Files
optimclaw/src/config.rs
T
33ef0a6ea5 fix: security hardening across all layers (#35)
* fix: comprehensive security hardening across all layers

Critical:
- Replace --dangerously-skip-permissions with explicit tool allowlist
  via settings.json (Claude Code bridge)
- Constant-time token comparison (subtle crate) in web auth and
  orchestrator auth to prevent timing attacks

High:
- Revoke tokens and clean up handles on container creation failure
- Drop SETUID/SETGID capabilities from containers (keep only CHOWN)
- Disable redirect following in HTTP tool and WASM wrapper (SSRF)
- Reject URL userinfo (@) in WASM allowlist parser (host confusion)
- Fix binary body bypassing leak detection (from_utf8 -> from_utf8_lossy)
- Protect identity files from LLM overwrites (prompt injection defense)
- Prevent tool shadowing: built-in tools cannot be replaced dynamically
- User-scoped job APIs: list/detail/cancel/restart/prompt/events/files
- CORS restricted to localhost origins, WebSocket origin validation
- Sandbox shell fail-closed: no silent fallback to unsandboxed execution
- Scrub secrets from log broadcaster before SSE broadcast
- XSS sanitization on rendered markdown in web UI
- WASM epoch ticker thread so timeout deadlines actually fire

Medium:
- Cap state transition history at 200 entries
- SSE/WebSocket connection limit (100 max)
- Request body size limit (1MB)
- Response body size limit enforcement in WASM HTTP
- UTF-8 safe string truncation (routine engine, shell tool)
- Fix PolicyAction::Sanitize to actually run the sanitizer
- TOCTOU fix in scheduler and context manager (hold write lock)
- Project file serving moved behind auth
- Path traversal guard on project_id
- Session file permissions set to 0600 on unix
- AtomicUsize for routine running_count (panic-safe)
- Completion detection hardened against false positives and tool injection
- Tool output no longer drives job completion (only LLM response)

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix: address security review findings across all layers

- Fix path traversal sandbox bypass via lexical normalization (file.rs)
- Fix SSRF via DNS rebinding with pre-request hostname resolution (http.rs)
- Add token budget enforcement on LLM calls (reasoning.rs, state.rs)
- Fix cross-user chat history leak with ownership verification (store.rs, server.rs)
- Add sliding-window rate limiter on gateway chat endpoint (server.rs)
- Harden extension install: HTTPS-only, 50MB cap, WASM magic validation (manager.rs)
- Add destructive command blocklist that overrides shell auto-approval (shell.rs)
- Add 5MB response body size cap to HTTP tool (http.rs)

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* refactor: deduplicate shared helpers and remove dead code

Extract floor_char_boundary and llm_signals_completion into src/util.rs,
unifying diverging phrase lists from agent/worker.rs and worker/runtime.rs.
Remove dead RespondResult::usage(), duplicate PROTECTED_IDENTITY_FILES
constant, double LeakDetector scanning in WebLogLayer, and invalid
0.0.0.0 origin from WebSocket allow list.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix: address PR review findings and CI test failures

- Fix record_failed_approve: .truncate(true) wiped the attempts file
  before reading, so failed pairing attempts never accumulated and
  rate limiting never triggered.
- Guard wizard WASM test: skip gracefully when channel build artifacts
  are absent (CI doesn't compile wasm32-wasip2 targets).
- Fix DNS rebinding check: use port 0 instead of hardcoded 443, since
  the port is irrelevant for hostname resolution.
- Remove hardcoded CORS port 3001: the dynamic addr.port() entries
  already cover the actual server port.
- Require WebSocket Origin header: reject connections that omit it
  entirely, since browsers always send Origin for WS upgrades and a
  missing header indicates a non-browser client bypassing the check.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix: address second round of PR review findings

- store.rs: reintroduce file locking around read-modify-write in
  record_failed_approve (concurrent callers could clobber each other).
- sse.rs: replace load+check+fetch_add with atomic fetch_update in both
  subscribe_raw() and subscribe() to prevent overshooting max_connections.
- ws.rs: decrement WS tracker before early return when subscribe_raw()
  returns None (connection limit reached), fixing a counter leak.
- server.rs: parse WS Origin host exactly instead of prefix matching,
  preventing bypass via crafted origins like http://localhost.evil.com.
- workspace_integration.rs: skip tests gracefully when Postgres is
  unreachable instead of panicking (fixes 10 CI failures).

Co-Authored-By: Claude Opus 4.6 <[email protected]>

* fix: add Origin header to WS integration tests

The Origin header requirement added in a3b0190 broke the WS gateway
integration tests. Test clients now send Origin: http://127.0.0.1:{port}
to match the server's localhost validation.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

