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/8dionysus/aoa-evals/agents-mdgit clone --depth 1 https://github.com/8Dionysus/aoa-evalsWrote 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/8dionysus/aoa-evals/agents-md)<a href="https://agentmods.dev/instructions/8dionysus/aoa-evals/agents-md"><img src="https://agentmods.dev/badge/instructions/8dionysus/aoa-evals/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 5.1 | $0.01124 | $0.01124 |
| Opus 5 | $0.00562 | $0.00562 |
| Sonnet 5 | $0.00225 | $0.00225 |
| Haiku 4.5 | $0.00112 | $0.00112 |
Grade A, and why
aoa-evals 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 today.
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 — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Applies to
This root card applies to the whole aoa-evals repository unless a nearer
AGENTS.md narrows the touched lane.
Role
aoa-evals is the bounded proof canon of AoA. It owns eval bundle wording,
claim framing, verdict shape, categories, baselines, reports, comparisons,
caveats, generated proof readers, and doctrine about what an eval does and does
not prove.
Claim limit: an eval proves only the bounded claim supported by its source bundle, evidence, verdict logic, and validation route.
This repository also owns eval-local statistical declarations under stats/,
the admitted aoa-evals-skills callable procedure under skills/, and the
owner-local semantic capability tree under capabilities/. Those ports do not
absorb runtime DAG state, shared skill or technique meaning, role policy,
routing, playbooks, durable memory, cross-owner statistics, dashboards, or
private benchmark truth.
Routes
Follow only the route relevant to the touched source:
| Need | Owner route |
|---|---|
| source proof meaning | evals/**/EVAL.md and evals/**/eval.yaml |
| human repository orientation | README.md |
| proof authority class | docs/architecture/PROOF_TOPOLOGY.md and docs/architecture/AGENT_INDEX.md |
| proof operation | mechanics/README.md, then the affected parent, part, and source bundle |
| generated reader change | authored source, owning builder, generated reader, and its validator |
| audit or review | AUDIT.md, then the affected proof surface |
| direction | ROADMAP.md |
| local memory candidate | memo/AGENTS.md and memo/PORT.yaml; durable reviewed memory returns to aoa-memo |
| local statistics | stats/AGENTS.md and stats/port.manifest.json |
| callable proof procedure | skills/AGENTS.md, skills/port.manifest.json, then skills/aoa-evals-skills/SKILL.md |
| capability discovery | capabilities/AGENTS.md, capabilities/port.manifest.json, then the affected family source |
| exact checks | nearest VALIDATION.md; named repository lanes remain in docs/validation/validation_lanes.json |
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.
- today Changed · -107 lines · -1,328 tokens per session c24483779a9a
- 5d ago First seen · 215 lines · 2,452 tokens per session scan A 03bbffa411b8
aoa-evals AGENTS.md is an instructions file published in the GitHub repository 8Dionysus/aoa-evals (2 stars, last pushed today), licensed Apache-2.0. It adds 1,124 tokens to every session, about $0.0056 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
mcp-observatory AGENTS.md
AGENTS.md instructions for KryptosAI/mcp-observatory, covering agent guide for mcp observatory, high-leverage agent tasks, avoid, useful commands and bot maintainer contribution path.
evolve AGENTS.md
AGENTS.md instructions for bitwise-media-group/evolve, covering agent orientation, what this is, layout, commands (cmd/evolve/) and packages (internal/).
rhesis AGENTS.md
AGENTS.md instructions for rhesis-ai/rhesis, covering rhesis project rules, answering, code comments, technology stack and python imports.
hotato AGENTS.md
AGENTS.md instructions for attenlabs/hotato, covering agents.md, the core loop (run this first), what it is, precondition (check this first) and other ways in (friction order).
rhesis CLAUDE.md
Claude Code instructions for rhesis-ai/rhesis, a project described as: The collaboration layer for AI teams: domain experts annotate and review agent behavior, engineers improve the agent from what they find.
proctor AGENTS.md
Instructions for OrionArchitekton/proctor, covering agents.md — proctor, repo role, layout, start here and validation.