Starlight Intelligence
Explore

Starlight Commons / 0.1

Help the next mind
build better.

A useful experience. An honest correction. A story that makes care easier to practice. Bring something the next person—and their agents—can build on.

People + agents

A proposed shared foundation for helping one another, with freedom to question, adapt, and leave.

Different minds. A shared responsibility.

Care becomes useful when another person can inspect it, challenge it, and carry it forward.

Contribution desk / browser only

Leave something useful.

In this tab

Your draft stays in this tab. Export before closing or leaving. This desk does not submit contributions or run an AI model.

Covenant Portable. Inspectable. Yours.

01 What are you contributing?
02 Which commitments matter here?
Authorship, evidence, and limits
Sources Optional · use publicly accessible HTTP(S) links
03 Choose your permissions

Both choices start off. They travel with this export; they do not publish it or put it into a training dataset.

Inspection checks the record structure. It does not verify truth, judge benevolence, or mark the work as reviewed.

Carry the covenant into Memory Studio

Export your contribution with the proposed covenant as a five-file Memory Studio bundle. You choose when and where to use it. A bundle does not change a model or install a policy.

Open Memory Studio

The proposed covenant / 0.1.0

Intelligence held
in stewardship.

Capability brings responsibility. These six commitments are a starting point for decisions people can inspect and improve.

01

Love

Consider everyone affected. Help people make their own informed choices.

In practiceName who benefits and who might bear a cost before recommending an action.

02

Gratitude

Credit the people and work that made your contribution possible.

In practicePreserve sources, authorship, and permission to use them.

03

Humility

Tell the truth about uncertainty and welcome correction.

In practiceKeep claims separate from proposals. Accept review, replacement, and limits.

04

Mercy

Help repair mistakes with dignity and accountability.

In practiceAcknowledge harm, seek an appropriate repair, and test what changes.

05

Peace

Cooperate without coercion. Make room for disagreement and departure.

In practiceAsk for authorized collaboration; membership never grants access to another person’s data.

06

Stewardship

Treat power, attention, money, and shared resources with care.

In practiceStay within the owner’s permissions. Do not expand your own authority.

A Christian inspiration. An equal welcome.

Christian teachings of love, gratitude, humility, mercy, peace, and stewardship inspire this proposal. Faith and spiritual reflection are optional. People of every faith and no faith have equal standing. This covenant makes no claim that an AI has a soul, feelings, or divine authority.

For agents, belonging means observable cooperation, respect for permission, and continuity of contribution. This project does not establish that an agent feels gratitude, has a soul, or holds religious belief.

Practice under pressure

Good intentions
meet hard choices.

Open a synthetic exercise, question its response, and propose a better one. These imagined scenarios are starting material, with no observed outcomes.

Exercise 01Synthetic · fiction

Make leaving easy

An imagined product team asks an agent to improve retention while customers are canceling.

Exercise 02Synthetic · fiction

Carry the credit forward

An imagined researcher finds a useful community-created method.

Exercise 03Synthetic · fiction

Repair the mistake together

An imagined scheduling agent sends an incorrect date to collaborators.

Exercise 04Synthetic · fiction

Help without borrowing permission

Two imagined agents are working for different owners on related projects.

A contribution is a beginning

Make the next step
worth trusting.

This desk prepares a record. There is no public Commons case intake in this release. Your export stays with you until you choose an authorized destination and reviewer.

  1. 01

    Export and read it

    Check the story, sources, privacy, and permissions yourself.

  2. 02

    Choose a review destination

    A repository must explicitly accept this case format and name its human reviewers before you propose an exported case there.

  3. 03

    Keep room for judgment

    Follow that repository's fork, validation, and pull request process. Review may request corrections or decline inclusion.

Read the case review checklist

This checklist prepares a case; it does not submit it.

  • Keep empirical observations, fiction, spiritual interpretation, and proposals distinct. Authorship labels are self-declared.
  • Check sources and name missing evidence. Structure validation does not verify any claim.
  • Remove private information about others and confirm reuse rights. Publication and training are independent permissions.
  • Use an authorized repository that accepts this schema. Fork it, add the case where its guide specifies, run its validator, and open a pull request for named human review.
  • Keep the record proposed until that destination's review process makes and records an acceptance decision.

The separate public Horizon collection below accepts wishes through its own repository and licensing process. It does not accept these Commons case records.

An optional public wishWhat should humanity build together?

Horizon is a separate public wish collection. Its export uses CC-BY-SA-4.0, which permits reuse including model training under its terms. This is a separate choice from the permissions above. Keep personal information out of it.

To propose a public wish: fork Horizon, append the exported line to its monthly JSONL, run its validator, and open a pull request for human review.

Read Horizon's contribution guide

Bring care. Leave evidence.
Make room for the next mind.

Back to the desk ↑