feat(cli): add cn config command for model management

[aaronlippold]

PR: Add cn config Command for Model Management

Description

Adds a comprehensive cn config command group to the CLI for managing Continue configuration files. This enables users to verify, sync, and manage models in their config.yaml without manual editing.

Problem

Currently, users must manually edit config.yaml to add/remove models, and there's no way to verify that configured models actually exist on their provider. This leads to:

• Configuration errors when models are deprecated or renamed
• Manual work to discover available models
• No tooling for CI/CD integration

Solution

The cn config command provides:

• Validation: Verify configured models exist on the provider
• Sync: Automatically remove unavailable models
• Discovery: List available models from any OpenAI-compatible API
• CRUD: Add/remove models programmatically
• Generation: Bootstrap config from available models
• Backup/Restore: Safe config management with automatic backups

Commands

Command	Description
cn config verify	Check if configured models exist on provider
cn config sync	Remove unavailable models (supports --dry-run)
cn config validate	Validate config file structure
cn config sections	Show config sections
cn config list	List available models from provider
cn config test	Test each chat model with sample prompt
cn config add <model>	Add a model with --name, --role options
cn config remove <model>	Remove a model from config
cn config generate	Bootstrap config from available models
cn config show	Display current config
cn config diff	Compare config vs available models
cn config backups	List backup files
cn config restore <backup>	Restore from backup

Global Options

Option	Description
--json	Output as JSON for CI/scripting
--provider <name>	Use preset: openai, anthropic, azure, ollama, together, groq, mistral
--api-base <url>	Custom API base URL
--api-key-env <var>	Environment variable for API key
--auth-header <header>	Custom auth header (default: Authorization)
--chat-only	Filter to chat models only
--embed-only	Filter to embedding models only
--rerank-only	Filter to rerank models only
--filter <pattern>	Filter models by pattern
--dry-run	Preview changes without applying

Provider Presets

Provider	API Base	Auth Header	Env Var
openai	api.openai.com/v1	Authorization	OPENAI_API_KEY
anthropic	api.anthropic.com/v1	x-api-key	ANTHROPIC_API_KEY
azure	(user provided)	api-key	AZURE_API_KEY
ollama	localhost:11434/v1	(none)	-
together	api.together.xyz/v1	Authorization	TOGETHER_API_KEY
groq	api.groq.com/openai/v1	Authorization	GROQ_API_KEY
mistral	api.mistral.ai/v1	Authorization	MISTRAL_API_KEY

Example Usage

# Verify models exist on provider
cn config verify --provider openai

# Sync config - remove models that don't exist
cn config sync --provider openai --dry-run
cn config sync --provider openai

# List available models
cn config list --provider openai --chat-only
cn config list --provider ollama --filter llama

# Add a model
cn config add gpt-4o --provider openai --role chat

# Generate fresh config from available models
cn config generate --provider openai --chat-only

# Compare config to available models
cn config diff --provider openai --json

# Backup management
cn config backups
cn config restore config.backup-2026-01-27.yaml

Key Features

• Comment preservation: Uses yaml package (eemeli/yaml) to preserve comments in config files
• Safe writes: Atomic file operations with automatic backup before any modification
• PRESERVED_SECTIONS: mcpServers, rules, prompts, context, data, docs are never modified by sync/generate
• Provider-agnostic: Works with any OpenAI-compatible API
• CI/CD friendly: --json output for programmatic use

Bug Fix

Also includes a fix for CLIPlatformClient.ts:

• .env files are now checked BEFORE process.env, matching IDE behavior
• Previously caused inconsistency between CLI and IDE environments

Checklist

• I've read the contributing guide
• The relevant docs, if any, have been updated or created
• The relevant tests, if any, have been updated or created

Tests

File	Tests	Coverage
configModels.test.ts	63	Core functions
index.test.ts	8	Command handlers
CLIPlatformClient.test.ts	2	Env priority fix
Total new	73

All 1867 tests passing.

Screen recording or screenshot

CLI command - text output only, no UI changes.

$ cn config verify --provider openai
Checking 3 models against https://api.openai.com/v1...
✓ gpt-4o exists
✓ gpt-4o-mini exists
✗ gpt-3.5-turbo-0301 NOT FOUND
2 of 3 models verified

---

Continue Tasks: ▶️ 1 queued — View all

---

Summary by cubic

Adds a new cn config command group to manage models in config.yaml from the CLI, including verification, syncing, discovery, CRUD, and safe backups. Also fixes env loading to prefer .env over process.env for consistent behavior with the IDE.

• New Features

  • Introduces cn config to manage models: verify/sync, list/test, add/remove, generate/diff/show, backups/restore, validate/sections.
  • Preserves comments and non-model sections; only models are modified. Atomic writes with automatic backups.
  • Works with any OpenAI-compatible API; presets for openai, anthropic, azure, ollama, together, groq, mistral.
  • Supports filters (chat/embed/rerank, pattern) and --json output for CI. Includes 73 new tests.

• Bug Fixes

  • CLI now checks .env before process.env in CLIPlatformClient, matching IDE behavior.

^{Written for commit 4609b83. Summary will update on new commits.}

[cubic-dev-ai]

1 issue found across 7 files

Prompt for AI agents (all issues)

Check if these issues are valid — if so, understand the root cause of each and fix them.


<file name="extensions/cli/src/commands/config/index.ts">

<violation number="1" location="extensions/cli/src/commands/config/index.ts:127">
P1: Rule violated: **Don't use console.log**

Replace console.log/error with the standard logger (e.g., logger.info/error) per the no-console rule.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

[cubic-dev-ai]

P1: Rule violated: Don't use console.log

Replace console.log/error with the standard logger (e.g., logger.info/error) per the no-console rule.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At extensions/cli/src/commands/config/index.ts, line 127:

<comment>Replace console.log/error with the standard logger (e.g., logger.info/error) per the no-console rule.</comment>

<file context>
@@ -0,0 +1,891 @@
+ * Verify command - Check if config models are available
+ */
+async function runVerify(options: ConfigCommandOptions): Promise<void> {
+  console.log("Verifying config against available models...\n");
+
+  const continueHome = getContinueHome(options.continueHome);
</file context>