| 1 | use std::collections::{BTreeMap, BTreeSet, HashMap}; |
| 2 | use std::fs; |
| 3 | use std::io::Read; |
| 4 | use std::path::{Component, Path, PathBuf}; |
| 5 | |
| 6 | use semver::Version; |
| 7 | use serde::{Deserialize, Serialize}; |
| 8 | use sha2::{Digest, Sha256}; |
| 9 | |
| 10 | use super::path_identity::metadata_is_link_or_reparse; |
| 11 | #[cfg(windows)] |
| 12 | use super::path_identity::windows_file_identity; |
| 13 | use crate::mcp::{McpServerConfig, is_relative_stdio_path_arg}; |
| 14 | |
| 15 | pub const CURRENT_SCHEMA_VERSION: u32 = 1; |
| 16 | const MAX_PLUGIN_NAME_CHARS: usize = 64; |
| 17 | const MAX_COMPONENT_PATHS: usize = 64; |
| 18 | const MAX_MANIFEST_BYTES: u64 = 1024 * 1024; |
| 19 | const MAX_HASHED_FILES: usize = 4_096; |
| 20 | const MAX_HASHED_BYTES: u64 = 64 * 1024 * 1024; |
| 21 | const MAX_MCP_ARGS: usize = 64; |
| 22 | const MAX_MCP_ENV: usize = 64; |
| 23 | const MAX_MCP_HEADERS: usize = 64; |
| 24 | const MAX_MCP_SCOPES: usize = 64; |
| 25 | const MAX_MCP_TOOL_FILTERS: usize = 256; |
| 26 | const MAX_MCP_TIMEOUT_SECS: u64 = 3_600; |
| 27 | |
| 28 | #[derive(Debug, Clone, Deserialize, Serialize)] |
| 29 | #[serde(deny_unknown_fields)] |
| 30 | pub struct PluginManifest { |
| 31 | /// Missing means the legacy, pre-versioned Codewhale manifest. Legacy |
| 32 | /// manifests remain readable, but `/plugin validate` reports the migration. |
| 33 | #[serde(default)] |
| 34 | pub schema_version: u32, |
| 35 | pub plugin: PluginMeta, |
| 36 | #[serde(default)] |
| 37 | pub skills: Option<PluginPathSpec>, |
| 38 | #[serde(default)] |
| 39 | pub commands: Option<PluginPathSpec>, |
| 40 | #[serde(default, alias = "profiles")] |
| 41 | pub agents: Option<PluginPathSpec>, |
| 42 | #[serde(default)] |
| 43 | pub hooks: Option<PluginPathSpec>, |
| 44 | #[serde(default, alias = "lsp_servers")] |
| 45 | pub lsp: Option<PluginPathSpec>, |
| 46 | #[serde(default, alias = "native_extension")] |
| 47 | pub native: Option<PluginPathSpec>, |
| 48 | #[serde(default)] |
| 49 | pub mcp_servers: Option<HashMap<String, McpServerConfig>>, |
| 50 | #[serde(default)] |
| 51 | pub capabilities: PluginCapabilities, |
| 52 | #[serde(default)] |
| 53 | pub when: Option<PluginWhen>, |
| 54 | } |
| 55 | |
| 56 | #[derive(Debug, Clone, Deserialize, Serialize)] |
| 57 | #[serde(deny_unknown_fields)] |
| 58 | pub struct PluginMeta { |
| 59 | pub name: String, |
| 60 | #[serde(default)] |
| 61 | pub description: Option<String>, |
| 62 | #[serde(default)] |
| 63 | pub version: String, |
| 64 | #[serde(default)] |
| 65 | pub author: Option<String>, |
| 66 | } |
| 67 | |
| 68 | /// A declarative component location. `path` preserves the original manifest |
| 69 | /// shape; `paths` lets a bundle split one component kind across directories. |
| 70 | #[derive(Debug, Clone, Default, Deserialize, Serialize)] |
| 71 | #[serde(deny_unknown_fields)] |
| 72 | pub struct PluginPathSpec { |
| 73 | #[serde(default)] |
| 74 | pub path: Option<String>, |
| 75 | #[serde(default)] |
| 76 | pub paths: Vec<String>, |
| 77 | } |
| 78 | |
| 79 | impl PluginPathSpec { |
| 80 | fn declared_paths(&self, default: Option<&str>) -> Result<Vec<String>, String> { |
