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/andypgray/loadbearing/agents-mdgit clone --depth 1 https://github.com/andypgray/loadbearingWrote 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/andypgray/loadbearing/agents-md)<a href="https://agentmods.dev/instructions/andypgray/loadbearing/agents-md"><img src="https://agentmods.dev/badge/instructions/andypgray/loadbearing/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 | $0.04949 | $0.04949 |
| Opus 5 | $0.02475 | $0.02475 |
| Sonnet 5 | $0.00990 | $0.00990 |
| Haiku 4.5 | $0.00495 | $0.00495 |
Grade A, and why
loadbearing 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 — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
LoadBearing — architecture for agents
The self-hosting principle
A compiler that cannot compile itself is not trusted with anyone else's source. LoadBearing accepts the same test: this repository is the tool's own first consumer. The spec in arch/Zphil.LoadBearing.ArchSpec governs this code, and every read verb runs against this solution in CI: check, render, status, graph, explain. The xUnit adapter runs the same rules as individually named tests. The renderer writes nine committed artifacts here, the block below and the diagram in ARCHITECTURE.md and seven per-directory cards, each recomposed by a test that fails on drift. One of those tests goes further and fails on a committed card that no rule placement produced, so a card orphaned by a spec change cannot sit here being read. CI re-renders the repository and requires a zero diff, unscoped, because a card written somewhere nobody predicted still has to be committed. Only baseline has no gate pointing at this repository: it is the human escape valve, and a gate that ran it would be moving the ratchet.
The feedback loop is live in this working copy rather than in a clone. A PostToolUse hook runs check after every code edit and blocks the edit on a red rule, a second hook fires the warm server's arch_check with the diff tripwire armed, and a registered MCP server answers architecture queries in the session. That wiring is machine-local, so what a clone gets is hooks/: both wrappers, the config snippet, the launcher that keeps a warm server from holding this repository's build outputs, and what to paste where. The contract those wrappers implement does not rest on anyone's local setup. Tests execute them as real child processes across the whole exit-code mapping, and CI installs the packed global tool, walks the README quickstart end to end, and scaffolds the spec project the derive_spec recipe publishes — PackageReference and all — out of that same pack, once on a repository that pins package versions per project and again on one that manages them centrally, holding it green on conforming code, red on a violation, and red again when the one line that keeps a single csproj valid on both is deleted. The repository is the evidence for the product's claim: rules that cannot drift, prose that cannot rot.
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 · 76 lines · 4,949 tokens per session scan A 11877a58f621
loadbearing AGENTS.md is an instructions file published in the GitHub repository andypgray/loadbearing (0 stars, last pushed 8d ago), licensed MIT. It adds 4,949 tokens to every session, about $0.0247 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
truecourse CLAUDE.md
Claude Code instructions for truecourse-ai/truecourse, covering truecourse — claude instructions, key files to keep updated, project layout, development commands and storage.
truecourse AGENTS.md
AGENTS.md instructions for truecourse-ai/truecourse, covering truecourse — codex instructions, key files to keep updated, project layout, development commands and rules.
PhpCodeArcheology CLAUDE.md
Instructions for PhpCodeArcheology/PhpCodeArcheology, covering claude.md — phpcodearcheology, project overview, language, tech stack and project structure.
drift copilot-instructions.md
Instructions for mick-gsk/drift, covering drift — verbindliche arbeitsgrundlage für alle agenten, primärmodus bei prompt-engineering, zuerst das richtige primitive wählen, nicht verhandelbare schärferegeln and mindestvertrag für jeden guten prompt.
drift drift-push-gates.instructions.md
Nutze diese Instruction, wenn ein Commit, ein Push oder eine Release-Vorbereitung im Drift-Repo ansteht. Sie ist die autoritative Gate-Quelle fuer Pre-Push-Anforderungen, Bypaesse und lokale CI-Pflichten.
drift drift-quality-workflow.instructions.md
Nutze diese Instruction bei nicht-trivialen Aenderungen, Reviews oder Fix-Loops im Drift-Repo. Fokus: adversarialer Gegenreview, iterativer Fix-Loop, expliziter Freigabestatus und Human-in-the-Loop vor Merge.