Add embedding profile size aliases (small/base/large)

[Copilot]

• Centralizes CLI profile choices in the registry so aliases and allowed options stay in sync (including default/quality/code).
• Expands codexa init --profile help text to mention both size and named aliases now accepted.
• Updates codexa models profiles guidance to show canonical profiles with a concise alias note.
• Parameterizes the init profile alias test to cover all aliases uniformly and assert resolved profiles via output/config.

[Copilot]

Pull request overview

Adds size-oriented embedding profile aliases to make codexa init --profile easier to choose on constrained hardware, while keeping resolution consistent via the model registry.

Changes:

• Extend embedding profile alias mapping to include base → balanced and large → precise (alongside small → fast).
• Update codexa init --profile option validation/help text to accept and describe the new aliases.
• Add docs and CLI tests covering the new alias selections.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
semantic_code_intelligence/tests/test_cli.py	Adds a new CLI test ensuring small/base/large map to the expected underlying profiles.
semantic_code_intelligence/embeddings/model_registry.py	Extends PROFILE_ALIASES to include base and large aliases.
semantic_code_intelligence/cli/commands/init_cmd.py	Expands Click --profile choices and updates help text to mention aliases.
README.md	Documents the new alias mappings in the init section.
docs/guide/quickstart.md	Adds a quickstart note listing the alias mappings.

[M9nx]

@copilot apply changes based on the comments in this thread

[Copilot]

@copilot apply changes based on the comments in this thread

Added explicit output checks so the base alias must resolve and emit the profile message; test now fails if resolution silently falls back (d5126fc).

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

[M9nx]

@copilot apply changes based on the comments in this thread

[Copilot]

@copilot apply changes based on the comments in this thread

Updated init CLI to consume centralized profile choices from the registry (including aliases) and parameterized the profile alias test for easier future additions; covered in 1752a7d.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 3 comments.

[Copilot]

all_profile_names is a module-level constant used to build CLI_PROFILE_CHOICES, but it’s named like a local variable while surrounding constants are uppercase (PROFILE_ALIASES, CORE_PROFILES, CLI_PROFILE_CHOICES). Renaming it to ALL_PROFILE_NAMES (or inlining it) would keep naming consistent and make it clearer this is part of the public/constant configuration.

[M9nx]

@copilot apply changes based on this feedback

[Copilot]

Renamed the profile-name union constant to ALL_PROFILE_NAMES and kept it alongside the other registry constants; changes in f1cf67d.

[Copilot]

The --profile option now accepts all values in CLI_PROFILE_CHOICES, which (via PROFILE_ALIASES) includes aliases like default, quality, and code in addition to small/base/large. The help text only mentions the size aliases, so CLI help/error output can suggest values that aren’t described. Either update the help text to mention the full alias set, or narrow CLI_PROFILE_CHOICES to only the intended user-facing aliases.

Suggested change

	help="Embedding model profile: fast (low RAM), balanced (default), or precise (code-optimized). Size aliases (small/base/large) are supported.",
	help=(
	"Embedding model profile: fast (low RAM), balanced (default), or precise (code-optimized). "
	"Size aliases (small/base/large) and named aliases (default, quality, code) are supported."
	),

[Copilot]

Printing Use: codexa init --profile <...> by joining CLI_PROFILE_CHOICES can produce a long/cryptic list (e.g., including default|quality|code alongside canonical profiles), which may be confusing since models profiles itself only displays the canonical profiles. Consider formatting this as fast|balanced|precise plus a short note listing the supported aliases/mappings, rather than dumping every accepted token.

Suggested change

	print_info(f"Use: codexa init --profile <{'|'.join(CLI_PROFILE_CHOICES)}>")
	canonical_names = sorted({p.name for p in MODEL_PROFILES.values()})
	print_info(f"Use: codexa init --profile <{'|'.join(canonical_names)}>")
	# Show a concise note about additional aliases, if any, without dumping every token inline.
	alias_map: dict[str, set[str]] = {}
	for choice in CLI_PROFILE_CHOICES:
	try:
	profile = resolve_profile(choice)
	except Exception:
	# Be defensive: ignore any unexpected tokens rather than breaking CLI output.
	continue
	if profile is None:
	continue
	if choice == profile.name:
	# Skip canonical names; we already show them above.
	continue
	alias_map.setdefault(profile.name, set()).add(choice)
	if alias_map:
	alias_parts = []
	for name in sorted(alias_map.keys()):
	aliases = "/".join(sorted(alias_map[name]))
	alias_parts.append(f"{name} ({aliases})")
	print_info(f"Aliases also supported: {', '.join(alias_parts)}")

[M9nx]

@copilot apply changes based on the comments in this thread

[Copilot]

@copilot apply changes based on the comments in this thread

Expanded the init help text to mention all supported aliases and updated models profiles guidance to show canonical profiles with a concise alias note (see f1cf67d).

[M9nx]

Seems good !