| 81 | let mut paths = Vec::new(); |
| 82 | if let Some(path) = self.path.as_deref() { |
| 83 | paths.push(path.to_string()); |
| 84 | } |
| 85 | paths.extend(self.paths.iter().cloned()); |
| 86 | if paths.is_empty() |
| 87 | && let Some(default) = default |
| 88 | { |
| 89 | paths.push(default.to_string()); |
| 90 | } |
| 91 | if paths.is_empty() { |
| 92 | return Err("component table must declare `path` or `paths`".to_string()); |
| 93 | } |
| 94 | if paths.len() > MAX_COMPONENT_PATHS { |
| 95 | return Err(format!( |
| 96 | "component declares {} paths; maximum is {MAX_COMPONENT_PATHS}", |
| 97 | paths.len() |
| 98 | )); |
| 99 | } |
| 100 | let mut seen = BTreeSet::new(); |
| 101 | for path in &paths { |
| 102 | if !seen.insert(path.clone()) { |
| 103 | return Err(format!( |
| 104 | "component path `{path}` is declared more than once" |
| 105 | )); |
| 106 | } |
| 107 | } |
| 108 | Ok(paths) |
| 109 | } |
| 110 | } |
| 111 | |
| 112 | #[derive(Debug, Clone, Default, Deserialize, Serialize)] |
| 113 | #[serde(deny_unknown_fields)] |
| 114 | pub struct PluginCapabilities { |
| 115 | /// Requested filesystem roots are inventory-only in v0.9.1. Declaring any |
| 116 | /// keeps the bundle inactive until a later permission adapter exists. |
| 117 | #[serde(default)] |
| 118 | pub filesystem_roots: Vec<String>, |
| 119 | /// Requested hosts are inventory-only. MCP URL hosts are added to the |
| 120 | /// effective capability inventory automatically. |
| 121 | #[serde(default)] |
| 122 | pub network_hosts: Vec<String>, |
| 123 | /// Lifecycle mutation is inventoried but unsupported in v0.9.1. |
| 124 | #[serde(default)] |
| 125 | pub lifecycle_mutation: bool, |
| 126 | } |
| 127 | |
| 128 | #[derive(Debug, Clone, Deserialize, Serialize)] |
| 129 | #[serde(deny_unknown_fields)] |
| 130 | pub struct PluginWhen { |
| 131 | #[serde(default)] |
| 132 | pub os: Option<Vec<String>>, |
| 133 | #[serde(default)] |
| 134 | pub binaries: Option<Vec<String>>, |
| 135 | } |
| 136 | |
| 137 | #[derive(Debug, Clone, Default, PartialEq, Eq)] |
| 138 | pub struct ResolvedPluginComponents { |
| 139 | pub skills: Vec<PathBuf>, |
| 140 | pub commands: Vec<PathBuf>, |
| 141 | pub agents: Vec<PathBuf>, |
| 142 | pub hooks: Vec<PathBuf>, |
| 143 | pub lsp: Vec<PathBuf>, |
| 144 | pub native: Vec<PathBuf>, |
| 145 | } |
| 146 | |
| 147 | impl ResolvedPluginComponents { |
| 148 | pub fn all_paths(&self) -> impl Iterator<Item = &PathBuf> { |
| 149 | self.skills |
| 150 | .iter() |
| 151 | .chain(&self.commands) |
| 152 | .chain(&self.agents) |
| 153 | .chain(&self.hooks) |
| 154 | .chain(&self.lsp) |
| 155 | .chain(&self.native) |
| 156 | } |
| 157 | } |
| 158 | |
| 159 | #[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)] |
| 160 | #[serde(deny_unknown_fields)] |
| 161 | pub struct PluginInventory { |
| 162 | pub skills: usize, |
| 163 | pub mcp_servers: usize, |
| 164 | /// MCP servers that launch a child process under the Codewhale user's |
| 165 | /// host permissions. Kept separate from remote MCP so the review screen |
| 166 | /// cannot imply that an empty declared filesystem/network list is a |
