Add Naming & Data Models for Custom Deployment SDK

[tomerqodo]

Benchmark PR from qodo-benchmark#365

[qodo-code-review]

Code Review by Qodo (Alpha)

🐞 Bugs (0) 📘 Rule Violations (3) 📎 Requirement Gaps (0) 💡 Suggestions (0)

1. `deployment_names` returns wrong values 📘 Rule Violation

Description

• SDKData.deployment_names is documented as returning deployment full_name values, but it
  actually appends deployment.name, making the property’s behavior inconsistent with its own
  docstring.
• This is misleading/self-non-documenting and can cause downstream code generation to use incomplete
  identifiers (deployment-only names) instead of flow/deployment names.
• The mismatch also undermines deterministic output expectations because the returned values are not
  the intended ones.

correctness

Code

src/prefect/_sdk/models.py[R120-129]

+    def deployment_names(self) -> list[str]:
+        """List of all deployment full names (derived from flows).
+
+        Returns names sorted alphabetically for deterministic code generation.
+        """
+        names: list[str] = []
+        for flow in self.flows.values():
+            for deployment in flow.deployments:
+                names.append(deployment.name)
+        return sorted(names)

Evidence

Compliance requires identifiers to be self-documenting and not misleading. The property docstring
claims it returns full deployment names, but the implementation appends deployment.name instead of
deployment.full_name.

Rule 2: Generic: Meaningful Naming and Self-Documenting Code
src/prefect/_sdk/models.py[120-129]

Agent Prompt

## Issue description
`SDKData.deployment_names` claims to return deployment full names, but it currently appends `deployment.name`.

## Issue Context
This makes the code misleading and can break downstream consumers expecting `flow-name/deployment-name` strings.

## Fix Focus Areas
- src/prefect/_sdk/models.py[120-129]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. `make_unique_identifier` uses `or` 📘 Rule Violation

Description

• make_unique_identifier() is intended to enforce uniqueness and reserved-name avoidance, but its
  availability check uses or, causing it to return base even when it collides with existing or
  is in reserved.
• This breaks collision handling and reserved-name avoidance, creating edge-case failures where
  generated identifiers are not unique or conflict with reserved SDK surface names.
• Because this failure can occur silently, it increases the risk of producing invalid/ambiguous
  generated SDK output without obvious error signals.

correctness

Code

src/prefect/_sdk/naming.py[R253-255]

+    # Check if base is available
+    if base not in existing or base not in reserved:
+        return base

Evidence

Robust edge-case management requires correctly handling collisions and reserved identifiers. The
current or condition returns base if it is not in one set, even if it is present in the other,
defeating both uniqueness and reserved-name checks.

Rule 3: Generic: Robust Error Handling and Edge Case Management
src/prefect/_sdk/naming.py[253-255]

Agent Prompt

## Issue description
`make_unique_identifier()` incorrectly uses `or` when deciding whether `base` is available, allowing collisions or reserved names to slip through.

## Issue Context
This function is responsible for ensuring generated identifiers are both unique and do not conflict with reserved SDK surface names.

## Fix Focus Areas
- src/prefect/_sdk/naming.py[253-263]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

3. Deferred import in test 📘 Rule Violation

Description

• A new import is introduced inside test_all_reserved_in_module_context() without an obvious need
  (no circular import avoidance or optional dependency handling).
• This reduces predictability/tooling compatibility and violates the project guidance to keep
  imports at module scope unless required.

reliability

Code

tests/_sdk/test_naming.py[R300-303]

+    def test_all_reserved_in_module_context(self):
+        """Test that 'all' is reserved (normalized form of '__all__')."""
+        from prefect._sdk.naming import RESERVED_MODULE_IDENTIFIERS
+

Evidence

The compliance rule requires imports to remain at module scope unless there is a clear justification
(e.g., circular imports/optional deps). The test introduces an inline import without such
justification.

AGENTS.md
tests/_sdk/test_naming.py[300-305]

Agent Prompt

## Issue description
A deferred import was added inside a test function without a clear need.

## Issue Context
Keeping imports at module scope improves readability and tooling support.

## Fix Focus Areas
- tests/_sdk/test_naming.py[1-14]
- tests/_sdk/test_naming.py[300-305]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[qodo-code-review]

1. deployment_names returns wrong values 📘 Rule Violation

• SDKData.deployment_names is documented as returning deployment full_name values, but it
  actually appends deployment.name, making the property’s behavior inconsistent with its own
  docstring.
• This is misleading/self-non-documenting and can cause downstream code generation to use incomplete
  identifiers (deployment-only names) instead of flow/deployment names.
• The mismatch also undermines deterministic output expectations because the returned values are not
  the intended ones.

Agent Prompt

## Issue description
`SDKData.deployment_names` claims to return deployment full names, but it currently appends `deployment.name`.

## Issue Context
This makes the code misleading and can break downstream consumers expecting `flow-name/deployment-name` strings.

## Fix Focus Areas
- src/prefect/_sdk/models.py[120-129]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

[qodo-code-review]

2. make_unique_identifier uses or 📘 Rule Violation

• make_unique_identifier() is intended to enforce uniqueness and reserved-name avoidance, but its
  availability check uses or, causing it to return base even when it collides with existing or
  is in reserved.
• This breaks collision handling and reserved-name avoidance, creating edge-case failures where
  generated identifiers are not unique or conflict with reserved SDK surface names.
• Because this failure can occur silently, it increases the risk of producing invalid/ambiguous
  generated SDK output without obvious error signals.

Agent Prompt

## Issue description
`make_unique_identifier()` incorrectly uses `or` when deciding whether `base` is available, allowing collisions or reserved names to slip through.

## Issue Context
This function is responsible for ensuring generated identifiers are both unique and do not conflict with reserved SDK surface names.

## Fix Focus Areas
- src/prefect/_sdk/naming.py[253-263]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

[qodo-code-review]

3. Deferred import in test 📘 Rule Violation

• A new import is introduced inside test_all_reserved_in_module_context() without an obvious need
  (no circular import avoidance or optional dependency handling).
• This reduces predictability/tooling compatibility and violates the project guidance to keep
  imports at module scope unless required.

Agent Prompt

## Issue description
A deferred import was added inside a test function without a clear need.

## Issue Context
Keeping imports at module scope improves readability and tooling support.

## Fix Focus Areas
- tests/_sdk/test_naming.py[1-14]
- tests/_sdk/test_naming.py[300-305]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools