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/hackbyrd/orbital-express/agents-mdgit clone --depth 1 https://github.com/Hackbyrd/orbital-expressWhat 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 | $0.02415 | $0.02415 |
| Opus 5 | $0.01208 | $0.01208 |
| Sonnet 5 | $0.00483 | $0.00483 |
| Haiku 4.5 | $0.00242 | $0.00242 |
Grade A, and why
orbital-express 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 — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Agent guide for Orbital-Express
Canonical, tool-agnostic instructions for any AI agent working in this repo (Claude Code reads it via CLAUDE.md; Cursor/Codex/Copilot read AGENTS.md directly). Keep this file concise — it points to the deep docs rather than duplicating them.
What this is
Orbital-Express is an opinionated Express.js + Sequelize (PostgreSQL) framework for building really good backend APIs. Feature-folder architecture (Django + Rails hybrid): everything for a feature — model, routes, controller, actions, tasks, tests, i18n, mailers — lives in one folder under app/. A background-job system (Bull/Redis) and Socket.IO layer sit alongside.
Where the knowledge lives (read these for depth)
README.md— the full human onboarding doc. Deep explanations of every folder, pattern, and flow. Read the relevant section before non-trivial work.docs/conventions.txt— the terse, authoritative rulebook (naming, file structure, DB, auth, sockets, etc.). When a rule is ambiguous, this wins.database/schema.sql— documentation of every table (not executed). The column-order / naming template is at the top.docs/auth-migration.md— the access+refresh auth design and status.docs/google-oauth-setup.md— how to provision Google OAuth for "Sign in with Google".docs/workflow.md— the feature-development lifecycle (Path A new feature, Path B modify existing). The high-level map; thecreate-feature/modify-featureskills are the step-by-step..claude/skills/— step-by-step playbooks for common tasks (see "Skills" below). Prefer running these.
⚠️ Historical/outdated:
docs/tests.txtpredates the current setup (it mentions mocha). Ignore it —write-tests, the README, andconventions.txtare authoritative.
Golden rules (non-negotiable)
- Plan and get sign-off before edits or scaffolding, for new and existing features. Then use the generator:
yarn gen Feature,yarn gen Feature -a V1Action,yarn gen Feature -t V1Task,yarn gen Feature -m Mailer. Never hand-create feature files when scaffolding exists. Generator names/targets are validated before writes; existing source/test/mailer targets abort without writes; new source/test/templates are exclusive-created; indexes are deduplicated and sorted. Remove placeholder actions/tasks withyarn del(neverrm), then runyarn repair Feature --dry-runbeforeyarn repair Feature. 1b. Everyapp/<Feature>has the complete standard structure. There are no model-only/table-only exceptions. Extra domain files are allowed. Repair creates only missing files from generator templates, never overwrites existing files, restores missing action/task tests, creates README markers for otherwise-empty standard directories, and reports ambiguous route/controller/worker wiring instead of inventing policy. - Install exact versions only:
yarn add <pkg> --exact(and--devfor dev deps). Never^/~. - JS file structure (every
.jsfile): header comment →'use strict'→ env → built-ins → third-party → services → helpers → models → queues (queue.get('XQueue')instances, right after models; the queue service is required up in services) → module-level consts →module.exports(before the methods) → method definitions. Imports ordered by increasing length, plain requires before destructured. Close every function with// END <name>. (README: "JavaScript File Structure".) - Naming: actions
V{version}{Action}[By{Role}][On{Device}]; tasks appendTask; feature folders singular PascalCase; controllers plural, actions singular; constantsUPPER_CASE; booleans start withis/has/can/does; FK columns<entity>Id→ PascalCase plural table. 4b. No magic strings. Any string literal used — or likely to be used — in more than one place (statuses, types, roles, locales, enum-like values) lives once inhelpers/constants.jsand is referenced, e.g.LOCALE.ENnot'en'. Use theadd-constantskill. (Migrations stay literal — frozen snapshots.) - HTTP: POST and GET only. Use
req.args(neverreq.body/req.query). Responses are flat:{ status, success: true, ...payload }— nodatanesting. Status:200default,201on create,202on background-job handoff. Route URLs are lowercase, no separators, even multi-word (V1LogoutAll→/v1/users/logoutall, notlogout_all/-all/camelCase). - Errors: HTTP actions return
errorResponse(req, ERROR_CODES.X, ...); tasks & socket-invoked actionsthrow. Never return a 500 manually — let it propagate tomiddleware/error.js. - Models: UUID v7 PKs (
defaultValue: () => uuidv7()— require{ v7: uuidv7 }from'uuid';validate: { isUUID: 7 });paranoid: truesoft-deletes (usescope(null)to bypass); explicit static PascalCase pluraltableName(including irregular plurals); always index FKs; carry the owner FK onto every descendant + protect with a composite FK; named indexes use that exact table name in{Table}_{col}_{idx|unique}in BOTH model and migration. - i18n: keys are
NAMESPACE[snake_case]; edit featurelanguages/*.js, then runyarn lang(it compileslocales/and validates keys — required, andyarn testruns it first). - Tests: every action and task has an active test; required counterparts cannot use
describe/it/test.skip,xit,xtest,it.todo, ortest.todo. EveryERROR_CODEin the JSDoc has a test; test who cannot do something; fixtures are baselines you mutate in-test; run with--runInBand. Test location mirrors source: actions →app/<Feature>/tests/integration/, tasks →app/<Feature>/tests/tasks/, global helpers/services →test/helpers/ortest/services/. - One user type = one table (Admin, User, …) — never a single table with a role column. Auth is access token + revocable refresh token; see the
add-auth-user-typeskill. - Docs travel together. The same conventions are documented in
documentation.html,README.md,docs/conventions.txt,AGENTS.md,CLAUDE.md,database/schema.sql, and the.claude/skills/. Editing one is not done until you've reconciled the others that cover the same thing — including the relevant skill(s). Follow thesync-docsskill. (A PostToolUse hook reminds you on every doc edit.) - Convention enforcement is mandatory. Run
yarn conventions:checkand thereview-conventionsskill after every non-trivial change. Root routes, models, workers, and errors auto-discover complete feature folders; actions live inactions/.
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 · 78 lines · 2,415 tokens per session scan A b7afc182c866
orbital-express AGENTS.md is an instructions file published in the GitHub repository Hackbyrd/orbital-express (14 stars, last pushed 13d ago), licensed MIT. It adds 2,415 tokens to every session, about $0.0121 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
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
next.js AGENTS.md
Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.