| 167 | /// sandbox boundary. |
| 168 | #[serde(default)] |
| 169 | pub stdio_mcp_servers: usize, |
| 170 | /// MCP servers contacted over HTTP(S) without launching a local child. |
| 171 | #[serde(default)] |
| 172 | pub remote_mcp_servers: usize, |
| 173 | pub commands: usize, |
| 174 | pub agents: usize, |
| 175 | pub hooks: usize, |
| 176 | pub lsp: usize, |
| 177 | pub native: usize, |
| 178 | pub filesystem_roots: Vec<String>, |
| 179 | pub network_hosts: Vec<String>, |
| 180 | pub lifecycle_mutation: bool, |
| 181 | } |
| 182 | |
| 183 | impl PluginInventory { |
| 184 | #[must_use] |
| 185 | pub fn unsupported_labels(&self) -> Vec<&'static str> { |
| 186 | let mut labels = Vec::new(); |
| 187 | if self.commands > 0 { |
| 188 | labels.push("commands"); |
| 189 | } |
| 190 | if self.agents > 0 { |
| 191 | labels.push("agents"); |
| 192 | } |
| 193 | if self.hooks > 0 { |
| 194 | labels.push("hooks"); |
| 195 | } |
| 196 | if self.lsp > 0 { |
| 197 | labels.push("lsp"); |
| 198 | } |
| 199 | if self.native > 0 { |
| 200 | labels.push("native"); |
| 201 | } |
| 202 | if !self.filesystem_roots.is_empty() { |
| 203 | labels.push("filesystem-roots"); |
| 204 | } |
| 205 | if self.lifecycle_mutation { |
| 206 | labels.push("lifecycle-mutation"); |
| 207 | } |
| 208 | labels |
| 209 | } |
| 210 | |
| 211 | #[must_use] |
| 212 | pub fn has_unsupported_capabilities(&self) -> bool { |
| 213 | !self.unsupported_labels().is_empty() |
| 214 | } |
| 215 | |
| 216 | #[must_use] |
| 217 | pub fn summary(&self) -> String { |
| 218 | format!( |
| 219 | "skills={} mcp={} (stdio={} remote={}) commands={} agents={} hooks={} lsp={} native={}", |
| 220 | self.skills, |
| 221 | self.mcp_servers, |
| 222 | self.stdio_mcp_servers, |
| 223 | self.remote_mcp_servers, |
| 224 | self.commands, |
| 225 | self.agents, |
| 226 | self.hooks, |
| 227 | self.lsp, |
| 228 | self.native |
| 229 | ) |
| 230 | } |
| 231 | } |
| 232 | |
| 233 | #[derive(Debug, Clone)] |
| 234 | pub struct ValidatedManifest { |
| 235 | pub manifest: PluginManifest, |
| 236 | pub canonical_root: PathBuf, |
| 237 | pub components: ResolvedPluginComponents, |
| 238 | pub inventory: PluginInventory, |
| 239 | pub content_hash: String, |
| 240 | /// Digest of the exact bytes read for every regular bundle file, keyed by |
| 241 | /// its lossless relative OS path. Runtime adapters use this to bind parsed |
| 242 | /// representations to the same bytes that produced `content_hash`. |
| 243 | pub(crate) file_hashes: BTreeMap<PathBuf, String>, |
| 244 | pub capability_hash: String, |
| 245 | pub applicable: bool, |
| 246 | pub warnings: Vec<String>, |
| 247 | } |
| 248 | |
| 249 | impl PluginManifest { |
| 250 | pub fn from_path(path: &Path) -> Result<Self, String> { |
| 251 | let metadata = fs::symlink_metadata(path) |
| 252 | .map_err(|e| format!("failed to inspect plugin.toml: {e}"))?; |
| 253 | if metadata_is_link_or_reparse(&metadata) { |
| 254 | return Err("plugin.toml may not be a symbolic link".to_string()); |
| 255 | } |
| 256 | let bytes = read_manifest_bytes(path)?; |
| 257 | let content = std::str::from_utf8(&bytes) |