---------

Co-authored-by: Claude Opus 4.6 <[email protected]>
2026-02-13 05:25:20 +00:00

1297 lines
46 KiB
Rust

//! Configuration for IronClaw.
//!
//! Settings are loaded with priority: env var > database > default.
//! The database replaces the old `settings.json` file for all settings
//! except the 4 bootstrap fields (database_url, pool_size, secrets key
//! source, onboard_completed) which live in `~/.ironclaw/bootstrap.json`.
use std::path::PathBuf;
use std::time::Duration;
use secrecy::{ExposeSecret, SecretString};
use crate::error::ConfigError;
use crate::settings::Settings;
/// Main configuration for the agent.
#[derive(Debug, Clone)]
pub struct Config {
pub database: DatabaseConfig,
pub llm: LlmConfig,
pub embeddings: EmbeddingsConfig,
pub tunnel: TunnelConfig,
pub channels: ChannelsConfig,
pub agent: AgentConfig,
pub safety: SafetyConfig,
pub wasm: WasmConfig,
pub secrets: SecretsConfig,
pub builder: BuilderModeConfig,
pub heartbeat: HeartbeatConfig,
pub routines: RoutineConfig,
pub sandbox: SandboxModeConfig,
pub claude_code: ClaudeCodeConfig,
}
impl Config {
/// Load configuration from environment variables and the database.
///
/// Priority: env var > DB settings > default.
/// This is the primary way to load config after DB is connected.
pub async fn from_db(
store: &crate::history::Store,
user_id: &str,
bootstrap: &crate::bootstrap::BootstrapConfig,
) -> Result<Self, ConfigError> {
let _ = dotenvy::dotenv();
// Load all settings from DB into a Settings struct
let db_settings = match store.get_all_settings(user_id).await {
Ok(map) => Settings::from_db_map(&map),
Err(e) => {
tracing::warn!("Failed to load settings from DB, using defaults: {}", e);
Settings::default()
}
};
Self::build(bootstrap, &db_settings).await
}
/// Load configuration from environment variables only (no database).
///
/// Used during early startup before the database is connected,
/// and by CLI commands that don't have DB access.
/// Falls back to legacy `settings.json` on disk if present.
pub async fn from_env() -> Result<Self, ConfigError> {
let _ = dotenvy::dotenv();
let bootstrap = crate::bootstrap::BootstrapConfig::load();
let settings = Settings::load();
Self::build(&bootstrap, &settings).await
}
/// Build config from bootstrap + settings (shared by from_env and from_db).
async fn build(
bootstrap: &crate::bootstrap::BootstrapConfig,
settings: &Settings,
) -> Result<Self, ConfigError> {
Ok(Self {
database: DatabaseConfig::resolve(bootstrap)?,
llm: LlmConfig::resolve(settings)?,
embeddings: EmbeddingsConfig::resolve(settings)?,
tunnel: TunnelConfig::resolve(settings)?,
channels: ChannelsConfig::resolve(settings)?,
agent: AgentConfig::resolve(settings)?,
safety: SafetyConfig::resolve()?,
wasm: WasmConfig::resolve()?,
secrets: SecretsConfig::resolve(bootstrap).await?,
builder: BuilderModeConfig::resolve()?,
heartbeat: HeartbeatConfig::resolve(settings)?,
routines: RoutineConfig::resolve()?,
sandbox: SandboxModeConfig::resolve()?,
claude_code: ClaudeCodeConfig::resolve()?,
})
}
}
/// Tunnel configuration for exposing the agent to the internet.
///
/// Used by channels and tools that need public webhook endpoints.
/// The tunnel URL is shared across all channels (Telegram, Slack, etc.).
#[derive(Debug, Clone, Default)]
pub struct TunnelConfig {
/// Public URL from tunnel provider (e.g., "https://abc123.ngrok.io").
pub public_url: Option<String>,
}
impl TunnelConfig {
fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
let public_url = optional_env("TUNNEL_URL")?
.or_else(|| settings.tunnel.public_url.clone().filter(|s| !s.is_empty()));
if let Some(ref url) = public_url {
if !url.starts_with("https://") {
return Err(ConfigError::InvalidValue {
key: "TUNNEL_URL".to_string(),
message: "must start with https:// (webhooks require HTTPS)".to_string(),
});
}
}
Ok(Self { public_url })
}
/// Check if a tunnel is configured.
pub fn is_enabled(&self) -> bool {
self.public_url.is_some()
}
/// Get the webhook URL for a given path.
pub fn webhook_url(&self, path: &str) -> Option<String> {
self.public_url.as_ref().map(|base| {
let base = base.trim_end_matches('/');
let path = path.trim_start_matches('/');
format!("{}/{}", base, path)
})
}
}
/// Database configuration.
#[derive(Debug, Clone)]
pub struct DatabaseConfig {
pub url: SecretString,
pub pool_size: usize,
}
impl DatabaseConfig {
fn resolve(bootstrap: &crate::bootstrap::BootstrapConfig) -> Result<Self, ConfigError> {
let url = optional_env("DATABASE_URL")?
.or_else(|| bootstrap.database_url.clone())
.ok_or_else(|| ConfigError::MissingRequired {
key: "database_url".to_string(),
hint: "Run 'ironclaw onboard' or set DATABASE_URL environment variable".to_string(),
})?;
let pool_size = optional_env("DATABASE_POOL_SIZE")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "DATABASE_POOL_SIZE".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.or(bootstrap.database_pool_size)
.unwrap_or(10);
Ok(Self {
url: SecretString::from(url),
pool_size,
