adapt package to reorganized backend

.env.example

@@ -39,5 +39,5 @@ GUARD_SPACE_ID=
 #GUARD_ENV_FILE=.env.staging
 
 # (OPTIONAL) Default media file to analyze
-# Used only by scripts/create_activity.py
+# Used only by scripts/create_prediction.py
 #GUARD_MEDIA=path/to/photo.jpg

CONTRIBUTING.md

@@ -42,7 +42,7 @@ The `docs` group is also excluded since it is only used for building documentati
 
 To keep the repository organized, please use descriptive branch names based on the type of work you are doing:
 
-* **Features:** `feature/activity-pagination` or `feat/share-expiry-filters`
+* **Features:** `feature/prediction-pagination` or `feat/share-expiry-filters`
 * **Bug Fixes:** `fix/presigned-upload-retry`
 * **Documentation:** `docs/update-readme`
 
@@ -127,23 +127,23 @@ you are finished, running `uv sync` returns you to the default environment.
 ### Testing against a Live API
 
 The test suite is fully mocked and never touches the network. To exercise the real lifecycle end-to-end, create an 
-activity. **This runs against production by default and spends real tokens** — one activity per run. You can point it 
+prediction. **This runs against production by default and spends real tokens** — one prediction per run. You can point it 
 elsewhere using the `GUARD_BASE_URL` environment variable.
 
 ```bash
 export GUARD_API_KEY=... # a token_raw from POST /api/v1/tokens/
 export GUARD_SPACE_ID=...
 
 # see the space and the projected cost, creating nothing
-uv run python scripts/create_activity.py path/to/photo.jpg --list-only
+uv run python scripts/create_prediction.py path/to/photo.jpg --list-only
 
-uv run python scripts/create_activity.py path/to/photo.jpg
+uv run python scripts/create_prediction.py path/to/photo.jpg
 
 # against a local dev server instead
-GUARD_BASE_URL=http://localhost:8000 uv run python scripts/create_activity.py path/to/photo.jpg
+GUARD_BASE_URL=http://localhost:8000 uv run python scripts/create_prediction.py path/to/photo.jpg
 
 # on-device, which needs no space, no network and no tokens
-uv run python scripts/create_activity.py path/to/photo.jpg --engine local
+uv run python scripts/create_prediction.py path/to/photo.jpg --engine local
 ```
 
 Credentials resolve the same way everywhere in this client: explicit arguments, followed by `GUARD_* `environment 
@@ -157,15 +157,15 @@ If you do not have a `space_id` yet, four more scripts can help:
 # list the spaces your key can see, with their ids
 uv run python scripts/list_spaces.py
 
-# the two ids a new space needs: a predictor, and optionally some tasks
+# the two ids a new space needs: a predictor, and optionally some targets
 uv run python scripts/list_predictors.py
-uv run python scripts/list_tasks.py --predictor-id <uuid>
+uv run python scripts/list_targets.py --predictor-id <uuid>
 
-# walk predictors -> tasks -> create a space (--list-only creates nothing)
+# walk predictors -> targets -> create a space (--list-only creates nothing)
 uv run python scripts/create_space.py --list-only
 
