osdu-spi-stack
AGENTS.md0 starsChanged 42 days agoApache-2.0View on GitHub
Azure-native OSDU deployment. A Python CLI (spi) provisions Azure infra with az + Bicep, bootstraps AKS Automatic, then Flux CD reconciles Kubernetes workloads from this repository. Repo: Azure/osdu-spi-stack. uv is the only tool needed for repo work; it manages Python, dependencies, and every command below. Deploying to Azure also needs az, bicep, kubectl, kubelogin, and flux, which uv run spi check verifies. Directory and file names are self-describing; explore per task rather than from a maintained map. The CLI's command…
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 31,700 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 repository6 other files it gives its agents.
- .github/copilot-instructions.mdCopilot instructions
- code-reviewSkill
- gh-voiceSkill
- osdu-apiSkill
- primeSkill
- resolve-review-threadsSkill
The file
We split this AGENTS.md into the rules it gives an agent
8 rules. About 1 is common, 2 are distinctive, and 5 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 43
Commits and PR titles: follow Conventional Commits (feat, fix, docs, refactor, test, ci, style, chore).
- Very specific: line 53
Prose: no em dashes; use commas, periods, or semicolons.
- Distinctive: line 55
Comments: only where they add something the code cannot say: cross-file coupling, an external contract, or why the obvious approach was not …
- Distinctive: line 57
Keep them to a line or two.
- Very specific: line 58
Transparency: every az and kubectl command that changes Azure or cluster state is shown to the user via a Rich panel before execution.
- Very specific: line 60
Read-only queries backing status, verify, and lock reads run silently.
- Very specific: line 64
Azure only: When reading cloned OSDU service repos, only *-azure/ providers (and shared *-core/) matter.
- Common: line 68
Never commit secrets or hardcode credentials.
≈ “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.
Files
AGENTS.md
Azure-native OSDU deployment. A Python CLI (spi) provisions Azure infra with
az + Bicep, bootstraps AKS Automatic, then Flux CD reconciles Kubernetes
workloads from this repository. Repo: Azure/osdu-spi-stack.
uv is the only tool needed for repo work; it manages Python, dependencies, and
every command below. Deploying to Azure also needs az, bicep, kubectl,
kubelogin, and flux, which uv run spi check verifies.
Repository map
| Path | What lives here |
src/spi/ | The spi CLI (Typer + Rich + Pydantic) |
infra/ | Bicep templates for Azure PaaS provisioning |
software/ | Helm charts, middleware components, OSDU service manifests |
docs/ | Architecture, ADRs (decisions/), design docs, prose STYLE.md |
scripts/, tests/ | Helper scripts and the pytest suite |
Directory and file names are self-describing; explore per task rather than from
a maintained map. The CLI's command surface is defined in src/spi/cli.py.
Working commands
uv sync # install the CLI plus pytest, ruff, ty, pre-commit
uv run pre-commit run --all-files # lint, format, types, and tests in one pass
uv run spi --help # sanity-check the CLI itselfFrom a checkout, invoke the CLI as uv run spi. The bare spi commands in
README.md assume the released wheel installed via uv tool install; do not
expect spi on PATH here.
pre-commit lints, formats, type-checks (ty), and tests in one pass. The
ruff hooks auto-fix what they can and fail the run so you re-stage the
corrected files. Run it before every PR; run individual tools from
.pre-commit-config.yaml when iterating on one kind of failure.
Conventions
- Commits and PR titles follow Conventional Commits
(
feat,fix,docs,refactor,test,ci,style,chore). Squash-merge uses the PR title as the commit subject release-please versions from (featminor, breaking!major, anything else patch); thepr-titlecheck enforces it. PR descriptions follow the shape inCONTRIBUTING.md: why first, then what changed, then notes on anything a reader would stop and question. No validation checklist; CI results are already on the pull request. - Branches are named
<type>/<short-name>, for examplefeat/add-redis-component. - Code: ruff line length 100 with import sorting; keep
tyclean. - Prose: no em dashes; use commas, periods, or semicolons. Files under
docs/additionally followdocs/STYLE.md. - Comments only where they add something the code cannot say: cross-file coupling, an external contract, or why the obvious approach was not taken. Keep them to a line or two; delete comments that restate the code.
- Transparency: every
azandkubectlcommand that changes Azure or cluster state is shown to the user via a Rich panel before execution. Read-only queries backingstatus,verify, and lock reads run silently.
Scope notes
- Azure only. When reading cloned OSDU service repos, only
*-azure/providers (and shared*-core/) matter; skip*-aws/,*-gc/,*-ibm/,*-core-plus/. - No stored credentials. Workload Identity is the only data-plane path; local (key/SAS) authentication is disabled on Cosmos and Service Bus (ADR-023). Never commit secrets or hardcode credentials.
- Decisions are records.
docs/decisions/governs the deployment model. Read the record that owns a subject before changing it;docs/design/explains how the subsystems actually work.
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
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.

