refactor(plugins): pin anthropic to 1.48, drop redundant colon check, trim docs

.kiro/skills/library-development/references/project-map.md

@@ -113,7 +113,7 @@ for the chapter-by-chapter reference.
 
 - **Python ≥ 3.11** (ruff/ty target 3.13). Runtime deps: `strands-agents`
   (>=1.48,<2), `pydantic` v2, `pyyaml`, `mcp`. Optional extras:
-  `agentcore-memory`, `ollama`, `openai`, `gemini`. `anthropic`.
+  `agentcore-memory`, `ollama`, `openai`, `gemini`, `anthropic`.
 - **MCP servers** run on a background daemon thread with a self-managed
   `uvicorn.Server` (HTTP transports only — `streamable-http`, `sse`); `stdio`
   is client-side (the client spawns a subprocess).

docs/configuration/Chapter_19.md

@@ -103,68 +103,15 @@ agents:
           skills: ./skills/     # one skill directory, or a parent of several
 ```
 
-> **Paths in `params` are not rewritten.** Unlike a plugin's `type:` (which is resolved
-> relative to the config file), everything under `params` is forwarded to the plugin
-> verbatim. `skills: ./skills/` therefore resolves against the process working directory —
-> run the config from the directory that makes that path valid.
+> **Note — path params are relative to the working directory.** A plugin's `type:` is
+> resolved relative to the config file, but any path *inside* `params` (such as a skills
+> directory) is forwarded verbatim and resolves against the current working directory. Run
+> the config from a directory where that path is valid. (This may change in a future
+> release.)
 
-### Authoring a skill (SKILL.md)
-
-Agent Skills is an open, cross-vendor format — originated by Anthropic and adopted across
-agent tooling. A skill is a directory (named after the skill) containing a `SKILL.md`: YAML
-frontmatter followed by a markdown body. It may also ship resource directories the agent
-reads on demand.
-
-```
-conventional-commits/
-├── SKILL.md          # required — frontmatter + instructions
-├── scripts/          # optional — runnable scripts
-├── references/       # optional — docs loaded only when needed
-└── assets/           # optional — templates, data, images
-```
-
-Frontmatter fields, per the [Agent Skills specification](https://agentskills.io/specification):
-
-| Field | Required | Notes |
-|-------|----------|-------|
-| `name` | yes | 1–64 chars; lowercase `a–z`, `0–9`, `-`; no leading/trailing or consecutive hyphens. **Must match the directory name.** |
-| `description` | yes | 1–1024 chars. State *what* it does **and** *when* to use it, with keywords the model can match — this is all the agent sees until it activates the skill. |
-| `license` | no | A license name or a bundled license file. |
-| `compatibility` | no | ≤500 chars. Environment requirements (target product, system packages, network access). |
-| `metadata` | no | Arbitrary string-to-string map for client-specific data. |
-| `allowed-tools` | no | Space-separated pre-approved tool names (experimental; support varies). |
-
-```markdown
----
-name: conventional-commits
-description: Write git commit messages that follow the Conventional Commits standard. Use when the user asks for a commit message or wants a change summary written up for git.
-license: Apache-2.0
----
-
-# Conventional Commits
-
-## When to use
-Activate when the user asks for a commit message...
-
-## Steps
-1. Choose the single type that best fits the change...
-
-## Examples
-Input: "added pagination to the users endpoint" -> `feat(api): add pagination to the users endpoint`
-
-## Edge cases
-- Several unrelated changes -> recommend separate commits.
-```
-
-**Write for progressive disclosure.** The `description` is the model's only cue to activate
-the skill, so make it specific. Keep the body focused — the spec recommends under ~500 lines
-(≈5000 tokens) — and move long reference material into `references/` files that load only
-when the task needs them.
-
-> **Resource files need a tool.** `AgentSkills` handles discovery and activation only. A