| 258 | .map_err(|_| "plugin.toml must be valid UTF-8".to_string())?; |
| 259 | validate_nested_mcp_schema(content)?; |
| 260 | toml::from_str(content).map_err(|error| safe_toml_parse_error(&error)) |
| 261 | } |
| 262 | |
| 263 | pub fn validate_from_path(path: &Path) -> Result<ValidatedManifest, String> { |
| 264 | let manifest_metadata = fs::symlink_metadata(path) |
| 265 | .map_err(|e| format!("failed to inspect plugin.toml: {e}"))?; |
| 266 | if metadata_is_link_or_reparse(&manifest_metadata) || !manifest_metadata.is_file() { |
| 267 | return Err("plugin.toml must be a regular file, not a symbolic link".to_string()); |
| 268 | } |
| 269 | let root = path |
| 270 | .parent() |
| 271 | .ok_or_else(|| "plugin.toml has no parent directory".to_string())?; |
| 272 | let root_metadata = fs::symlink_metadata(root) |
| 273 | .map_err(|e| format!("failed to inspect plugin root: {e}"))?; |
| 274 | if metadata_is_link_or_reparse(&root_metadata) || !root_metadata.is_dir() { |
| 275 | return Err("plugin root must be a directory, not a symbolic link".to_string()); |
| 276 | } |
| 277 | let canonical_root = root |
| 278 | .canonicalize() |
| 279 | .map_err(|e| format!("failed to canonicalize plugin root: {e}"))?; |
| 280 | if !canonical_root.is_dir() { |
| 281 | return Err("plugin root is not a directory".to_string()); |
| 282 | } |
| 283 | |
| 284 | let manifest_bytes = read_manifest_bytes(path)?; |
| 285 | let manifest_text = std::str::from_utf8(&manifest_bytes) |
| 286 | .map_err(|_| "plugin.toml must be valid UTF-8".to_string())?; |
| 287 | validate_nested_mcp_schema(manifest_text)?; |
| 288 | let mut manifest: Self = |
| 289 | toml::from_str(manifest_text).map_err(|error| safe_toml_parse_error(&error))?; |
| 290 | let warnings = if manifest.schema_version == 0 { |
| 291 | let mut warnings = vec![format!( |
| 292 | "legacy manifest: add `schema_version = {CURRENT_SCHEMA_VERSION}`" |
| 293 | )]; |
| 294 | if manifest.plugin.version.trim().is_empty() { |
| 295 | manifest.plugin.version = "0.0.0".to_string(); |
| 296 | warnings.push( |
| 297 | "legacy manifest: add a semantic `[plugin].version`; displaying `0.0.0`" |
| 298 | .to_string(), |
| 299 | ); |
| 300 | } |
| 301 | warnings |
| 302 | } else { |
| 303 | Vec::new() |
| 304 | }; |
| 305 | manifest.validate_metadata()?; |
| 306 | |
| 307 | let components = manifest.resolve_components(&canonical_root)?; |
| 308 | manifest.validate_mcp_servers(&canonical_root)?; |
| 309 | let inventory = manifest.inventory(&components)?; |
| 310 | let (content_hash, file_hashes) = hash_bundle(&canonical_root, &manifest_bytes)?; |
| 311 | let capability_hash = hash_inventory(&inventory); |
| 312 | let applicable = manifest.check_when(); |
| 313 | if read_manifest_bytes(path)? != manifest_bytes { |
| 314 | return Err( |
| 315 | "plugin.toml changed while it was being validated; retry discovery".to_string(), |
| 316 | ); |
| 317 | } |
| 318 | |
| 319 | Ok(ValidatedManifest { |
| 320 | manifest, |
| 321 | canonical_root, |
| 322 | components, |
| 323 | inventory, |
| 324 | content_hash, |
| 325 | file_hashes, |
| 326 | capability_hash, |
| 327 | applicable, |
| 328 | warnings, |
| 329 | }) |
