fix: persist OpenAI-compatible provider and respect embeddings disable (#177)

* fix: persist OpenAI-compatible provider and respect embeddings disable (#129)

Three interrelated bugs caused the agent to ignore user choices made
during onboarding when using an OpenAI-compatible LLM provider:

1. Session auth ran before DB config reload, so Config::from_env()
   defaulted to NearAi and attempted Clerk auth before the real
   backend was known. Moved session auth to after final config
   resolution.

2. EmbeddingsConfig::resolve() force-enabled embeddings whenever
   OPENAI_API_KEY was present, ignoring the user's explicit disable.
   Changed to respect the stored setting as source of truth.

3. LLM_BACKEND was not saved to the bootstrap .env file, so
   Config::from_env() always defaulted to NearAi before the DB
   was connected. Now saves LLM_BACKEND, LLM_BASE_URL, and
   OLLAMA_BASE_URL alongside the database bootstrap vars.

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

* fix: add SAFETY comments and sanitize .env value escaping

Address PR review feedback:

- Add SAFETY comments to all unsafe env var manipulation in config
  tests (gemini-code-assist).
- Escape backslashes and double quotes in save_bootstrap_env() to
  prevent env var injection via malicious URLs (gemini-code-assist).
- Add test verifying injection attempt is neutralized.

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

* fix: incorporate PR #138 changes (chat completions, model sorting, tool schemas)

Includes all changes from bigguybobby's PR #138:

- Use Chat Completions API for OpenAI-compatible providers (avoids
  Responses API assumptions like required tool call IDs)
- Fall back to settings.selected_model when LLM_MODEL env var is unset
- Update OpenAI model list (add gpt-5 family) with priority-based sorting
- Add is_openai_chat_model() filter with broader exclusion patterns
- Fix http tool: headers schema → array of {name,value}, body → string type,
  parse_headers_param() accepts both legacy object and array formats
- Fix json tool: data schema → string type, parse_json_input() normalizer,
  validate uses strict string-only check
- Add mutex-serialized config tests for env var manipulation
- Update NEAR AI config comment for accuracy

Co-Authored-By: Bobby (bigguybobby) <[email protected]>
Co-Authored-By: Claude Opus 4.6 <[email protected]>

