chore: pre-public-release cleanup

- Remove internal document reference from models.py (personal doc path)
- Remove duplicate 'Built-in -> Amplifier Capability Mapping' header in _constants.py
- Fix provider name in __init__.py docstring (copilot-sdk -> github-copilot)
- Clarify amplifier-core is runtime-provided in README dependencies
- Broaden pyproject.toml keywords to reflect multi-model support

🤖 Generated with [Amplifier](https://github.com/microsoft/amplifier)

Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>

Author: robotdad

README.md

@@ -152,7 +152,7 @@ make check         # Full check (lint + test)
 
 ## Dependencies
 
-- `amplifier-core>=1.0.0`
+- `amplifier-core` (provided by Amplifier runtime, not installed separately)
 - `github-copilot-sdk>=0.1.0,<0.2.0`
 
 

amplifier_module_provider_github_copilot/__init__.py

@@ -21,7 +21,7 @@
     ```yaml
     # In amplifier.yaml
     providers:
-      copilot-sdk:
+      github-copilot:
         model: claude-opus-4.5
         # timeout: 3600        # 1 hour default (override if needed)
         # thinking_timeout: 3600  # Same as regular (override if needed)
@@ -30,7 +30,7 @@
 
     Or via CLI:
     ```bash
-    amplifier chat --provider copilot-sdk
+    amplifier chat --provider github-copilot
     ```
 
 Prerequisites:

amplifier_module_provider_github_copilot/_constants.py

@@ -182,10 +182,6 @@ class LoopExitMethod(Enum):
 # Evidence: Same (delegate, report_intent) pair repeated 303 times
 DEDUPLICATE_TOOL_CALLS = True
 
-# ═══════════════════════════════════════════════════════════════════════════════
-# Built-in → Amplifier Capability Mapping
-# ═══════════════════════════════════════════════════════════════════════════════
-
 BUILTIN_TO_AMPLIFIER_CAPABILITY: dict[str, frozenset[str]] = {
     # File operation overlaps
     "view": frozenset({"read_file"}),

amplifier_module_provider_github_copilot/models.py

@@ -93,7 +93,6 @@ def copilot_model_to_internal(raw_model: Any) -> CopilotModelInfo:
                 # Derive max output from context window minus prompt allocation
                 # NOTE: Do NOT cap this value. The SDK provides authoritative limits.
                 # Capping causes incorrect budget calculation in context manager.
-                # See: myDocs-Amp-CLI-SDK-provider/SDK-TOKEN-LIMITS-RESEARCH-20260208.md
                 max_output_tokens = context_window - limits.max_prompt_tokens
 
         if hasattr(caps, "supports") and caps.supports:

pyproject.toml

@@ -8,7 +8,7 @@ requires-python = ">=3.11"
 authors = [
     { name = "Microsoft MADE:Explorations Team" },
 ]
-keywords = ["amplifier", "copilot", "llm", "ai", "claude"]
+keywords = ["amplifier", "copilot", "github", "llm", "ai", "provider"]
 classifiers = [
     "Development Status :: 3 - Alpha",
     "Intended Audience :: Developers",