SaudSatopaySaudSatopay/bullseye/AGENTS.md

bullseye

AGENTS.md4 starsChanged 24 days agoMITView on GitHub

This repository is a playbook for coding agents that build hackathon projects. It is a set of Agent Skills (skills/<name>/SKILL.md), Claude Code subagents (agents/), and the docs and scripts behind them. Two kinds of agent read this file; find yours. Read CONTRIBUTING.md. In short:

Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,542 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository13 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- brief-decoderSkill
- build-sprintSkill
- bullseyeSkill
- demo-videoSkill
- finale-pitchSkill
- hackathon-trackerSkill
- idea-roundSkill
- judge-panelSkill
- missileSkill
- pitch-deckSkill
- prove-itSkill
- submission-packSkill

The file

We split this AGENTS.md into the rules it gives an agent

5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Read skills/bullseye/SKILL.md first and follow it.
line 8
Very specific:
Read CONTRIBUTING.md. In short:
line 32
Very specific:
Every new rule carries a field note saying where it was learned: > Field note (Project · Event): what happened and the lesson.
line 35
Distinctive:
Do not copy third-party skills into this repository.
line 38
Very specific:
Before finishing: node scripts/validate.mjs and node --test must both pass.
line 39

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

AGENTS.md

AGENTS.md · BULLSEYE

This repository is a playbook for coding agents that build hackathon projects. It is a set of Agent Skills (skills/<name>/SKILL.md), Claude Code subagents (agents/), and the docs and scripts behind them. Two kinds of agent read this file; find yours.

If a user asked you to build or run a hackathon entry with BULLSEYE

Work in the user's project directory, never inside this repository. If this repository was cloned into the project as .bullseye/, every path below is relative to .bullseye/.
Read skills/bullseye/SKILL.md first and follow it. It is the flight director: it creates BULLSEYE.md in the user's project, decides the phase, and tells you which skill to read next.
When a skill is not installed in your agent, read its folder here and follow it the same way: skills/<name>/SKILL.md first, then the references/, assets/ and scripts/ files it names, resolving every relative path from that skill's folder. Run a bundled script by its path, for example node .bullseye/skills/prove-it/scripts/claims-check.mjs claims.json.
The skills, in the order a hackathon uses them:

Skill	Phase	Read when
`bullseye`	All	Always first; it sequences everything else
`brief-decoder`	Intake	A new event, problem statement, rules page or rubric arrives
`idea-round`	Idea gate	The event has an idea, proposal or concept submission
`pitch-deck`	Any document	A deck, concept brief, proposal or organiser template must be produced
`build-sprint`	Build	Writing the product
`missile`	Polish	Raising product craft with screenshots, measurements and a separate critic
`prove-it`	Proof	Any number, claim, benchmark or evaluation
`judge-panel`	Judging	Simulating judges or rehearsing Q&A
`demo-video`	Package	The demo video or a launch film
`submission-pack`	Package	README for judges, form answers, final sweep, submitting
`finale-pitch`	Finale	Live pitch, live demo, Q&A, surprise challenges
`hackathon-tracker`	Portfolio	Tracking several events and deadlines

Subagents in agents/ work only in Claude Code. Elsewhere, when a skill calls for a fresh-context critic or judge, open a new session, paste the prompt from that skill's assets/ with the evidence it lists, and bring back only the verdict.
Respect the law in skills/bullseye/references/law.md and the event's own rules, which win over anything here.

If you are improving BULLSEYE itself

Read CONTRIBUTING.md. In short:

Every skill follows the Agent Skills specification: name equals the folder name (lowercase, hyphens), description says what it does and when to use it (at most 1024 characters; all descriptions together under about 8,000, Codex's listing budget), the SKILL.md body stays under 500 lines, and depth goes into references/ one level deep.
Every new rule carries a field note saying where it was learned: > Field note (Project · Event): what happened and the lesson. No invented results; no personal data; no secrets.
House style: plain, imperative, concrete. No em dash characters anywhere (the validator fails on them); no emoji; no hype words.
Scripts: Node 20+ ESM with zero dependencies where possible, or Python 3.10+ with optional libraries checked at import and an install hint. Every script has --help, clear errors, and a test in tests/ (node:test).
Do not copy third-party skills into this repository; reference them with install commands.
Before finishing: node scripts/validate.mjs and node --test must both pass.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free