})
}
/// Get the database URL (exposes the secret).
pub fn url(&self) -> &str {
self.url.expose_secret()
}
}
/// Which LLM backend to use.
///
/// Defaults to `NearAi` to keep IronClaw close to the NEAR ecosystem.
/// Users can override with `LLM_BACKEND` env var to use their own API keys.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
pub enum LlmBackend {
/// NEAR AI proxy (default) -- session or API key auth
#[default]
NearAi,
/// Direct OpenAI API
OpenAi,
/// Direct Anthropic API
Anthropic,
/// Local Ollama instance
Ollama,
/// Any OpenAI-compatible endpoint (e.g. vLLM, LiteLLM, Together)
OpenAiCompatible,
}
impl std::str::FromStr for LlmBackend {
type Err = String;
fn from_str(s: &str) -> Result<Self, Self::Err> {
match s.to_lowercase().as_str() {
"nearai" | "near_ai" | "near" => Ok(Self::NearAi),
"openai" | "open_ai" => Ok(Self::OpenAi),
"anthropic" | "claude" => Ok(Self::Anthropic),
"ollama" => Ok(Self::Ollama),
"openai_compatible" | "openai-compatible" | "compatible" => Ok(Self::OpenAiCompatible),
_ => Err(format!(
"invalid LLM backend '{}', expected one of: nearai, openai, anthropic, ollama, openai_compatible",
s
)),
}
}
}
impl std::fmt::Display for LlmBackend {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Self::NearAi => write!(f, "nearai"),
Self::OpenAi => write!(f, "openai"),
Self::Anthropic => write!(f, "anthropic"),
Self::Ollama => write!(f, "ollama"),
Self::OpenAiCompatible => write!(f, "openai_compatible"),
}
}
}
/// Configuration for direct OpenAI API access.
#[derive(Debug, Clone)]
pub struct OpenAiDirectConfig {
pub api_key: SecretString,
pub model: String,
}
/// Configuration for direct Anthropic API access.
#[derive(Debug, Clone)]
pub struct AnthropicDirectConfig {
pub api_key: SecretString,
pub model: String,
}
/// Configuration for local Ollama.
#[derive(Debug, Clone)]
pub struct OllamaConfig {
pub base_url: String,
pub model: String,
}
/// Configuration for any OpenAI-compatible endpoint.
#[derive(Debug, Clone)]
pub struct OpenAiCompatibleConfig {
pub base_url: String,
pub api_key: Option<SecretString>,
pub model: String,
}
/// LLM provider configuration.
///
/// NEAR AI remains the default backend. Users can switch to other providers
/// by setting `LLM_BACKEND` (e.g. `openai`, `anthropic`, `ollama`).
#[derive(Debug, Clone)]
pub struct LlmConfig {
/// Which backend to use (default: NearAi)
pub backend: LlmBackend,
/// NEAR AI config (always populated for NEAR AI embeddings, etc.)
pub nearai: NearAiConfig,
/// Direct OpenAI config (populated when backend=openai)
pub openai: Option<OpenAiDirectConfig>,
/// Direct Anthropic config (populated when backend=anthropic)
pub anthropic: Option<AnthropicDirectConfig>,
/// Ollama config (populated when backend=ollama)
pub ollama: Option<OllamaConfig>,
/// OpenAI-compatible config (populated when backend=openai_compatible)
pub openai_compatible: Option<OpenAiCompatibleConfig>,
}
/// API mode for NEAR AI.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
pub enum NearAiApiMode {
/// Use the Responses API (chat-api proxy) - session-based auth
#[default]
Responses,
/// Use the Chat Completions API (cloud-api) - API key auth
ChatCompletions,
}
impl std::str::FromStr for NearAiApiMode {
type Err = String;
fn from_str(s: &str) -> Result<Self, Self::Err> {
match s.to_lowercase().as_str() {
"responses" | "response" => Ok(Self::Responses),
"chat_completions" | "chatcompletions" | "chat" | "completions" => {
Ok(Self::ChatCompletions)
}
_ => Err(format!(
"invalid API mode '{}', expected 'responses' or 'chat_completions'",
s
)),
}
}
}
/// NEAR AI chat-api configuration.
#[derive(Debug, Clone)]
pub struct NearAiConfig {
/// Model to use (e.g., "claude-3-5-sonnet-20241022", "gpt-4o")
pub model: String,
/// Base URL for the NEAR AI API (default: https://api.near.ai)
pub base_url: String,
/// Base URL for auth/refresh endpoints (default: https://private.near.ai)
pub auth_base_url: String,
/// Path to session file (default: ~/.ironclaw/session.json)
pub session_path: PathBuf,
/// API mode: "responses" (chat-api) or "chat_completions" (cloud-api)
pub api_mode: NearAiApiMode,
/// API key for cloud-api (required for chat_completions mode)
pub api_key: Option<SecretString>,
}
impl LlmConfig {
fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
// Determine backend (default: NearAi)
let backend: LlmBackend = if let Some(b) = optional_env("LLM_BACKEND")? {
b.parse().map_err(|e| ConfigError::InvalidValue {
key: "LLM_BACKEND".to_string(),
message: e,
})?
} else {
LlmBackend::NearAi
};
// Always resolve NEAR AI config (used as fallback and for embeddings)
let nearai_api_key = optional_env("NEARAI_API_KEY")?.map(SecretString::from);
let api_mode = if let Some(mode_str) = optional_env("NEARAI_API_MODE")? {
mode_str.parse().map_err(|e| ConfigError::InvalidValue {
key: "NEARAI_API_MODE".to_string(),
message: e,
})?
} else if nearai_api_key.is_some() {
NearAiApiMode::ChatCompletions
} else {
NearAiApiMode::Responses
};