-> skill that ships scripts or references also needs a file-reading tool such as `file_read`
-> from [`strands-agents-tools`](https://strandsagents.com/docs/user-guide/concepts/plugins/skills/#providing-tools-for-resource-access);
-> add it under the agent's `tools:`. Instruction-only skills need none.
+A skill is just a `SKILL.md` directory that `AgentSkills` discovers. Authoring one — the
+frontmatter fields, progressive disclosure, resource files — is owned by strands and the
+open Agent Skills format, not by strands-compose; `examples/15_plugins/` ships a working one.
 
 ### Further reading
 
@@ -178,8 +125,7 @@ when the task needs them.
 
 | Condition | Exception |
 |-----------|-----------|
-| `type` has no `:` separator | `ValueError` |
-| Malformed spec / missing file / module / attribute | `ImportResolutionError` (a `ValueError` subclass, from `load_object`) |
+| Malformed spec (no `:` separator) / missing file / module / attribute | `ImportResolutionError` (a `ValueError` subclass, from `load_object`) |
 | Resolved object is not callable | `TypeError` |
 | Resolved object is callable but does not return a `Plugin` | `TypeError` |
 | Constructor or factory raises | the original exception, unwrapped |

pyproject.toml

@@ -45,7 +45,7 @@ gemini = [
     "strands-agents[gemini]>=1.48.0,<2.0.0",
 ]
 anthropic = [
-    "strands-agents[anthropic]>=1.35.0,<2.0.0",
+    "strands-agents[anthropic]>=1.48.0,<2.0.0",
 ]
 
 [project.urls]

src/strands_compose/config/resolvers/hooks.py

@@ -27,24 +27,17 @@ def resolve_hook(hook_def: HookDef) -> HookProvider:
         Instantiated HookProvider.
 
     Raises:
-        ValueError: If ``type`` is not in ``module:Class`` format.
+        ImportResolutionError: If ``type`` is not a valid ``module:Class`` /
+            ``./file.py:Class`` spec (a ``ValueError`` subclass, from ``load_object``).
         TypeError: If the resolved object is not a HookProvider subclass.
     """
 
-    type_str = hook_def.type
-    if ":" not in type_str:
-        raise ValueError(
-            f"Hook type {type_str!r} is not a valid import spec.\n"
-            f"Use 'module.path:ClassName' (e.g. 'strands_compose.hooks:StopGuard') "
-            f"or './path/to/file.py:ClassName'."
-        )
-
-    cls = load_object(type_str, target="hook")
+    cls = load_object(hook_def.type, target="hook")
 
     hook = cls(**hook_def.params)
     if not isinstance(hook, HookProvider):
         raise TypeError(
-            f"Hook {type_str!r} returned {type(hook).__name__}, expected HookProvider subclass."
+            f"Hook {hook_def.type!r} returned {type(hook).__name__}, expected HookProvider subclass."
         )
     return hook
 

src/strands_compose/config/resolvers/plugins.py

@@ -27,24 +27,17 @@ def resolve_plugin(plugin_def: PluginDef) -> Plugin:
         Instantiated Plugin.
 
     Raises:
-        ValueError: If ``type`` is not in ``module:Class`` format.
+        ImportResolutionError: If ``type`` is not a valid ``module:Class`` /
+            ``./file.py:Class`` spec (a ``ValueError`` subclass, from ``load_object``).
         TypeError: If calling the resolved object does not produce a Plugin.
     """
 
-    type_str = plugin_def.type
-    if ":" not in type_str:
-        raise ValueError(
-            f"Plugin type {type_str!r} is not a valid import spec.\n"
-            f"Use 'module.path:ClassName' (e.g. 'strands:AgentSkills') "
-            f"or './path/to/file.py:ClassName'."
-        )
-
-    obj = load_object(type_str, target="plugin")
+    obj = load_object(plugin_def.type, target="plugin")
 
     plugin = obj(**plugin_def.params)
     if not isinstance(plugin, Plugin):
         raise TypeError(
-            f"Plugin {type_str!r} returned {type(plugin).__name__}, "
+            f"Plugin {plugin_def.type!r} returned {type(plugin).__name__}, "
             f"expected strands.plugins.Plugin subclass."
         )
     return plugin