agentleFS
Microsoft AzureKnown publisherAzure/azure-dev/cli/azd/extensions/azure.ai.toolboxes/AGENTS.md

azure-dev / azure.ai.toolboxes

AGENTS.md573 starsChanged 8 months agoMITView on GitHub

Use this file together with cli/azd/AGENTS.md. This guide supplements the root azd instructions with the conventions that are specific to this extension. azure.ai.toolboxes is a first-party azd extension under cli/azd/extensions/azure.ai.toolboxes/. It runs as a separate Go binary and talks to the azd host over gRPC. The user-facing surface is azd ai toolbox <verb> — CRUD plus a connection subgroup for managing the connection-backed tools attached to a Foundry toolbox. Useful places to start: - internal/cmd/: Cobra commands, verb implementations, and…

The file

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

12 rules. About 0 are common, 5 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

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

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

    Use this file together with cli/azd/AGENTS.md.

    line 3
  2. Distinctive:

    If extension work depends on a new azd core change, plan for two PRs:

    line 35
  3. Very specific:

    No toolbox-specific logic inside internal/foundry/: Keep that package to primitives any sibling extension would also need …

    line 61
  4. Very specific:

    Use plain Go errors by default.

    line 74
  5. Distinctive:

    Create a structured error once, as close as possible to the place where you know the final category, code, and suggestion.

    line 90
  6. Very specific:

    If err is already a structured error, usually return it unchanged.

    line 93
  7. Very specific:

    Do not add context with fmt.Errorf("context: %w", err) after err is already structured.

    line 95
  8. Distinctive:

    Prefer the dedicated helpers instead of hand-rolling conversions:

    line 138
  9. Distinctive:

    keep them stable once introduced

    line 153
  10. Very specific:

    After the GitHub release is live, a follow-up PR updates cli/azd/extensions/registry.json so azd users can install the new version.

    line 173
  11. Very specific:

    Do not use log.Fatal or log.Panic for expected failures — return a structured error via exterrors instead.

    line 188
  12. Distinctive:

    Use modern Go 1.26 patterns where they help readability.

    line 207

≈ “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 AI Toolboxes Extension - Agent Instructions

Use this file together with cli/azd/AGENTS.md. This guide supplements the root azd instructions with the conventions that are specific to this extension.

Overview

azure.ai.toolboxes is a first-party azd extension under cli/azd/extensions/azure.ai.toolboxes/. It runs as a separate Go binary and talks to the azd host over gRPC.

The user-facing surface is azd ai toolbox <verb> — CRUD plus a connection subgroup for managing the connection-backed tools attached to a Foundry toolbox.

Useful places to start:

  • internal/cmd/: Cobra commands, verb implementations, and connection subgroup
  • internal/pkg/azure/: FoundryToolboxClient (toolbox CRUD data-plane client)
  • internal/foundry/: Foundry primitives shared with sibling extensions (project-endpoint cascade, credential factory, single-connection lookup)
  • internal/exterrors/: structured error factories and toolbox-specific codes

Build and test

From cli/azd/extensions/azure.ai.toolboxes:

# Build using developer extension (for local development)
azd x build

# Or build using Go directly
go build

If extension work depends on a new azd core change, plan for two PRs:

  1. Land the core change in cli/azd first.
  2. Land the extension change after that, updating this module to the newer azd dependency with go get github.com/azure/azure-dev/cli/azd && go mod tidy.

For local development, draft work, or validating both sides together before the core PR is merged, you may temporarily add:

replace github.com/azure/azure-dev/cli/azd => ../../

That replace points this extension at your local cli/azd checkout instead of the version in go.mod. Do not merge the extension with that replace still present.

Package boundaries

internal/foundry/ holds Foundry data-plane primitives that are currently duplicated from azure.ai.agents and intended to be lifted into a shared module later. Two rules keep that future lift mechanical:

  1. One-way import contract. Files under internal/foundry/ MUST NOT import anything under internal/cmd/. Files under internal/cmd/ may import internal/foundry/... freely.
  2. No toolbox-specific logic inside internal/foundry/. Keep that package to primitives any sibling extension would also need (project-endpoint resolution, single-connection lookup, credential factory). Toolbox-specific concepts (projectConnection, connectionResolver, the FoundryToolboxClient) live in internal/cmd/ and internal/pkg/azure/.