let nearai = NearAiConfig {
model: optional_env("NEARAI_MODEL")?
.or_else(|| settings.selected_model.clone())
.unwrap_or_else(|| {
"fireworks::accounts/fireworks/models/llama4-maverick-instruct-basic"
.to_string()
}),
base_url: optional_env("NEARAI_BASE_URL")?
.unwrap_or_else(|| "https://cloud-api.near.ai".to_string()),
auth_base_url: optional_env("NEARAI_AUTH_URL")?
.unwrap_or_else(|| "https://private.near.ai".to_string()),
session_path: optional_env("NEARAI_SESSION_PATH")?
.map(PathBuf::from)
.unwrap_or_else(default_session_path),
api_mode,
api_key: nearai_api_key,
};
// Resolve provider-specific configs based on backend
let openai = if backend == LlmBackend::OpenAi {
let api_key = optional_env("OPENAI_API_KEY")?
.map(SecretString::from)
.ok_or_else(|| ConfigError::MissingRequired {
key: "OPENAI_API_KEY".to_string(),
hint: "Set OPENAI_API_KEY when LLM_BACKEND=openai".to_string(),
})?;
let model = optional_env("OPENAI_MODEL")?.unwrap_or_else(|| "gpt-4o".to_string());
Some(OpenAiDirectConfig { api_key, model })
} else {
None
};
let anthropic = if backend == LlmBackend::Anthropic {
let api_key = optional_env("ANTHROPIC_API_KEY")?
.map(SecretString::from)
.ok_or_else(|| ConfigError::MissingRequired {
key: "ANTHROPIC_API_KEY".to_string(),
hint: "Set ANTHROPIC_API_KEY when LLM_BACKEND=anthropic".to_string(),
})?;
let model = optional_env("ANTHROPIC_MODEL")?
.unwrap_or_else(|| "claude-sonnet-4-20250514".to_string());
Some(AnthropicDirectConfig { api_key, model })
} else {
None
};
let ollama = if backend == LlmBackend::Ollama {
let base_url = optional_env("OLLAMA_BASE_URL")?
.unwrap_or_else(|| "http://localhost:11434".to_string());
let model = optional_env("OLLAMA_MODEL")?.unwrap_or_else(|| "llama3".to_string());
Some(OllamaConfig { base_url, model })
} else {
None
};
let openai_compatible = if backend == LlmBackend::OpenAiCompatible {
let base_url =
optional_env("LLM_BASE_URL")?.ok_or_else(|| ConfigError::MissingRequired {
key: "LLM_BASE_URL".to_string(),
hint: "Set LLM_BASE_URL when LLM_BACKEND=openai_compatible".to_string(),
})?;
let api_key = optional_env("LLM_API_KEY")?.map(SecretString::from);
let model = optional_env("LLM_MODEL")?.unwrap_or_else(|| "default".to_string());
Some(OpenAiCompatibleConfig {
base_url,
api_key,
model,
})
} else {
None
};
Ok(Self {
backend,
nearai,
openai,
anthropic,
ollama,
openai_compatible,
})
}
}
/// Embeddings provider configuration.
#[derive(Debug, Clone)]
pub struct EmbeddingsConfig {
/// Whether embeddings are enabled.
pub enabled: bool,
/// Provider to use: "openai" or "nearai"
pub provider: String,
/// OpenAI API key (for OpenAI provider).
pub openai_api_key: Option<SecretString>,
/// Model to use for embeddings.
pub model: String,
}
impl Default for EmbeddingsConfig {
fn default() -> Self {
Self {
enabled: false,
provider: "openai".to_string(),
openai_api_key: None,
model: "text-embedding-3-small".to_string(),
}
}
}
impl EmbeddingsConfig {
fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
let openai_api_key = optional_env("OPENAI_API_KEY")?.map(SecretString::from);
let provider = optional_env("EMBEDDING_PROVIDER")?
.unwrap_or_else(|| settings.embeddings.provider.clone());
let model =
optional_env("EMBEDDING_MODEL")?.unwrap_or_else(|| settings.embeddings.model.clone());
let enabled = optional_env("EMBEDDING_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "EMBEDDING_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or_else(|| settings.embeddings.enabled || openai_api_key.is_some());
Ok(Self {
enabled,
provider,
openai_api_key,
model,
})
}
/// Get the OpenAI API key if configured.
pub fn openai_api_key(&self) -> Option<&str> {
self.openai_api_key.as_ref().map(|s| s.expose_secret())
}
}
/// Get the default session file path (~/.ironclaw/session.json).
fn default_session_path() -> PathBuf {
dirs::home_dir()
.unwrap_or_else(|| PathBuf::from("."))
.join(".ironclaw")
.join("session.json")
}
/// Channel configurations.
#[derive(Debug, Clone)]
pub struct ChannelsConfig {
pub cli: CliConfig,
pub http: Option<HttpConfig>,
pub gateway: Option<GatewayConfig>,
/// Directory containing WASM channel modules (default: ~/.ironclaw/channels/).
pub wasm_channels_dir: std::path::PathBuf,
/// Whether WASM channels are enabled.
pub wasm_channels_enabled: bool,
/// Telegram owner user ID. When set, the bot only responds to this user.
pub telegram_owner_id: Option<i64>,
}
#[derive(Debug, Clone)]
pub struct CliConfig {
pub enabled: bool,
}
#[derive(Debug, Clone)]
pub struct HttpConfig {
pub host: String,
pub port: u16,
pub webhook_secret: Option<SecretString>,
pub user_id: String,
}
/// Web gateway configuration.
#[derive(Debug, Clone)]
pub struct GatewayConfig {
pub host: String,
pub port: u16,
/// Bearer token for authentication. Random hex generated at startup if unset.
pub auth_token: Option<String>,
pub user_id: String,
}
impl ChannelsConfig {
fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
let http = if optional_env("HTTP_PORT")?.is_some() || optional_env("HTTP_HOST")?.is_some() {