 # and then, with a space id in hand, analyze something in it
-uv run python scripts/create_activity.py path/to/photo.jpg --list-only
+uv run python scripts/create_prediction.py path/to/photo.jpg --list-only
 ```
 
 ## What to Watch Out For
@@ -190,8 +190,8 @@ exceptions do not subclass it, so `local.py` translates each one before it escap
 that does not come from `guard_local` propagates untouched. This is intentional, as a bug in the engine should surface 
 exactly as the bug it is.
 
-**The task labels are a public API.** `_adapt` derives each local result's id from `uuid5(namespace, label)`. Renaming a 
-label silently changes the ID that callers rely on. These labels also mirror the tasks the cloud API seeds, allowing 
+**The target labels are a public API.** `_adapt` derives each local result's id from `uuid5(namespace, label)`. Renaming a 
+label silently changes the ID that callers rely on. These labels also mirror the targets the cloud API seeds, allowing 
 users to test locally and then route to the cloud without changing their code.
 
 **Engine scores must be floats between `0.0` and `1.0`.** The `_score_to_int` function rescales by value rather than by 

README.md

@@ -22,7 +22,7 @@ and test locally without an API key. Both return the exact same strictly typed m
 **⚡ Sync & Async Support:** Natively supports both synchronous operations and `async/await` out of the box, making it 
 a perfect fit for high-performance frameworks like FastAPI.
 
-**🛠️ Comprehensive API Bindings:** Typed bindings for the entire API including spaces, activities, tasks, predictors, 
+**🛠️ Comprehensive API Bindings:** Typed bindings for the entire API including spaces, predictions, targets, predictors, 
 runners, shares, and reactions. We also include token estimation, so you can price a job before you spend on it.
 
 ## Installation
@@ -51,10 +51,10 @@ pip install "guard-client[local]"
 
 ## Quick Start
 
-The `analyze()` method runs the entire detection lifecycle for you. It creates an activity, uploads the media, confirms 
+The `analyze()` method runs the entire detection lifecycle for you. It creates a prediction, uploads the media, confirms 
 the upload, polls until processing finishes, and returns the result.
 
-You will need an **API key** and a space to create the activity in. You can create a new access token in the **Elhio 
+You will need an **API key** and a space to create the prediction in. You can create a new access token in the **Elhio 
 dashboard** (under Settings > Account Settings > Access Tokens). If you do not know your space ID yet, you can 
 ask the API for it. The `spaces.list()` method returns every space the key can see.
 
@@ -66,15 +66,15 @@ with GuardClient(api_key="your_api_key_here") as client:
     # 1. Get an available space
     space = client.spaces.list()[0]
 
-    # 2. Analyze the media (creates activity, uploads, and polls for results)
+    # 2. Analyze the media (creates prediction, uploads, and polls for results)
     result = client.analyze("photo.jpg", space_id=space.id)
 
     # 3. Print the detection results
     for item in result.results:
         print(f"{item.label}: {item.score}/100")
 
 # AI-Generated: 87/100
-# Violence: 2/100
+# Violent: 2/100
 # Explicit: 1/100
 ```
 
@@ -110,11 +110,11 @@ with GuardClient(engine="local") as client:
         print(f"{item.label}: {item.score}/100")
 
