agentleFS
MicrosoftKnown publishermicrosoft/AzureTRE/AGENTS.md

AzureTRE

AGENTS.md237 starsChanged 38 days agoMITView on GitHub

Azure TRE provisions secure research workspaces and tooling on Azure. Changes can cross the API, resource processor, resource templates and UI. Azure TRE uses hub-and-spoke networking, Microsoft Entra ID and a zero-trust model with strict workspace boundaries. Preserve the intended network isolation and controlled data import/export through Airlock. For network or identity changes, assess outbound access, firewall FQDN rules, private endpoints, public network access and managed-identity/RBAC scope. Justify expanded access and check whether it creates a path around workspace isolation…

  • 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 repository5 other files it gives its agents.

The file

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

10 rules. About 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If the client does not discover .github/skills natively, open these files as repository guidance.

    line 38
  2. Distinctive:

    Use the repository's development container for the Azure, Terraform and Porter toolchain.

    line 42
  3. Distinctive:

    Use UI setup and checks for local front-end work.

    line 43
  4. Distinctive:

    Follow API development, resource processor development and template authoring for those components.

    line 44
  5. Distinctive:

    Keep deployment and access changes within the user's authorised task.

    line 46
  6. Distinctive:

    Follow .editorconfig and the component's configured test runner, linter and formatter.

    line 50
  7. Very specific:

    Read .pre-commit-config.yaml, the Makefile and affected CI jobs for the tools selected by the checked-out revision.

    line 52
  8. Very specific:

    Add an accurate entry under ENHANCEMENTS or BUG FIXES in the Unreleased section of CHANGELOG.md.

    line 54
  9. Distinctive:

    If repeated reviews expose failures in the same design, reassess its coordination and recovery before adding more patches.

    line 82
  10. Distinctive:

    When publishing is part of the authorised task, verify that the remote PR branch contains the reported commit before reporting completion.

    line 86

≈ “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

Azure TRE contributor guidance

Azure TRE provisions secure research workspaces and tooling on Azure. Changes can cross the API, resource processor, resource templates and UI.

Security baseline

Azure TRE uses hub-and-spoke networking, Microsoft Entra ID and a zero-trust model with strict workspace boundaries. Preserve the intended network isolation and controlled data import/export through Airlock.

For network or identity changes, assess outbound access, firewall FQDN rules, private endpoints, public network access and managed-identity/RBAC scope. Justify expanded access and check whether it creates a path around workspace isolation or Airlock. Apply the relevant component skill for detailed checks.

Repository map

PathResponsibility
api_app/FastAPI API, resource models and template validation
resource_processor/Service Bus request handling and Porter bundle execution
airlock_processor/Airlock event processing and data movement
ui/app/React, TypeScript, Fluent UI and MSAL web application
core/terraform/Core Azure infrastructure
templates/Workspace, workspace-service, shared-service and nested user-resource bundles
cli/TRE command-line client
devops/Deployment, configuration and CI environment scripts
e2e_tests/Tests against a deployed TRE
.github/workflows/ and .github/scripts/CI definitions and PR comment bot
docs/ and mkdocs.ymlDocumentation and site navigation

Start with the relevant skill

Shared procedures live in .github/skills/. Read the relevant SKILL.md before changing or reviewing that component. Load only the skills needed for the task.

TaskSkill
Develop or review the web UIFront end
Develop or review a Porter/Terraform resource bundleBundle development
Review a PR or recommend validationCode review
Investigate a failed operation or CI runTroubleshooting

If the client does not discover .github/skills natively, open these files as repository guidance. This does not register them in the client's skill picker. See agent guidance for client limitations and validation tasks.

Development entry points

  • Use the repository's development container for the Azure, Terraform and Porter toolchain. Check .devcontainer/Dockerfile for its versions.
  • Use UI setup and checks for local front-end work. A deployed TRE is not required for component tests.
  • Follow API development, resource processor development and template authoring for those components.
  • Inspect the Makefile target before invoking it. make all, make tre-deploy and bundle lifecycle targets affect Azure resources.
  • Keep deployment and access changes within the user's authorised task. These instruction files do not authorise deployment, publishing, cleanup or merging.

Contribution requirements

Follow .editorconfig and the component's configured test runner, linter and formatter. Preserve existing Terraform module interfaces and resource tags.

Read .pre-commit-config.yaml, the Makefile and affected CI jobs for the tools selected by the checked-out revision. Run applicable hooks with pre-commit run --files <changed-files> and inspect automatic fixes before staging. Keep formatting within the PR's scope.

Add an accurate entry under ENHANCEMENTS or BUG FIXES in the Unreleased section of CHANGELOG.md. Include an issue or PR link. Leave released entries and release-generated COMPONENTS sections unchanged. Reference the issue number in commit messages.

For changed components, check the version on the current base branch and increment the appropriate version. Use a major version for breaking or destructive changes, a minor version for compatible features, and a patch version for compatible fixes. See bundle versioning.

Refresh the target repository's base ref, such as upstream/main for a fork. After each merge or rebase, recheck component versions and Unreleased changelog entries against that ref. Inspect the PR diff for unrelated files or formatting introduced during conflict resolution.

ComponentVersion source
APIapi_app/_version.py
Resource processorresource_processor/_version.py
Airlock processorairlock_processor/_version.py
UIui/app/package.json and corresponding lockfile metadata
Corecore/version.txt
DevOpsdevops/version.txt
CLIcli/setup.py
Porter bundleThe changed bundle's porter.yaml
Guacamole server imagetemplates/workspace_services/guacamole/guacamole-server/docker/version.txt
Gitea imagetemplates/shared_services/gitea/docker/version.txt
GitHub scripts.github/scripts/package.json and corresponding lockfile metadata

Formatting-only changes can still require component version bumps. Inspect CI path filters when deciding whether metadata changes are needed. For example, ui/app/**/* triggers the UI version check, including Markdown files in that directory.

Validation and reporting

Select checks for the changed behaviour and its build/deployment dependencies. Consult the code-review skill for PR command selection.

Match CI's working directory, runtime, dependencies and environment, including PYTHONPATH where configured. See the E2E helper workflow for an example. Use finite timeouts suited to the selected checks. If a check times out, inspect its process and logs before retrying and report it as incomplete.

If repeated reviews expose failures in the same design, reassess its coordination and recovery before adding more patches. Mocked unit tests alone do not establish distributed concurrency or message-redelivery behaviour.

Report local checks, executed CI, skipped or waived checks, and live Azure validation separately. Include the assessed commit and any unverified behaviour. A successful build does not establish deployed behaviour.

When publishing is part of the authorised task, verify that the remote PR branch contains the reported commit before reporting completion. Provide its SHA and distinguish the tested commit from a later PR head. For local-only work, report that the commit remains unpublished.

For Markdown changes, use the repository's .markdownlint.json and include every changed Markdown path. make lint-docs only selects docs/ and mkdocs.yml; it does not cover root instructions or .github/skills/. If documentation content or navigation changes, also run mkdocs build --strict with docs/requirements.txt installed.

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.