| 330 | } |
| 331 | |
| 332 | fn validate_metadata(&self) -> Result<(), String> { |
| 333 | if self.schema_version > CURRENT_SCHEMA_VERSION { |
| 334 | return Err(format!( |
| 335 | "unsupported schema_version {}; maximum is {CURRENT_SCHEMA_VERSION}", |
| 336 | self.schema_version |
| 337 | )); |
| 338 | } |
| 339 | validate_plugin_name(&self.plugin.name)?; |
| 340 | Version::parse(self.plugin.version.trim()).map_err(|e| { |
| 341 | format!( |
| 342 | "plugin version `{}` is not valid semantic versioning: {e}", |
| 343 | self.plugin.version |
| 344 | ) |
| 345 | })?; |
| 346 | validate_optional_text("description", self.plugin.description.as_deref(), 1_024)?; |
| 347 | validate_optional_text("author", self.plugin.author.as_deref(), 256)?; |
| 348 | validate_unique_texts("filesystem root", &self.capabilities.filesystem_roots, 512)?; |
| 349 | validate_unique_texts("network host", &self.capabilities.network_hosts, 253)?; |
| 350 | let declared_network_hosts = self |
| 351 | .capabilities |
| 352 | .network_hosts |
| 353 | .iter() |
| 354 | .map(|host| normalize_network_host(host)) |
| 355 | .collect::<Result<BTreeSet<_>, _>>()?; |
| 356 | let remote_network_hosts = self |
| 357 | .mcp_servers |
| 358 | .as_ref() |
| 359 | .into_iter() |
| 360 | .flat_map(|servers| servers.values()) |
| 361 | .filter_map(|server| server.url.as_deref()) |
| 362 | .map(|url| { |
| 363 | reqwest::Url::parse(url) |
| 364 | .map_err(|_| "remote MCP URL is invalid".to_string())? |
| 365 | .host_str() |
| 366 | .map(str::to_string) |
| 367 | .ok_or_else(|| "remote MCP URL is missing a host".to_string()) |
| 368 | }) |
| 369 | .collect::<Result<BTreeSet<_>, _>>()?; |
| 370 | if declared_network_hosts != remote_network_hosts { |
| 371 | return Err( |
| 372 | "capabilities.network_hosts must exactly match the normalized host set of all remote MCP endpoints" |
| 373 | .to_string(), |
| 374 | ); |
| 375 | } |
| 376 | if let Some(when) = &self.when { |
| 377 | if let Some(os_values) = &when.os { |
| 378 | validate_unique_texts("OS", os_values, 32)?; |
| 379 | const SUPPORTED: &[&str] = &[ |
| 380 | "windows", "linux", "macos", "freebsd", "openbsd", "netbsd", "android", "ios", |
| 381 | ]; |
| 382 | for os in os_values { |
| 383 | if !SUPPORTED.contains(&os.to_ascii_lowercase().as_str()) { |
| 384 | return Err(format!("unsupported OS selector `{os}`")); |
| 385 | } |
| 386 | } |
| 387 | } |
| 388 | if let Some(binaries) = &when.binaries { |
| 389 | validate_unique_texts("binary", binaries, 128)?; |
| 390 | for binary in binaries { |
| 391 | if binary.contains('/') |
| 392 | || binary.contains('\\') |
| 393 | || looks_windows_absolute(binary) |
| 394 | { |
| 395 | return Err(format!( |
| 396 | "binary condition `{binary}` must be a bare executable name" |
| 397 | )); |
| 398 | } |
| 399 | } |
| 400 | } |
| 401 | } |
| 402 | Ok(()) |
| 403 | } |
| 404 | |
| 405 | fn resolve_components(&self, root: &Path) -> Result<ResolvedPluginComponents, String> { |
| 406 | Ok(ResolvedPluginComponents { |
| 407 | skills: resolve_spec(root, "skills", self.skills.as_ref(), Some("skills"))?, |
| 408 | commands: resolve_spec(root, "commands", self.commands.as_ref(), None)?, |