---------

Co-authored-by: Illia Polosukhin <[email protected]>
Co-authored-by: Claude Opus 4.6 <[email protected]>
Co-authored-by: Bobby (bigguybobby) <[email protected]>
This commit is contained in:
Illia Polosukhin
2026-02-18 08:29:53 +00:00
committed by GitHub
co-authored by Illia Polosukhin Claude Opus 4.6 Bobby
parent c3340c60ef
commit 750a94030b
9 changed files with 527 additions and 55 deletions
+115 -18
View File
@@ -106,6 +106,46 @@ fn is_disallowed_ip(ip: &IpAddr) -> bool {
}
}
fn parse_headers_param(
headers: Option<&serde_json::Value>,
) -> Result<Vec<(String, String)>, ToolError> {
match headers {
None => Ok(Vec::new()),
Some(serde_json::Value::Object(map)) => {
let mut out = Vec::with_capacity(map.len());
for (k, v) in map {
let value = v.as_str().ok_or_else(|| {
ToolError::InvalidParameters(format!("header '{}' must have a string value", k))
})?;
out.push((k.clone(), value.to_string()));
}
Ok(out)
}
Some(serde_json::Value::Array(items)) => {
let mut out = Vec::with_capacity(items.len());
for (idx, item) in items.iter().enumerate() {
let obj = item.as_object().ok_or_else(|| {
ToolError::InvalidParameters(format!(
"headers[{}] must be an object with 'name' and 'value'",
idx
))
})?;
let name = obj.get("name").and_then(|v| v.as_str()).ok_or_else(|| {
ToolError::InvalidParameters(format!("headers[{}].name must be a string", idx))
})?;
let value = obj.get("value").and_then(|v| v.as_str()).ok_or_else(|| {
ToolError::InvalidParameters(format!("headers[{}].value must be a string", idx))
})?;
out.push((name.to_string(), value.to_string()));
}
Ok(out)
}
Some(_) => Err(ToolError::InvalidParameters(
"'headers' must be an object or an array of {name, value}".to_string(),
)),
}
}
impl Default for HttpTool {
fn default() -> Self {
Self::new()
@@ -136,12 +176,21 @@ impl Tool for HttpTool {
"description": "The URL to request"
},
"headers": {
"type": "object",
"additionalProperties": { "type": "string" },
"description": "HTTP headers to include"
"type": "array",
"description": "Optional headers as a list of {name, value} objects",
"items": {
"type": "object",
"properties": {
"name": { "type": "string" },
"value": { "type": "string" }
},
"required": ["name", "value"],
"additionalProperties": false
}
},
"body": {
"description": "Request body (for POST/PUT/PATCH)"
"type": "string",
"description": "Request body. Use plain text or serialized JSON."
},
"timeout_secs": {
"type": "integer",
@@ -165,14 +214,7 @@ impl Tool for HttpTool {
let parsed_url = validate_url(url)?;
// Parse headers
let headers: HashMap<String, String> = params
.get("headers")
.and_then(|v| serde_json::from_value(v.clone()).ok())
.unwrap_or_default();
let headers_vec: Vec<(String, String)> = headers
.iter()
.map(|(k, v)| (k.clone(), v.clone()))
.collect();
let headers_vec = parse_headers_param(params.get("headers"))?;
// Build request
let mut request = match method.to_uppercase().as_str() {
@@ -190,16 +232,31 @@ impl Tool for HttpTool {
};
// Add headers
for (key, value) in headers {
request = request.header(&key, &value);
for (key, value) in &headers_vec {
request = request.header(key.as_str(), value.as_str());
}
// Add body if present
let body_bytes = if let Some(body) = params.get("body") {
let bytes = serde_json::to_vec(body)
.map_err(|e| ToolError::InvalidParameters(format!("invalid body JSON: {}", e)))?;
request = request.json(body);
Some(bytes)
if let Some(body_str) = body.as_str() {
if let Ok(json_body) = serde_json::from_str::<serde_json::Value>(body_str) {
let bytes = serde_json::to_vec(&json_body).map_err(|e| {
ToolError::InvalidParameters(format!("invalid body JSON: {}", e))
})?;
request = request.json(&json_body);
Some(bytes)
} else {
let bytes = body_str.as_bytes().to_vec();
request = request.body(body_str.to_string());
Some(bytes)
}
} else {
let bytes = serde_json::to_vec(body).map_err(|e| {
ToolError::InvalidParameters(format!("invalid body JSON: {}", e))
})?;
request = request.json(body);
Some(bytes)
}
} else {
None
};
@@ -304,6 +361,20 @@ impl Tool for HttpTool {
mod tests {
use super::*;
#[test]
fn test_http_tool_schema_body_has_type() {
let tool = HttpTool::new();
let schema = tool.parameters_schema();
assert_eq!(schema["properties"]["body"]["type"], "string");
}
#[test]
fn test_http_tool_schema_headers_is_array() {
let tool = HttpTool::new();
let schema = tool.parameters_schema();
assert_eq!(schema["properties"]["headers"]["type"], "array");
}
#[test]
fn test_validate_url_rejects_http() {
let err = validate_url("http://example.com").unwrap_err();
@@ -363,4 +434,30 @@ mod tests {
// MAX_RESPONSE_SIZE should be 5 MB to prevent OOM while allowing typical API responses.
assert_eq!(MAX_RESPONSE_SIZE, 5 * 1024 * 1024);
}
#[test]
fn test_parse_headers_param_accepts_object_legacy_shape() {
let headers = serde_json::json!({"Authorization": "Bearer token"});
let parsed = parse_headers_param(Some(&headers)).unwrap();
assert_eq!(
parsed,
vec![("Authorization".to_string(), "Bearer token".to_string())]
);
}
#[test]
fn test_parse_headers_param_accepts_array_shape() {
let headers = serde_json::json!([
{"name": "Authorization", "value": "Bearer token"},
{"name": "X-Test", "value": "1"}
]);
let parsed = parse_headers_param(Some(&headers)).unwrap();
assert_eq!(
parsed,
vec![
("Authorization".to_string(), "Bearer token".to_string()),
("X-Test".to_string(), "1".to_string())
]
);
}
}
+39 -8
View File
@@ -28,7 +28,8 @@ impl Tool for JsonTool {
"description": "The JSON operation to perform"
},
"data": {
"description": "The JSON data to operate on (string for parse, object otherwise)"
"type": "string",
"description": "JSON input string. For query/stringify/validate, pass serialized JSON."
},
"path": {
"type": "string",
@@ -64,7 +65,8 @@ impl Tool for JsonTool {
parsed
}
"stringify" => {
let json_str = serde_json::to_string_pretty(data).map_err(|e| {
let value = parse_json_input(data)?;
let json_str = serde_json::to_string_pretty(&value).map_err(|e| {
ToolError::ExecutionFailed(format!("failed to stringify: {}", e))
})?;
@@ -75,14 +77,14 @@ impl Tool for JsonTool {
ToolError::InvalidParameters("missing 'path' parameter for query".to_string())
})?;
query_json(data, path)?
let value = parse_json_input(data)?;
query_json(&value, path)?
}
"validate" => {
let is_valid = if let Some(s) = data.as_str() {
serde_json::from_str::<serde_json::Value>(s).is_ok()
} else {
true // Already a valid JSON value
};
let is_valid = data
.as_str()
.map(|s| serde_json::from_str::<serde_json::Value>(s).is_ok())
.unwrap_or(false);
serde_json::json!({ "valid": is_valid })
}
@@ -102,6 +104,14 @@ impl Tool for JsonTool {
}
}
fn parse_json_input(data: &serde_json::Value) -> Result<serde_json::Value, ToolError> {
let json_str = data
.as_str()
.ok_or_else(|| ToolError::InvalidParameters("'data' must be a JSON string".to_string()))?;
serde_json::from_str(json_str)
.map_err(|e| ToolError::InvalidParameters(format!("invalid JSON input: {}", e)))
}
/// Simple JSONPath-like query implementation.
fn query_json(data: &serde_json::Value, path: &str) -> Result<serde_json::Value, ToolError> {
let mut current = data;
@@ -144,6 +154,13 @@ fn query_json(data: &serde_json::Value, path: &str) -> Result<serde_json::Value,
mod tests {
use super::*;
#[test]
fn test_json_tool_schema_data_has_type() {
let tool = JsonTool;
let schema = tool.parameters_schema();
assert_eq!(schema["properties"]["data"]["type"], "string");
}
#[test]
fn test_query_json() {
let data = serde_json::json!({
@@ -166,4 +183,18 @@ mod tests {
serde_json::json!(3)
);
}
#[test]
fn test_parse_json_input_accepts_valid_json_string() {
let input = serde_json::json!("{\"ok\":true}");
let parsed = parse_json_input(&input).unwrap();
assert_eq!(parsed, serde_json::json!({"ok": true}));
}
#[test]
fn test_parse_json_input_rejects_invalid_json_string() {
let input = serde_json::json!("{not valid json}");
let err = parse_json_input(&input).unwrap_err();
assert!(err.to_string().contains("invalid JSON input"));
}
}