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/dosco/aithy/agents-mdgit clone --depth 1 https://github.com/dosco/aithyWhat 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 | $0.01311 | $0.01311 |
| Opus 5 | $0.00656 | $0.00656 |
| Sonnet 5 | $0.00262 | $0.00262 |
| Haiku 4.5 | $0.00131 | $0.00131 |
Grade A, and why
aithy 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.
How it starts
The opening of the file, as written. The whole thing — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Aithy Agent Notes
Code Organization
- Keep code clean, simple, and boring.
- Put separate subsystems in separate files and folders.
- Avoid mixed-responsibility modules. Startup wiring belongs in
src/main.ts; behavior belongs in subsystem files. - Keep files short. The repo enforces a 500-line hard limit with
bun run check:lines; prefer smaller files before getting near that limit. - Follow existing patterns before adding abstractions. Add an abstraction only when it removes real duplication or isolates a clear subsystem boundary.
- When adding or materially changing user-facing features, update
README.mdin the same change so setup, security boundaries, and product capabilities stay accurate. - After upgrading
@ax-llm/ax, manually refresh the project-local Claude skills withbunx @ax-llm/ax setup-claude. Keep this manual rather than adding Ax totrustedDependencies, and remove skill directories that the upstream CLI no longer owns.
State And Sessions
- Runtime state is stored under
~/.config/aithy/<bot id>/by default. - The packaged bot id defaults to
default. - SQLite state lives at
~/.config/aithy/<bot id>/state.db. schema_migrationsis scoped by subsystem so session and soul migrations can each start at version1without colliding.- Session history is stored in
session_items. - Processed soul data is stored in
metadataunder thesoul.mdkey askey,value, and optionalhash. - Session transcripts are durable conversation memory.
- We do not persist Ax agent runtime state across turns. Each turn builds a fresh agent from the transcript; if the agent asks a clarification, the question is recorded in the transcript and the next user message is processed as a new turn (the agent re-derives any context from history).
- Responder learning persists only a bounded Ax playbook snapshot in
metadataunderplaybook.responder(64 KiB maximum). New agents apply the cached snapshot after soul guidance; noAxAgentStateis persisted. - Ax v23 can resume a clarification from
AxAgentState, but Aithy intentionally defers that path. A future implementation would need a metadata snapshot with a short TTL and strict size cap; persisted runtime bindings can contain stale fetched pages or sandbox-dependent values, so transcript reconstruction remains the safer continuation contract for now. - Persisted sessions are logical bot sessions. Live Microsandbox VMs are process resources and are recreated after restart.
bun run debuglists SQLite tables.bun run debug <table>inspects a table, andbun run debug <table> <id>looks up a row by the table's primary lookup column.
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.
- 2d ago First seen · 61 lines · 1,311 tokens per session scan A 5f98228bba5d
aithy AGENTS.md is an instructions file published in the GitHub repository dosco/aithy (107 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 1,311 tokens to every session, about $0.0066 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 2026-08-30.
Other instructions, from other repositories
personal-model AGENTS.md
Instructions for Intuition-Lab/personal-model, covering agents.md, runtime boundary, commands, pipeline and documentation map.
localvoxtral AGENTS.md
Instructions for T0mSIlver/localvoxtral, covering localvoxtral — agent guide, build & test — read this first on a non-mac dev box, proof culture — non-negotiable, test tiers — the short version and ci / shipping.
cai CLAUDE.md
Instructions for cai-layer/cai, covering claude.md, what is cai?, design system, build & run and dev.
Meldwork AGENTS.md
Instructions for Ryder-MHumble/Meldwork, covering meldwork development rules, working style, product boundaries, frontend and desktop and verification.
codesearch AGENTS.md
Instructions for flupkede/codesearch, covering agents.md — codesearch, current state, implemented features, open todos and ⚠️ branching & pr workflow (read first).
foundry-local cpp-formatting.instructions.md
Use when writing or modifying C++ source/header files. Covers vertical whitespace conventions for readability.