| 409 | agents: resolve_spec(root, "agents", self.agents.as_ref(), None)?, |
| 410 | hooks: resolve_spec(root, "hooks", self.hooks.as_ref(), None)?, |
| 411 | lsp: resolve_spec(root, "lsp", self.lsp.as_ref(), None)?, |
| 412 | native: resolve_spec(root, "native", self.native.as_ref(), None)?, |
| 413 | }) |
| 414 | } |
| 415 | |
| 416 | fn validate_mcp_servers(&self, root: &Path) -> Result<(), String> { |
| 417 | let Some(servers) = &self.mcp_servers else { |
| 418 | return Ok(()); |
| 419 | }; |
| 420 | if servers.len() > MAX_COMPONENT_PATHS { |
| 421 | return Err(format!( |
| 422 | "manifest declares {} MCP servers; maximum is {MAX_COMPONENT_PATHS}", |
| 423 | servers.len() |
| 424 | )); |
| 425 | } |
| 426 | for (name, server) in servers { |
| 427 | validate_component_name("MCP server", name)?; |
| 428 | if server.args.len() > MAX_MCP_ARGS { |
| 429 | return Err(format!( |
| 430 | "MCP server `{name}` declares too many arguments; maximum is {MAX_MCP_ARGS}" |
| 431 | )); |
| 432 | } |
| 433 | if server.env.len() > MAX_MCP_ENV { |
| 434 | return Err(format!( |
| 435 | "MCP server `{name}` declares too many environment mappings; maximum is {MAX_MCP_ENV}" |
| 436 | )); |
| 437 | } |
| 438 | if server.env_headers.len() > MAX_MCP_HEADERS { |
| 439 | return Err(format!( |
| 440 | "MCP server `{name}` declares too many environment-backed headers; maximum is {MAX_MCP_HEADERS}" |
| 441 | )); |
| 442 | } |
| 443 | if server.scopes.len() > MAX_MCP_SCOPES { |
| 444 | return Err(format!( |
| 445 | "MCP server `{name}` declares too many OAuth scopes; maximum is {MAX_MCP_SCOPES}" |
| 446 | )); |
| 447 | } |
| 448 | if server.enabled_tools.len() > MAX_MCP_TOOL_FILTERS |
| 449 | || server.disabled_tools.len() > MAX_MCP_TOOL_FILTERS |
| 450 | { |
| 451 | return Err(format!( |
| 452 | "MCP server `{name}` declares too many tool filters; maximum is {MAX_MCP_TOOL_FILTERS} per list" |
| 453 | )); |
| 454 | } |
| 455 | for (label, timeout) in [ |
| 456 | ("connect_timeout", server.connect_timeout), |
| 457 | ("execute_timeout", server.execute_timeout), |
| 458 | ("read_timeout", server.read_timeout), |
| 459 | ] { |
| 460 | if timeout.is_some_and(|seconds| !(1..=MAX_MCP_TIMEOUT_SECS).contains(&seconds)) { |
| 461 | return Err(format!( |
| 462 | "MCP server `{name}` {label} must be 1-{MAX_MCP_TIMEOUT_SECS} seconds" |
| 463 | )); |
| 464 | } |
| 465 | } |
| 466 | if server.required && !server.enabled { |
| 467 | return Err(format!( |
| 468 | "MCP server `{name}` cannot be required while disabled" |
| 469 | )); |
| 470 | } |
| 471 | for arg in &server.args { |
| 472 | validate_text("MCP argument", arg, 4_096)?; |
| 473 | } |
| 474 | validate_unique_texts("enabled MCP tool", &server.enabled_tools, 256)?; |
| 475 | validate_unique_texts("disabled MCP tool", &server.disabled_tools, 256)?; |
| 476 | if server.enabled_tools.iter().any(|tool| { |
| 477 | server |
| 478 | .disabled_tools |
| 479 | .iter() |
| 480 | .any(|disabled| disabled == tool) |
| 481 | }) { |
| 482 | return Err(format!( |
| 483 | "MCP server `{name}` declares a tool in both enabled_tools and disabled_tools" |
| 484 | )); |
