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/jetbrains/thinkrail/agents-mdgit clone --depth 1 https://github.com/JetBrains/thinkrailWhat 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 | $0.04503 | $0.04503 |
| Opus 5 | $0.02252 | $0.02252 |
| Sonnet 5 | $0.00901 | $0.00901 |
| Haiku 4.5 | $0.00450 | $0.00450 |
Grade A, and why
thinkrail 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 — 238 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ThinkRail
A ThinkRail-branded desktop-and-mobile client for the pi coding agent. The app is a thin host that
runs pi and bridges it to a rich UI; pi owns models, skills, compaction, cost, and session state.
Canonical specs (read these first):
goal-and-requirements.md— product goal + V1/V2 scopearchitecture.md— top-level architecture, decisions, invariants
Module structure & boundaries (top-priority requirement)
The app is built as a set of clearly bounded modules. This is a primary design requirement, not a nice-to-have — treat it with the same weight as the non-negotiable invariants below.
- Modules are fractal. The boundary rule applies at every level: each package is a module, and the
directories inside a package (
packages/server/src/agent/,apps/web/src/transport/, …) are modules too. A sub-module is a directory with anindex.tsbarrel as its only public surface; siblings import it through that barrel, never its internals. (Exception: where a barrel would defeat code-splitting or a library's per-file convention — e.g.apps/web/src/panelsandcomponents/ui, which lazy-load Monaco/shiki/xterm — imports stay per-file and the boundary is held by spec + convention.) - Every module has a
SPEC.mdthat states its boundary explicitly: what it owns, what it exposes as its public surface, and what it must not reach into (allowed deps and forbidden deps). The dependency edges between sibling sub-modules live in the parent module'sSPEC.md(a dependency graph), not in each leaf — leaves declare only their own external deps + forbidden reaches. - Boundaries should be covered by tests where practical — a module's public surface and its boundary rules are worth exercising with tests, not just relying on convention. This is a goal, not a hard gate: aim for coverage, but don't block on guaranteeing it everywhere.
- The spec leads the code. A change that moves or blurs a boundary updates the module's
SPEC.mdfirst, then the code and the tests that pin it.
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 · 238 lines · 4,503 tokens per session scan A 868e430dd04d
thinkrail AGENTS.md is an instructions file published in the GitHub repository JetBrains/thinkrail (340 stars, last pushed 2d ago), licensed Apache-2.0. It adds 4,503 tokens to every session, about $0.0225 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
LeAgent AGENTS.md
Instructions for vixues/LeAgent, covering leagent development guidelines, project overview, architecture, execution topology (one kernel, many ingresses) and surface map (where things live).
grix AGENTS.md
Instructions for askie/grix, covering grix agent guide, repository boundaries, required workflows, shared agent configuration and cross-component contracts.
rn-dev-agent AGENTS.md
Instructions for Lykhoyda/rn-dev-agent, covering repository guide for agents, repository map, editing rules, architecture rules and supported node runtimes.
get-fable AGENTS.md
Instructions for imMamdouhaboammar/get-fable, covering get-fable repository instructions, purpose, working contract, canonical lifecycle packs and runtime semantics.
metatron CLAUDE.md
Instructions for kerbelp/metatron, covering claude.md, what metatron is, tech stack (decided — not open for debate), core principle and source of truth (depends on deployment).
AmbyKit AGENTS.md
Instructions for ambystechcom/AmbyKit, covering agents.md — ambykit, what ambykit is, golden rules, project structure and build & test.