agentleFS
vivary-devvivary-dev/vivary-cli/AGENTS.md

vivary-cli

AGENTS.md4 starsChanged 4 months agoMITView on GitHub

Read ENGINEERING.md. It is the governing engineering policy. Older packet and runtime instructions cannot restore heavier process defaults. Read the current frontier and the owning packet. The execution contract owns program metadata and delivery details. Choose the next observable user capability and implement one coherent slice through the real application. Use the workspace and execution environment authorized for the task. Reuse the current checkout and handoff when provided. Inspect live Git and preserve dirty work before editing. Keep one writer…

  • 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 repository3 other files it gives its agents.

The file

Show its rules and text

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

13 rules. About 0 are common, 12 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read ENGINEERING.md. It is the governing engineering policy.

    line 3
  2. Distinctive:

    Read the current frontier and the owning packet.

    line 8
  3. Distinctive:

    Use the workspace and execution environment authorized for the task.

    line 13
  4. Distinctive:

    Use existing implementations and tests.

    line 17
  5. Distinctive:

    Use another reviewer when the risk or complexity warrants it.

    line 19
  6. Distinctive:

    Before unusually heavy work, check memory and disk.

    line 21
  7. Very specific:

    Follow CONTRIBUTING.md. Branch from current remote dev, use a typed topic branch, and integrate through a PR with the required CI and …

    line 30
  8. Distinctive:

    Use the user's existing authorization.

    line 39
  9. Distinctive:

    Do not modify the read-only source repositories loam, braincheck, throughline, or flywheel.

    line 44
  10. Distinctive:

    Follow the release workflow when release operations apply.

    line 47
  11. Distinctive:

    Use the source map when work crosses project identity, root custody, runtime, or project writes.

    line 53
  12. Distinctive:

    Read the architecture when a design decision needs it.

    line 55
  13. Distinctive:

    Update durable docs when behavior, architecture, contracts, or lasting constraints change.

    line 59

≈ “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

Vivary agent instructions

Read ENGINEERING.md. It is the governing engineering policy. Older packet and runtime instructions cannot restore heavier process defaults.

Start with the product

Read the current frontier and the owning packet. The execution contract owns program metadata and delivery details. Choose the next observable user capability and implement one coherent slice through the real application.

Use the workspace and execution environment authorized for the task. Reuse the current checkout and handoff when provided. Inspect live Git and preserve dirty work before editing. Keep one writer per shared file.

Use existing implementations and tests. Run the cheapest relevant checks, exercise the actual application, fix failures, and review the diff. Use another reviewer when the risk or complexity warrants it.

Before unusually heavy work, check memory and disk. Keep one heavy job active, bound runaway processes, and clean up task-owned servers, browsers, and children. Normal tests need ordinary timeouts and cleanup.

Invoke high-assurance mode only for a named dangerous failure. Select controls for that risk.

Delivery and authority

Follow CONTRIBUTING.md. Branch from current remote dev, use a typed topic branch, and integrate through a PR with the required CI and review gates. Never push directly to protected dev or legacy prod.

Make small, coherent, reviewed commits. Preserve unrelated work, accepted evidence, and historical budgets. Keep private handoffs, credentials, transcripts, and machine-specific continuity notes outside public source and Git history. Private source hosting does not establish agent-session capture.

Use the user's existing authorization. Complete reversible preparation before requesting any missing approval for pushes, PR creation, merges, publication, account changes, paid calls, scheduled activation, outbound messages, or destructive actions. A material scope or authority change needs alignment before that action. Do not modify the read-only source repositories loam, braincheck, throughline, or flywheel.

Follow the release workflow when release operations apply. Source implementation and passing tests do not establish publication. README.md owns shipped behavior.

Keep knowledge useful

Use the source map when work crosses project identity, root custody, runtime, or project writes. Read the architecture when a design decision needs it. Reuse existing Native owners and installed dependencies. Review relevant deny-lists and advisories before adding packages.

Update durable docs when behavior, architecture, contracts, or lasting constraints change. Give each fact one owner and refresh generated views when their inputs change. Keep the existing handoff concise. Failed or unrun verification remains incomplete regardless of a policy change.

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.