| 485 | } |
| 486 | match (server.command.as_deref(), server.url.as_deref()) { |
| 487 | (Some(command), None) => { |
| 488 | validate_text("MCP command", command, 512)?; |
| 489 | if server.transport.is_some() |
| 490 | || !server.headers.is_empty() |
| 491 | || !server.env_headers.is_empty() |
| 492 | || server.bearer_token_env_var.is_some() |
| 493 | || !server.scopes.is_empty() |
| 494 | || server.oauth.is_some() |
| 495 | || server.oauth_resource.is_some() |
| 496 | { |
| 497 | return Err(format!( |
| 498 | "stdio MCP server `{name}` may not declare remote transport or authentication fields" |
| 499 | )); |
| 500 | } |
| 501 | if command.contains('/') || command.contains('\\') { |
| 502 | let resolved = resolve_contained_path(root, command, "MCP command")?; |
| 503 | if !resolved.is_file() { |
| 504 | return Err(format!( |
| 505 | "MCP server `{name}` command is not a regular file" |
| 506 | )); |
| 507 | } |
| 508 | } else { |
| 509 | validate_bare_executable(command)?; |
| 510 | } |
| 511 | if let Some(cwd) = server.cwd.as_deref() { |
| 512 | let raw = cwd.to_string_lossy(); |
| 513 | let resolved = resolve_contained_path(root, &raw, "MCP cwd")?; |
| 514 | if !resolved.is_dir() { |
| 515 | return Err(format!( |
| 516 | "MCP server `{name}` cwd is not a directory: {}", |
| 517 | resolved.display() |
| 518 | )); |
| 519 | } |
| 520 | } |
| 521 | validate_mcp_argv_has_no_literal_credentials(name, &server.args)?; |
| 522 | for (index, arg) in server.args.iter().enumerate() { |
| 523 | if Path::new(arg).is_absolute() || looks_windows_absolute(arg) { |
| 524 | return Err(format!( |
| 525 | "MCP server `{name}` argument #{} must not use an absolute path", |
| 526 | index + 1 |
| 527 | )); |
| 528 | } |
| 529 | if is_relative_stdio_path_arg(arg) |
| 530 | && Path::new(arg) |
| 531 | .components() |
| 532 | .any(|part| matches!(part, Component::ParentDir)) |
| 533 | { |
| 534 | return Err(format!( |
| 535 | "MCP server `{name}` argument #{} escapes the plugin root", |
| 536 | index + 1 |
| 537 | )); |
| 538 | } |
| 539 | } |
| 540 | for (destination, source) in &server.env { |
| 541 | validate_environment_name("MCP environment destination", destination)?; |
| 542 | let source = exact_environment_placeholder(source).ok_or_else(|| { |
| 543 | format!( |
| 544 | "MCP server `{name}` environment values must be exact `${{SOURCE_ENV}}` references" |
| 545 | ) |
| 546 | })?; |
| 547 | validate_environment_name("MCP environment source", source)?; |
| 548 | } |
| 549 | } |
| 550 | (None, Some(url)) => { |
| 551 | if !server.scopes.is_empty() |
| 552 | || server.oauth.is_some() |
| 553 | || server.oauth_resource.is_some() |
| 554 | { |
| 555 | return Err(format!( |
| 556 | "remote MCP server `{name}` may not declare OAuth fields because plugin OAuth is disabled in v0.9.1; use env_headers or bearer_token_env_var" |
| 557 | )); |
| 558 | } |
| 559 | let parsed = reqwest::Url::parse(url) |
| 560 | .map_err(|e| format!("MCP server `{name}` URL is invalid: {e}"))?; |
| 561 | if !matches!(parsed.scheme(), "http" | "https") || parsed.host_str().is_none() { |
| 562 | return Err(format!( |