Error handling

This extension uses internal/exterrors so the azd host can show a useful message, attach an optional suggestion, and emit stable telemetry.

Default rule

Use plain Go errors by default. Switch to exterrors.* only when the current code can confidently answer all three of these:

  1. What category should telemetry see?
  2. What stable error code should be recorded?
  3. What suggestion, if any, should the user get?

That usually means:

  • lower-level helpers return fmt.Errorf("context: %w", err)
  • user-facing orchestration code classifies the failure with exterrors.*

In this extension, that classification usually happens in internal/cmd/.

Most important rule

Create a structured error once, as close as possible to the place where you know the final category, code, and suggestion.

If err is already a structured error, usually return it unchanged.

Do not add context with fmt.Errorf("context: %w", err) after err is already structured. During gRPC serialization, azd preserves the structured error's own message/code/category, not the outer wrapper text. If you need extra context, include it in the structured error message when you create it.

Choosing an Error Type

SituationPrefer
Invalid input, manifest, or option combinationexterrors.Validation
Missing environment value, missing resource, unavailable dependencyexterrors.Dependency
Auth or tenant/credential failureexterrors.Auth
User cancellationexterrors.Cancelled
Azure SDK HTTP failureexterrors.ServiceFromAzure
gRPC failure from an azd host Prompt callexterrors.FromPrompt
Unexpected bug or local failure with no better categoryexterrors.Internal

Recommended pattern

func parseInput(path string) error {
    if err := readFile(path); err != nil {
        return fmt.Errorf("read %s: %w", path, err)
    }

    return nil
}

func runCommand() error {
    if err := parseInput("tools.json"); err != nil {
        return exterrors.Validation(
            exterrors.CodeInvalidParameter,
            fmt.Sprintf("invalid input file: %s", err),
            "fix the file and retry",
        )
    }

    return nil
}

Azure and gRPC boundaries

Prefer the dedicated helpers instead of hand-rolling conversions:

  • exterrors.ServiceFromAzure(err, operation) for azcore.ResponseError (data-plane and ARM calls).
  • exterrors.FromPrompt(err, contextMessage) for azdClient.Prompt().* failures.

These helpers keep telemetry and user-facing behavior consistent.

Error codes

Define new codes in internal/exterrors/codes.go.

  • use lowercase snake_case
  • describe the specific failure, not the general category
  • keep them stable once introduced
  • co-locate any new Op* constant with the matching ServiceFromAzure call site

Release preparation

A new extension release ships in two PRs:

PR 1 — Version bump

Bumps the extension to the new version. Touches only:

  • version.txt — new semver string
  • extension.yaml — version: field
  • CHANGELOG.md — new release section at the top

Once merged, the team triggers the CI release pipeline, which builds, signs, and publishes the extension binaries as a GitHub release.

PR 2 — Registry update

After the GitHub release is live, a follow-up PR updates cli/azd/extensions/registry.json so azd users can install the new version. The contents of that file are produced by running azd x publish against the published release artifacts (which computes the artifact URLs and checksums). The resulting PR should contain only the regenerated registry.json entry for the extension, and in some cases updated test snapshots as well.

Output: log vs fmt

Extensions write directly to stdout/stderr — there is no Console abstraction from azd core.

  • fmt.Print* — user-facing output (stdout).
  • log.Print* — developer diagnostics (stderr). Hidden unless --debug is set. Never use log for anything the user needs to see.
  • Do not use log.Fatal or log.Panic for expected failures — return a structured error via exterrors instead.
// ✅ log — internal detail the user doesn't need to see
log.Printf("toolbox show: pending-toolbox read failed for %q: %v", name, err)

// ✅ fmt — user-facing status and results
fmt.Printf("Created toolbox %s at version %s.\n", name, version)

// ❌ fmt used for debug noise — user sees internal details they can't act on
fmt.Printf("Parsed endpoint: host=%s, path=%s\n", host, path)  // use log.Printf

// ❌ log used for user-facing info — user never sees it without --debug
log.Printf("No toolboxes found on project")                    // use fmt.Print*

Other extension conventions

  • Use modern Go 1.26 patterns where they help readability.
  • Reserved azd globals (--output, --no-prompt) are inherited from extCtx, not registered as flags on each verb.
  • Lowercase-normalize --output when reading it from extCtx so downstream branches can compare with == "json".

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.