 # AI-Generated: 90/100
-# Violence: 2/100
+# Violent: 2/100
 # Explicit: 1/100
 ```
 
-Two fields easily tell the engines apart. The `result.activity_id` is `None` because a local run creates nothing 
+Two fields easily tell the engines apart. The `result.prediction_id` is `None` because a local run creates nothing 
 server-side, meaning there is nothing to share or react to. The `result.detected` field carries the engine's own 
 per-category threshold verdict, which the cloud API does not report. Reading it means you are opting into extra detail, 
 not into a different data shape.

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "guard-client"
-version = "0.0.1"
+version = "0.0.2"
 description = "A Python client for seamlessly integrating visual safety filters into your applications"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -42,7 +42,7 @@ dependencies = [
 ]
 
 [project.optional-dependencies]
-local = ["guard-local-detector>=0.0.1"]
+local = ["guard-local-detector>=0.0.2"]
 
 [project.urls]
 Homepage = "https://elhio.com/"

scripts/create_activity.py → scripts/create_prediction.py

@@ -1,9 +1,9 @@
 #!/usr/bin/env python3
 """
-Create one activity after showing the space and its estimated cost.
+Create one prediction after showing the space and its estimated cost.
 
 This script runs the complete detection lifecycle through `analyze()`. It creates the
-activity, uploads the media, confirms it, polls until completion, and prints the final
+prediction, uploads the media, confirms it, polls until completion, and prints the final
 scores. Be aware that running this against a live API will spend real tokens. You can
 pass `--list-only` to see the cost estimate without creating anything, or use
 `--engine local` to run the detection on your device for free.
@@ -15,14 +15,14 @@
 Example:
     ```bash
     # preview the cost without creating anything
-    uv run python scripts/create_activity.py photo.jpg --list-only
+    uv run python scripts/create_prediction.py photo.jpg --list-only
 
-    # run the activity against the cloud API
-    uv run python scripts/create_activity.py photo.jpg
-    uv run python scripts/create_activity.py photo.jpg --space-id <uuid>
+    # run the prediction against the cloud API
+    uv run python scripts/create_prediction.py photo.jpg
+    uv run python scripts/create_prediction.py photo.jpg --space-id <uuid>
 
     # run on-device instead, which requires no space and consumes no tokens
-    uv run python scripts/create_activity.py photo.jpg --engine local
+    uv run python scripts/create_prediction.py photo.jpg --engine local
     ```
 """
 
@@ -41,8 +41,10 @@ def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(description=__doc__.splitlines()[0])
     parser.add_argument("media", nargs="?", help="media file; defaults to GUARD_MEDIA")
     parser.add_argument("--space-id", help="the space to run in; or GUARD_SPACE_ID")
-    parser.add_argument("--user-id", help="create a user-owned activity")
-    parser.add_argument("--account-id", help="create a service-account-owned activity")
+    parser.add_argument("--user-id", help="create a user-owned prediction")
+    parser.add_argument(
+        "--account-id", help="create a service-account-owned prediction"
+    )
     parser.add_argument(
         "--engine", choices=["cloud", "local"], help="client default: cloud"
     )
@@ -87,13 +89,13 @@ def print_media(path: Path) -> None:
 def print_space(client: GuardClient, space_id: str, media: Path) -> None:
     """Show the space this will run in, and what it is expected to cost."""
     space = client.spaces.get(space_id)
-    tasks = ", ".join(task.name for task in space.enabled_tasks) or "-"
+    targets = ", ".join(target.name for target in space.enabled_targets) or "-"
 
     print(f"\nSpace: {space.name}")
     print(f"  id         {space.id}")
     print(f"  predictor  {space.predictor_name} (x{space.predictor_multiplier})")
     print(f"  max media  {space.max_media_size or '-'} bytes")
-    print(f"  tasks      {tasks}")
+    print(f"  targets      {targets}")
 
     print(f"\nEstimated cost: {client.estimate_tokens(media, space_id=space_id)}")
     print("  An estimate: the API reserves the minimum, payed_tokens is final.")
@@ -133,7 +135,7 @@ def main() -> int:
             )
 
             print(f"\nDone: engine={result.engine.value}")
-            print(f"  activity  {result.activity_id or '-'}")
+            print(f"  prediction  {result.prediction_id or '-'}")
             for item in result.results:
                 print(f"  - {item.label}: {item.score}/100")
             if not result.results:

scripts/create_space.py

@@ -1,10 +1,10 @@
 #!/usr/bin/env python3
 """
-Create a space after showing the available predictors and tasks.
+Create a space after showing the available predictors and targets.
 
-A space needs a predictor and optionally a set of tasks. This script walks through the
-entire flow. It lists the predictors, lists the tasks for the chosen predictor, creates
-the space, and prints the new space ID.
