Skip to content

CONFIGURATION.md

Latest commit

 

History

History
120 lines (92 loc) · 4.87 KB

CONFIGURATION.md

File metadata and controls

120 lines (92 loc) · 4.87 KB

Configuration Guide — REslava Loom

The REslava Loom workflow system is configured via VS Code settings. These settings control user preferences, AI behavior, and system defaults.

User Personalization

Setting Type Default Description
reslava-loom.user.name string "" Your preferred name. Used in document headers (## Rafa:) and AI conversations.
reslava-loom.user.email string "" (Optional) Your email address. Reserved for future collaboration features.

Example settings.json:

{
  "reslava-loom.user.name": "Rafa",
  "reslava-loom.user.email": "rafa@example.com"
}

AI Provider Configuration

Setting Type Default Description
reslava-loom.ai.provider string "deepseek" AI provider: "deepseek", "openai", "anthropic", or "ollama".
reslava-loom.ai.apiKey string "" Your API key for the selected provider.
reslava-loom.ai.model string "deepseek-chat" Model name (e.g., "deepseek-chat", "deepseek-reasoner").
reslava-loom.ai.baseUrl string Provider default Override the API endpoint (useful for local proxies or Ollama).

Example:

{
  "reslava-loom.ai.provider": "deepseek",
  "reslava-loom.ai.apiKey": "sk-...",
  "reslava-loom.ai.model": "deepseek-chat"
}

For local Ollama:

{
  "reslava-loom.ai.provider": "ollama",
  "reslava-loom.ai.baseUrl": "http://localhost:11434/v1",
  "reslava-loom.ai.model": "deepseek-r1:7b"
}

Context & Token Management

Setting Type Default Description
reslava-loom.ai.maxContextTokens number 8000 Maximum tokens for the entire AI prompt.
reslava-loom.ai.designSummaryThreshold number 20000 Characters in design.md before auto‑summary triggers.
reslava-loom.ai.reservedResponseTokens number 1000 Tokens reserved for the AI's response.
reslava-loom.ai.summarizationModel string Same as ai.model Model used for generating context summaries (can be a cheaper model).
reslava-loom.ai.summarizationMaxTokens number 500 Maximum tokens for summary generation.

Security & Effects

Setting Type Default Description
reslava-loom.allowShellCommands boolean false Master switch for run_command effect.
reslava-loom.allowSensitiveEnvVars boolean false If true, environment variables containing SECRET, TOKEN, PASSWORD are passed to subprocess. Not recommended.
reslava-loom.allowOutsideCwd boolean false If true, run_command can execute outside the loom root.
reslava-loom.commandTimeoutMs number 60000 Global maximum timeout for run_command (milliseconds).
reslava-loom.denyListPatterns string[] [] Additional regex patterns to block for run_command.

UI & Tree View

Setting Type Default Description
reslava-loom.tree.defaultGrouping string "thread" Default grouping mode: "type", "thread", "status", or "release".
reslava-loom.tree.showArchived boolean false Show archived threads in the tree view.
reslava-loom.tree.showChats boolean true Show the "Chats" node in the tree view.
reslava-loom.fileWatcherDebounceMs number 300 Debounce delay for file watcher (milliseconds).

Experimental / Advanced

Setting Type Default Description
reslava-loom.logLevel string "info" Logging verbosity: "debug", "info", "warn", "error".
reslava-loom.enableTokenTracking boolean true Show session token usage in status bar.
reslava-loom.autoSummarize boolean true Automatically generate -ctx.md when threshold exceeded.

Full Example settings.json

{
  "reslava-loom.user.name": "Rafa",
  "reslava-loom.ai.provider": "deepseek",
  "reslava-loom.ai.apiKey": "sk-...",
  "reslava-loom.ai.model": "deepseek-chat",
  "reslava-loom.ai.maxContextTokens": 8000,
  "reslava-loom.allowShellCommands": false,
  "reslava-loom.tree.defaultGrouping": "thread",
  "reslava-loom.logLevel": "info"
}

See Also