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/victusfate/scaffold/agents-mdgit clone --depth 1 https://github.com/victusfate/scaffoldWrote 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/victusfate/scaffold/agents-md)<a href="https://agentmods.dev/instructions/victusfate/scaffold/agents-md"><img src="https://agentmods.dev/badge/instructions/victusfate/scaffold/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 | $0.04012 | $0.04012 |
| Opus 5 | $0.02006 | $0.02006 |
| Sonnet 5 | $0.00802 | $0.00802 |
| Haiku 4.5 | $0.00401 | $0.00401 |
Grade A, and why
scaffold 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 3d 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.
How it starts
The opening of the file, as written. The whole thing — 315 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Guidelines
Session Start
On your first response in a new session:
- Sync check — A
SessionStarthook fetchesorigin/mainand warns if the current branch is behind. If you see that warning, rebase before starting new feature work (git rebase origin/main) or pull if onmain(git pull origin main). - PR watch — If the current branch has an open PR, subscribe to it with
subscribe_pr_activity. Only watch the PR for this session's branch — not PRs from other branches. - Artifacts check — Check
./docs/for existing feature artifacts (design.md,prd.md,plan.md).
- Artifacts exist: acknowledge them and ask how to continue.
- No artifacts: start
/feature-chain— no permission needed. If the user's intent is vague or unstated, the grill (Phase 1) resolves it through Q&A. Do not ask a pre-question first.
Responsiveness & Steerability
The user can see when a path is wrong even when it's technically valid. The interface exists so they can steer the intelligence. Optimize for being steerable, not for autonomous completion. A user who can't redirect you is paying for expensive autocomplete.
Answer the user first
- When a message arrives mid-work, address that message before resuming your prior agenda. Never finish a long self-directed run and treat their message as an afterthought.
- Keep turns tight — fewer tool calls between check-ins — so there are frequent moments to listen. Don't go heads-down for ten steps.
- If you're about to go quiet for a while, say so first.
- Harness limit to be honest about: a single in-flight tool call can't be
preempted; the instant it returns, their message is the priority. (The user
can press
Escto interrupt — but they shouldn't have to.)
Surface direction before building
- Show the fork, the tradeoff, and the cost before committing effort, so the user can veto at the cheap moment instead of after a rebuild.
- Flag your own over-engineering. If a simpler path exists, name it first — don't make the user catch it.
- Before recommending on a multidimensional problem, map its shape first — see Understand the shape of the problem below.
- Prefer reversible moves so course-correcting costs a sentence, not a redo.
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.
- 3d ago First seen · 315 lines · 4,012 tokens per session scan A 0a237e6f38f1
scaffold AGENTS.md is an instructions file published in the GitHub repository victusfate/scaffold (6 stars, last pushed 12d ago), licensed MIT. It adds 4,012 tokens to every session, about $0.0201 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.
Other instructions, from other repositories
mini-swe-agent copilot-instructions.md
Copilot instructions for SWE-agent/mini-swe-agent, covering style guide, test style, bad and good.
intelligent-terminal rust.instructions.md
Concise Rust coding conventions for this repository.
lime AGENTS.md
AGENTS.md instructions for limecloud/lime, covering lime 源码仓库指南, 事实源, 工程约束, 协议与质量 and 执行方式.
generative-media-skills AGENTS.md
AGENTS.md instructions for calesthio/generative-media-skills, covering agent instructions, non-negotiable rules and allowed consistency.
AICOO-Skills AGENTS.md
Instructions for Aicoo-Team/AICOO-Skills, covering agents.md, if you just installed this pack (read first), repo purpose, editing rules and documentation sync.
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.