Some(HttpConfig {
host: optional_env("HTTP_HOST")?.unwrap_or_else(|| "0.0.0.0".to_string()),
port: optional_env("HTTP_PORT")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "HTTP_PORT".to_string(),
message: format!("must be a valid port number: {e}"),
})?
.unwrap_or(8080),
webhook_secret: optional_env("HTTP_WEBHOOK_SECRET")?.map(SecretString::from),
user_id: optional_env("HTTP_USER_ID")?.unwrap_or_else(|| "http".to_string()),
})
} else {
None
};
let gateway = if optional_env("GATEWAY_ENABLED")?
.map(|s| s.to_lowercase() == "true" || s == "1")
.unwrap_or(true)
{
Some(GatewayConfig {
host: optional_env("GATEWAY_HOST")?.unwrap_or_else(|| "127.0.0.1".to_string()),
port: optional_env("GATEWAY_PORT")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "GATEWAY_PORT".to_string(),
message: format!("must be a valid port number: {e}"),
})?
.unwrap_or(3000),
auth_token: optional_env("GATEWAY_AUTH_TOKEN")?,
user_id: optional_env("GATEWAY_USER_ID")?.unwrap_or_else(|| "default".to_string()),
})
} else {
None
};
let cli_enabled = optional_env("CLI_ENABLED")?
.map(|s| s.to_lowercase() != "false" && s != "0")
.unwrap_or(true);
Ok(Self {
cli: CliConfig {
enabled: cli_enabled,
},
http,
gateway,
wasm_channels_dir: optional_env("WASM_CHANNELS_DIR")?
.map(PathBuf::from)
.unwrap_or_else(default_channels_dir),
wasm_channels_enabled: optional_env("WASM_CHANNELS_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "WASM_CHANNELS_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
telegram_owner_id: optional_env("TELEGRAM_OWNER_ID")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "TELEGRAM_OWNER_ID".to_string(),
message: format!("must be an integer: {e}"),
})?
.or(settings.channels.telegram_owner_id),
})
}
}
/// Get the default channels directory (~/.ironclaw/channels/).
fn default_channels_dir() -> PathBuf {
dirs::home_dir()
.unwrap_or_else(|| PathBuf::from("."))
.join(".ironclaw")
.join("channels")
}
/// Agent behavior configuration.
#[derive(Debug, Clone)]
pub struct AgentConfig {
pub name: String,
pub max_parallel_jobs: usize,
pub job_timeout: Duration,
pub stuck_threshold: Duration,
pub repair_check_interval: Duration,
pub max_repair_attempts: u32,
/// Whether to use planning before tool execution.
pub use_planning: bool,
/// Session idle timeout. Sessions inactive longer than this are pruned.
pub session_idle_timeout: Duration,
/// Allow chat to use filesystem/shell tools directly (bypass sandbox).
pub allow_local_tools: bool,
}
impl AgentConfig {
fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
Ok(Self {
name: optional_env("AGENT_NAME")?.unwrap_or_else(|| settings.agent.name.clone()),
max_parallel_jobs: optional_env("AGENT_MAX_PARALLEL_JOBS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "AGENT_MAX_PARALLEL_JOBS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.agent.max_parallel_jobs as usize),
job_timeout: Duration::from_secs(
optional_env("AGENT_JOB_TIMEOUT_SECS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "AGENT_JOB_TIMEOUT_SECS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.agent.job_timeout_secs),
),
stuck_threshold: Duration::from_secs(
optional_env("AGENT_STUCK_THRESHOLD_SECS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "AGENT_STUCK_THRESHOLD_SECS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.agent.stuck_threshold_secs),
),
repair_check_interval: Duration::from_secs(
optional_env("SELF_REPAIR_CHECK_INTERVAL_SECS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "SELF_REPAIR_CHECK_INTERVAL_SECS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.agent.repair_check_interval_secs),
),
max_repair_attempts: optional_env("SELF_REPAIR_MAX_ATTEMPTS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "SELF_REPAIR_MAX_ATTEMPTS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.agent.max_repair_attempts),
use_planning: optional_env("AGENT_USE_PLANNING")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "AGENT_USE_PLANNING".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(settings.agent.use_planning),
session_idle_timeout: Duration::from_secs(
optional_env("SESSION_IDLE_TIMEOUT_SECS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "SESSION_IDLE_TIMEOUT_SECS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.agent.session_idle_timeout_secs),
),
allow_local_tools: optional_env("ALLOW_LOCAL_TOOLS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "ALLOW_LOCAL_TOOLS".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(false),
})
}
}
/// Safety configuration.
#[derive(Debug, Clone)]
pub struct SafetyConfig {
pub max_output_length: usize,
pub injection_check_enabled: bool,
}
impl SafetyConfig {
fn resolve() -> Result<Self, ConfigError> {
Ok(Self {
max_output_length: parse_optional_env("SAFETY_MAX_OUTPUT_LENGTH", 100_000)?,
injection_check_enabled: optional_env("SAFETY_INJECTION_CHECK_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "SAFETY_INJECTION_CHECK_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
