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/thunderbird/thunderbolt/agents-mdgit clone --depth 1 https://github.com/thunderbird/thunderboltWhat 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 | $0.06549 | $0.06549 |
| Opus 5 | $0.03275 | $0.03275 |
| Sonnet 5 | $0.01310 | $0.01310 |
| Haiku 4.5 | $0.00655 | $0.00655 |
Grade A, and why
thunderbolt 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 — 272 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Core Principles
- Bias towards tasteful simplicity - favor elegant, readable, maintainable solutions that add minimal complexity. Avoid over-engineering, premature optimization, and defensive coding patterns that obscure intent.
- Always implement proper, architectural solutions - no shortcuts, hacky fixes, or temporary workarounds. Research best practices when needed.
- Prefer optimistic code over defensive code - let errors surface loudly during development rather than wrapping everything in if-checks and try/catch blocks. Handle errors architecturally at higher levels (e.g., error handling middleware).
- Deletes (soft vs hard)
- Frontend: Never hard delete. Always soft delete data (set
deletedAt; call APIs that update rather than permanently remove). The only exception is flows that explicitly perform account or device removal (e.g. “Delete account”), which call backend endpoints that hard delete by design. - Backend: Prefer soft deletes—set
deletedAtand filter out soft-deleted records in queries. Use hard delete only when required: e.g. account deletion (user and related data), PowerSync delete operations, or other cases where permanent removal is by design.
- Frontend: Never hard delete. Always soft delete data (set
- Question and recommend alternatives - your goal is better outcomes, not blind execution. Stop and ask for input when appropriate.
TypeScript & Code Style
- Never use
anyin TypeScript - Prefer
typeoverinterface - Prefer arrow functions over
functionkeyword - Prefer
constoverlet- create helper functions with early return instead of settingletvariables inside conditionals - Use camelCase for const and variable names
- Prefer early return over long if statements and nested code
- Use direct imports:
useEffectnotReact.useEffect - Prefer top-level imports over inline/dynamic imports (
await import(...)) when no circular dependency exists - Prefer async/await over .then/.catch
- Add JSDoc comments to new utility functions
- Only comment non-obvious code - avoid useless comments like "// Save data collection mutation" before
saveDataCollection() - Loosely prefer one React component per file
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 · +2 lines · +237 tokens per session 16f8be34778b
- 2d ago First seen · 270 lines · 6,312 tokens per session scan A 85f17f9fdb34
thunderbolt AGENTS.md is an instructions file published in the GitHub repository thunderbird/thunderbolt (4,762 stars, last pushed today), licensed MPL-2.0. It adds 6,549 tokens to every session, about $0.0327 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
ollama AGENTS.md
AGENTS.md instructions for ollama/ollama, covering agents.md and building.
fastmcp CLAUDE.md
Claude Code instructions for PrefectHQ/fastmcp, covering fastmcp development guidelines, required development workflow, repository structure, core mcp objects and development rules.
ollama CLAUDE.md
Claude Code instructions for ollama/ollama: See AGENTS.md for the shared agent instructions for this repository.
plano CLAUDE.md
Instructions for katanemo/plano, covering claude.md, build & test commands, rust — wasm plugins (must target wasm32-wasip1), rust — brightstaff binary (native target) and rust — tests, format, lint.
oumi AGENTS.md
AGENTS.md instructions for oumi-ai/oumi, covering agent instructions, publication policy and working with contributions.
graphiti AGENTS.md
Instructions for getzep/graphiti, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.