+A space needs a predictor and optionally a set of targets. This script walks through the
+entire flow. It lists the predictors, lists the targets for the chosen predictor,
+creates the space, and prints the new space ID.
 
 Only GUARD_API_KEY is required. Everything else resolves through the standard client
 precedence using arguments, the environment, or a `.env` file.
@@ -44,14 +44,14 @@ def parse_args() -> argparse.Namespace:
 
     parser.add_argument("--public", action="store_true", help="make the space public")
     parser.add_argument(
-        "--all-tasks",
+        "--all-targets",
         action="store_true",
-        help="enable every task the predictor supports (default: none)",
+        help="enable every target the predictor supports (default: none)",
     )
     parser.add_argument(
         "--list-only",
         action="store_true",
-        help="print predictors and tasks, then exit without creating anything",
+        help="print predictors and targets, then exit without creating anything",
     )
     return parser.parse_args()
 
@@ -74,12 +74,12 @@ def main() -> int:
                 print(f"  {predictor.id}  {predictor.name:28} media={media}")
 
             predictor_id = args.predictor_id or str(predictors[0].id)
-            tasks = client.tasks.list(predictor_id=predictor_id)
+            targets = client.targets.list(predictor_id=predictor_id)
 
-            print(f"\nTasks for predictor {predictor_id}:")
-            for task in tasks:
-                print(f"  {task.id}  {task.name}")
-            if not tasks:
+            print(f"\nTargets for predictor {predictor_id}:")
+            for target in targets:
+                print(f"  {target.id}  {target.name}")
+            if not targets:
                 print("  (none)")
 
             if args.list_only:
@@ -90,7 +90,7 @@ def main() -> int:
                 print("\nPass --name to create a space.", file=sys.stderr)
                 return 1
 
-            task_ids = [t.id for t in tasks] if args.all_tasks else None
+            target_ids = [t.id for t in targets] if args.all_targets else None
             print(f"\nCreating space {args.name!r}...")
 
             space = client.spaces.create(
@@ -100,13 +100,13 @@ def main() -> int:
                 is_public=args.public,
                 user_id=args.user_id,
                 organization_id=args.organization_id,
-                enabled_task_ids=task_ids,
+                enabled_target_ids=target_ids,
             )
 
             print(f"\nCreated: {space.id}")
             print(f"  name    {space.name}")
             print(f"  public  {space.is_public}")
-            print(f"  tasks   {', '.join(space.enabled_task_names) or '-'}")
+            print(f"  targets   {', '.join(space.enabled_target_names) or '-'}")
             print(f"\nPut this in .env as GUARD_SPACE_ID={space.id}")
     except GuardError as exc:
         print(f"\nFAILED: {type(exc).__name__}: {exc}", file=sys.stderr)

scripts/list_predictors.py

@@ -7,14 +7,14 @@
 required. Everything else resolves through the standard client precedence using
 arguments, the environment, or a `.env` file.
 
-The `--task-id` argument may be repeated. It narrows the results to predictors
-supporting all of the given tasks. Run `scripts/list_tasks.py` to find those IDs.
+The `--target-id` argument may be repeated. It narrows the results to predictors
+supporting all of the given targets. Run `scripts/list_targets.py` to find those IDs.
 
 Example:
     ```bash
     uv run python scripts/list_predictors.py
     uv run python scripts/list_predictors.py --sort-by created_at --sort-order desc
-    uv run python scripts/list_predictors.py --task-id <uuid> --task-id <uuid>
+    uv run python scripts/list_predictors.py --target-id <uuid> --target-id <uuid>
     ```
 """
 
@@ -35,10 +35,10 @@ def parse_args() -> argparse.Namespace:
     )
 
     parser.add_argument(
-        "--task-id",
+        "--target-id",
         action="append",
         metavar="UUID",
-        help="only predictors supporting this task; repeatable",
+        help="only predictors supporting this target; repeatable",
     )
     parser.add_argument(
         "--sort-by", choices=["name", "created_at"], help="server default: name"
@@ -58,7 +58,7 @@ def main() -> int:
     filters = {
         "user_id": args.user_id,
         "organization_id": args.organization_id,
-        "supported_task_ids": args.task_id,
+        "supported_target_ids": args.target_id,
         "sort_by": args.sort_by,
         "sort_order": args.sort_order,
     }

scripts/list_spaces.py

@@ -2,7 +2,7 @@
 """
 Print the spaces available to your API key along with their IDs.
 
-Creating an activity requires a `space_id`. This script helps you find one to place in
+Creating a prediction requires a `space_id`. This script helps you find one to place in
 your `.env` file as `GUARD_SPACE_ID`. Only the `GUARD_API_KEY` is required. Everything
 else resolves through the standard client precedence using arguments, the environment,
 or a `.env` file.