})
}
}
/// WASM sandbox configuration.
#[derive(Debug, Clone)]
pub struct WasmConfig {
/// Whether WASM tool execution is enabled.
pub enabled: bool,
/// Directory containing installed WASM tools (default: ~/.ironclaw/tools/).
pub tools_dir: PathBuf,
/// Default memory limit in bytes (default: 10 MB).
pub default_memory_limit: u64,
/// Default execution timeout in seconds (default: 60).
pub default_timeout_secs: u64,
/// Default fuel limit for CPU metering (default: 10M).
pub default_fuel_limit: u64,
/// Whether to cache compiled modules.
pub cache_compiled: bool,
/// Directory for compiled module cache.
pub cache_dir: Option<PathBuf>,
}
/// Secrets management configuration.
#[derive(Clone, Default)]
pub struct SecretsConfig {
/// Master key for encrypting secrets.
pub master_key: Option<SecretString>,
/// Whether secrets management is enabled.
pub enabled: bool,
/// Source of the master key.
pub source: crate::settings::KeySource,
}
impl std::fmt::Debug for SecretsConfig {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_struct("SecretsConfig")
.field("master_key", &self.master_key.is_some())
.field("enabled", &self.enabled)
.field("source", &self.source)
.finish()
}
}
impl SecretsConfig {
async fn resolve(bootstrap: &crate::bootstrap::BootstrapConfig) -> Result<Self, ConfigError> {
use crate::settings::KeySource;
let (master_key, source) = if let Some(env_key) = optional_env("SECRETS_MASTER_KEY")? {
(Some(SecretString::from(env_key)), KeySource::Env)
} else {
match bootstrap.secrets_master_key_source {
KeySource::Keychain => {
// Try to load from OS keychain (async on Linux)
match crate::secrets::keychain::get_master_key().await {
Ok(key_bytes) => {
let key_hex: String =
key_bytes.iter().map(|b| format!("{:02x}", b)).collect();
(Some(SecretString::from(key_hex)), KeySource::Keychain)
}
Err(_) => {
// Keychain configured but key not found
// This might happen if keychain was cleared
tracing::warn!(
"Secrets configured for keychain but key not found. \
Run 'ironclaw onboard' to reconfigure."
);
(None, KeySource::None)
}
}
}
KeySource::Env => {
tracing::warn!(
"Secrets configured for env var but SECRETS_MASTER_KEY not set."
);
(None, KeySource::None)
}
KeySource::None => (None, KeySource::None),
}
};
let enabled = master_key.is_some();
if let Some(ref key) = master_key {
if key.expose_secret().len() < 32 {
return Err(ConfigError::InvalidValue {
key: "SECRETS_MASTER_KEY".to_string(),
message: "must be at least 32 bytes for AES-256-GCM".to_string(),
});
}
}
Ok(Self {
master_key,
enabled,
source,
})
}
/// Get the master key if configured.
pub fn master_key(&self) -> Option<&SecretString> {
self.master_key.as_ref()
}
}
impl Default for WasmConfig {
fn default() -> Self {
Self {
enabled: true,
tools_dir: default_tools_dir(),
default_memory_limit: 10 * 1024 * 1024, // 10 MB
default_timeout_secs: 60,
default_fuel_limit: 10_000_000,
cache_compiled: true,
cache_dir: None,
}
}
}
/// Get the default tools directory (~/.ironclaw/tools/).
fn default_tools_dir() -> PathBuf {
dirs::home_dir()
.unwrap_or_else(|| PathBuf::from("."))
.join(".ironclaw")
.join("tools")
}
impl WasmConfig {
fn resolve() -> Result<Self, ConfigError> {
Ok(Self {
enabled: optional_env("WASM_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "WASM_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
tools_dir: optional_env("WASM_TOOLS_DIR")?
.map(PathBuf::from)
.unwrap_or_else(default_tools_dir),
default_memory_limit: parse_optional_env(
"WASM_DEFAULT_MEMORY_LIMIT",
10 * 1024 * 1024,
)?,
default_timeout_secs: parse_optional_env("WASM_DEFAULT_TIMEOUT_SECS", 60)?,
default_fuel_limit: parse_optional_env("WASM_DEFAULT_FUEL_LIMIT", 10_000_000)?,
cache_compiled: optional_env("WASM_CACHE_COMPILED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "WASM_CACHE_COMPILED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
cache_dir: optional_env("WASM_CACHE_DIR")?.map(PathBuf::from),
})
}
/// Convert to WasmRuntimeConfig.
pub fn to_runtime_config(&self) -> crate::tools::wasm::WasmRuntimeConfig {
use crate::tools::wasm::{FuelConfig, ResourceLimits, WasmRuntimeConfig};
use std::time::Duration;
WasmRuntimeConfig {
default_limits: ResourceLimits {
memory_bytes: self.default_memory_limit,
fuel: self.default_fuel_limit,
timeout: Duration::from_secs(self.default_timeout_secs),
},
fuel_config: FuelConfig {
initial_fuel: self.default_fuel_limit,
enabled: true,
},
cache_compiled: self.cache_compiled,
cache_dir: self.cache_dir.clone(),
optimization_level: wasmtime::OptLevel::Speed,
}
}
}
/// Builder mode configuration.
#[derive(Debug, Clone)]
pub struct BuilderModeConfig {
/// Whether the software builder tool is enabled.
pub enabled: bool,
/// Directory for build artifacts (default: temp dir).
pub build_dir: Option<PathBuf>,
/// Maximum iterations for the build loop.
pub max_iterations: u32,
/// Build timeout in seconds.
pub timeout_secs: u64,
/// Whether to automatically register built WASM tools.
