiring and the input JSON shapes are Claude Code's. See "Port the hooks".
  • Commands and agent (commands/, agents/): Claude Code formats that also rely on its subagent dispatch. Reuse the recipe inside them, not the files.

Use the skills as plain content

Every skill is a self-contained markdown procedure. Point your agent at skills/<name>/SKILL.md, or copy the body into your rules file (AGENTS.md, .cursorrules, a system prompt). The frontmatter description says when each skill applies, which is enough for a router to pick one. On a smaller model, load the one you need explicitly rather than expecting it to self-select.

Mount the MCP servers

The servers speak MCP over stdio and need uv, which installs each one's single dependency on first run. A vendor-neutral config most clients accept:

{
  "mcpServers": {
    "think": {
      "command": "uv",
      "args": ["run", "/abs/path/agent-stdlib/mcp-servers/think/server.py"]
    },
    "tool-gateway": {
      "command": "uv",
      "args": ["run", "/abs/path/agent-stdlib/mcp-servers/tool-gateway/server.py"]
    }
  }
}

Read the full file on GitHub · 85 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 · 85 lines · 969 tokens per session scan A 2510a8ac604b

Subscribe to this mod's changes

agent-stdlib AGENTS.md is an instructions file published in the GitHub repository Hoja-Solutions/agent-stdlib (1 stars, last pushed 1mo ago), licensed MIT. It adds 969 tokens to every session, about $0.0048 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.