pub auto_register: bool,
}
impl Default for BuilderModeConfig {
fn default() -> Self {
Self {
enabled: true,
build_dir: None,
max_iterations: 20,
timeout_secs: 600,
auto_register: true,
}
}
}
impl BuilderModeConfig {
fn resolve() -> Result<Self, ConfigError> {
Ok(Self {
enabled: optional_env("BUILDER_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "BUILDER_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
build_dir: optional_env("BUILDER_DIR")?.map(PathBuf::from),
max_iterations: parse_optional_env("BUILDER_MAX_ITERATIONS", 20)?,
timeout_secs: parse_optional_env("BUILDER_TIMEOUT_SECS", 600)?,
auto_register: optional_env("BUILDER_AUTO_REGISTER")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "BUILDER_AUTO_REGISTER".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
})
}
/// Convert to BuilderConfig for the builder tool.
pub fn to_builder_config(&self) -> crate::tools::BuilderConfig {
crate::tools::BuilderConfig {
build_dir: self.build_dir.clone().unwrap_or_else(std::env::temp_dir),
max_iterations: self.max_iterations,
timeout: Duration::from_secs(self.timeout_secs),
cleanup_on_failure: true,
validate_wasm: true,
run_tests: true,
auto_register: self.auto_register,
wasm_output_dir: None,
}
}
}
/// Heartbeat configuration.
#[derive(Debug, Clone)]
pub struct HeartbeatConfig {
/// Whether heartbeat is enabled.
pub enabled: bool,
/// Interval between heartbeat checks in seconds.
pub interval_secs: u64,
/// Channel to notify on heartbeat findings.
pub notify_channel: Option<String>,
/// User ID to notify on heartbeat findings.
pub notify_user: Option<String>,
}
impl Default for HeartbeatConfig {
fn default() -> Self {
Self {
enabled: false,
interval_secs: 1800, // 30 minutes
notify_channel: None,
notify_user: None,
}
}
}
impl HeartbeatConfig {
fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
Ok(Self {
enabled: optional_env("HEARTBEAT_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "HEARTBEAT_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(settings.heartbeat.enabled),
interval_secs: optional_env("HEARTBEAT_INTERVAL_SECS")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "HEARTBEAT_INTERVAL_SECS".to_string(),
message: format!("must be a positive integer: {e}"),
})?
.unwrap_or(settings.heartbeat.interval_secs),
notify_channel: optional_env("HEARTBEAT_NOTIFY_CHANNEL")?
.or_else(|| settings.heartbeat.notify_channel.clone()),
notify_user: optional_env("HEARTBEAT_NOTIFY_USER")?
.or_else(|| settings.heartbeat.notify_user.clone()),
})
}
}
/// Routines configuration.
#[derive(Debug, Clone)]
pub struct RoutineConfig {
/// Whether the routines system is enabled.
pub enabled: bool,
/// How often (seconds) to poll for cron routines that need firing.
pub cron_check_interval_secs: u64,
/// Max routines executing concurrently across all users.
pub max_concurrent_routines: usize,
/// Default cooldown between fires (seconds).
pub default_cooldown_secs: u64,
/// Max output tokens for lightweight routine LLM calls.
pub max_lightweight_tokens: u32,
}
impl Default for RoutineConfig {
fn default() -> Self {
Self {
enabled: true,
cron_check_interval_secs: 15,
max_concurrent_routines: 10,
default_cooldown_secs: 300,
max_lightweight_tokens: 4096,
}
}
}
impl RoutineConfig {
fn resolve() -> Result<Self, ConfigError> {
Ok(Self {
enabled: optional_env("ROUTINES_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "ROUTINES_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
cron_check_interval_secs: parse_optional_env("ROUTINES_CRON_INTERVAL", 15)?,
max_concurrent_routines: parse_optional_env("ROUTINES_MAX_CONCURRENT", 10)?,
default_cooldown_secs: parse_optional_env("ROUTINES_DEFAULT_COOLDOWN", 300)?,
max_lightweight_tokens: parse_optional_env("ROUTINES_MAX_TOKENS", 4096)?,
})
}
}
/// Docker sandbox configuration.
#[derive(Debug, Clone)]
pub struct SandboxModeConfig {
/// Whether the Docker sandbox is enabled.
pub enabled: bool,
/// Sandbox policy: "readonly", "workspace_write", or "full_access".
pub policy: String,
/// Command timeout in seconds.
pub timeout_secs: u64,
/// Memory limit in megabytes.
pub memory_limit_mb: u64,
/// CPU shares (relative weight).
pub cpu_shares: u32,
/// Docker image for the sandbox.
pub image: String,
/// Whether to auto-pull the image if not found.
pub auto_pull_image: bool,
/// Additional domains to allow through the network proxy.
pub extra_allowed_domains: Vec<String>,
}
impl Default for SandboxModeConfig {
fn default() -> Self {
Self {
enabled: true,
policy: "readonly".to_string(),
timeout_secs: 120,
memory_limit_mb: 2048,
cpu_shares: 1024,
image: "ghcr.io/nearai/sandbox:latest".to_string(),
auto_pull_image: true,
extra_allowed_domains: Vec::new(),
}
}
}
impl SandboxModeConfig {
fn resolve() -> Result<Self, ConfigError> {
let extra_domains = optional_env("SANDBOX_EXTRA_DOMAINS")?
.map(|s| s.split(',').map(|d| d.trim().to_string()).collect())
.unwrap_or_default();
Ok(Self {
enabled: optional_env("SANDBOX_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "SANDBOX_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
policy: optional_env("SANDBOX_POLICY")?.unwrap_or_else(|| "readonly".to_string()),
timeout_secs: parse_optional_env("SANDBOX_TIMEOUT_SECS", 120)?,
memory_limit_mb: parse_optional_env("SANDBOX_MEMORY_LIMIT_MB", 2048)?,
cpu_shares: parse_optional_env("SANDBOX_CPU_SHARES", 1024)?,
image: optional_env("SANDBOX_IMAGE")?
.unwrap_or_else(|| "ghcr.io/nearai/sandbox:latest".to_string()),
auto_pull_image: optional_env("SANDBOX_AUTO_PULL")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "SANDBOX_AUTO_PULL".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(true),
extra_allowed_domains: extra_domains,
})
}
/// Convert to SandboxConfig for the sandbox module.
pub fn to_sandbox_config(&self) -> crate::sandbox::SandboxConfig {
use crate::sandbox::SandboxPolicy;
use std::time::Duration;
let policy = self.policy.parse().unwrap_or(SandboxPolicy::ReadOnly);
let mut allowlist = crate::sandbox::default_allowlist();
allowlist.extend(self.extra_allowed_domains.clone());
crate::sandbox::SandboxConfig {
enabled: self.enabled,
policy,
timeout: Duration::from_secs(self.timeout_secs),
memory_limit_mb: self.memory_limit_mb,
cpu_shares: self.cpu_shares,
network_allowlist: allowlist,
image: self.image.clone(),
auto_pull_image: self.auto_pull_image,
proxy_port: 0, // Auto-assign
}
}
}
/// Claude Code sandbox configuration.
#[derive(Debug, Clone)]
pub struct ClaudeCodeConfig {
/// Whether Claude Code sandbox mode is available.
pub enabled: bool,
/// Host directory containing Claude auth session (mounted read-only).
pub config_dir: std::path::PathBuf,
/// Claude model to use (e.g. "sonnet", "opus").
pub model: String,
/// Maximum agentic turns before stopping.
pub max_turns: u32,
/// Memory limit in MB for Claude Code containers (heavier than workers).
pub memory_limit_mb: u64,
/// Allowed tool patterns for Claude Code permission settings.
///
/// Written to `/workspace/.claude/settings.json` before spawning the CLI.
/// Provides defense-in-depth: only explicitly listed tools are auto-approved.
/// Any new/unknown tools would require interactive approval (which times out
/// in the non-interactive container, failing safely).
///
/// Patterns follow Claude Code syntax: `"Bash(*)"`, `"Read"`, `"Edit(*)"`, etc.
pub allowed_tools: Vec<String>,
}
/// Default allowed tools for Claude Code inside containers.
///
/// These cover all standard Claude Code tools needed for autonomous operation.
/// The Docker container provides the primary security boundary; this allowlist
/// provides defense-in-depth by preventing any future unknown tools from being
/// silently auto-approved.
fn default_claude_code_allowed_tools() -> Vec<String> {
[
"Bash(*)",
"Read",
"Edit(*)",
"Glob",
"Grep",
"WebFetch(*)",
"Task(*)",
]
.into_iter()
.map(String::from)
.collect()
}
impl Default for ClaudeCodeConfig {
fn default() -> Self {
Self {
enabled: false,
config_dir: dirs::home_dir()
.unwrap_or_else(|| std::path::PathBuf::from("."))
.join(".claude"),
model: "sonnet".to_string(),
max_turns: 50,
memory_limit_mb: 4096,
allowed_tools: default_claude_code_allowed_tools(),
}
}
}
impl ClaudeCodeConfig {
/// Load from environment variables only (used inside containers where
/// there is no database or full config).
pub fn from_env() -> Self {
match Self::resolve() {
Ok(c) => c,
Err(e) => {
tracing::warn!("Failed to resolve ClaudeCodeConfig: {e}, using defaults");
Self::default()
}
}
}
fn resolve() -> Result<Self, ConfigError> {
let defaults = Self::default();
Ok(Self {
enabled: optional_env("CLAUDE_CODE_ENABLED")?
.map(|s| s.parse())
.transpose()
.map_err(|e| ConfigError::InvalidValue {
key: "CLAUDE_CODE_ENABLED".to_string(),
message: format!("must be 'true' or 'false': {e}"),
})?
.unwrap_or(defaults.enabled),
config_dir: optional_env("CLAUDE_CONFIG_DIR")?
.map(std::path::PathBuf::from)
.unwrap_or(defaults.config_dir),
model: optional_env("CLAUDE_CODE_MODEL")?.unwrap_or(defaults.model),
max_turns: parse_optional_env("CLAUDE_CODE_MAX_TURNS", defaults.max_turns)?,
memory_limit_mb: parse_optional_env(
"CLAUDE_CODE_MEMORY_LIMIT_MB",
defaults.memory_limit_mb,
)?,
allowed_tools: optional_env("CLAUDE_CODE_ALLOWED_TOOLS")?
.map(|s| {
s.split(',')
.map(|t| t.trim().to_string())
.filter(|t| !t.is_empty())
.collect()
})
.unwrap_or(defaults.allowed_tools),
})
}
}
// Helper functions
fn optional_env(key: &str) -> Result<Option<String>, ConfigError> {
match std::env::var(key) {
Ok(val) if val.is_empty() => Ok(None),
Ok(val) => Ok(Some(val)),
Err(std::env::VarError::NotPresent) => Ok(None),
Err(e) => Err(ConfigError::ParseError(format!(
"failed to read {key}: {e}"
))),
}
}
fn parse_optional_env<T>(key: &str, default: T) -> Result<T, ConfigError>
where
T: std::str::FromStr,
T::Err: std::fmt::Display,
{
optional_env(key)?
.map(|s| {
s.parse().map_err(|e| ConfigError::InvalidValue {
key: key.to_string(),
message: format!("{e}"),
})
})
.transpose()
.map(|opt| opt.unwrap_or(default))
}