| 563 | "MCP server `{name}` URL must use http or https and include a host" |
| 564 | )); |
| 565 | } |
| 566 | if parsed.scheme() == "http" |
| 567 | && !parsed.host_str().is_some_and(|host| { |
| 568 | host.eq_ignore_ascii_case("localhost") |
| 569 | || host |
| 570 | .parse::<std::net::IpAddr>() |
| 571 | .is_ok_and(|address| address.is_loopback()) |
| 572 | }) |
| 573 | { |
| 574 | return Err(format!( |
| 575 | "MCP server `{name}` URL must use HTTPS unless it targets loopback" |
| 576 | )); |
| 577 | } |
| 578 | if !parsed.username().is_empty() || parsed.password().is_some() { |
| 579 | return Err(format!( |
| 580 | "MCP server `{name}` URL must not embed credentials; use environment-backed authentication" |
| 581 | )); |
| 582 | } |
| 583 | if parsed.query().is_some() || parsed.fragment().is_some() { |
| 584 | return Err(format!( |
| 585 | "MCP server `{name}` URL may not contain a query or fragment" |
| 586 | )); |
| 587 | } |
| 588 | if server.cwd.is_some() || !server.args.is_empty() || !server.env.is_empty() { |
| 589 | return Err(format!( |
| 590 | "remote MCP server `{name}` may not declare stdio cwd, args, or env" |
| 591 | )); |
| 592 | } |
| 593 | if !server.headers.is_empty() { |
| 594 | return Err(format!( |
| 595 | "remote MCP server `{name}` may not contain literal headers; use env_headers or bearer_token_env_var" |
| 596 | )); |
| 597 | } |
| 598 | if let Some(transport) = server.transport.as_deref() { |
| 599 | validate_text("MCP transport", transport, 32)?; |
| 600 | if !transport.eq_ignore_ascii_case("sse") { |
| 601 | return Err(format!( |
| 602 | "MCP server `{name}` transport must be `sse` when explicitly set" |
| 603 | )); |
| 604 | } |
| 605 | } |
| 606 | for (header, env_var) in &server.env_headers { |
| 607 | validate_http_header_name(header)?; |
| 608 | validate_environment_name("MCP header environment source", env_var)?; |
| 609 | } |
| 610 | if let Some(env_var) = server.bearer_token_env_var.as_deref() { |
| 611 | validate_environment_name("MCP bearer environment source", env_var)?; |
| 612 | } |
| 613 | validate_unique_texts("OAuth scope", &server.scopes, 256)?; |
| 614 | if let Some(oauth) = &server.oauth |
| 615 | && let Some(client_id) = oauth.client_id.as_deref() |
| 616 | { |
| 617 | validate_text("OAuth client id", client_id, 512)?; |
| 618 | } |
| 619 | if let Some(resource) = server.oauth_resource.as_deref() { |
| 620 | validate_safe_oauth_resource(resource)?; |
| 621 | } |
| 622 | } |
| 623 | (Some(_), Some(_)) => { |
| 624 | return Err(format!( |
| 625 | "MCP server `{name}` must declare exactly one of command or url" |
| 626 | )); |
| 627 | } |
| 628 | (None, None) => { |
| 629 | return Err(format!( |
| 630 | "MCP server `{name}` must declare exactly one of command or url" |
| 631 | )); |
| 632 | } |
| 633 | } |
| 634 | } |
| 635 | Ok(()) |
| 636 | } |
| 637 | |
| 638 | fn inventory(&self, components: &ResolvedPluginComponents) -> Result<PluginInventory, String> { |
| 639 | let stdio_mcp_servers = self.mcp_servers.as_ref().map_or(0, |servers| { |
| 640 | servers |
| 641 | .values() |
| 642 | .filter(|server| server.command.is_some() && server.url.is_none()) |
| 643 | .count() |
| 644 | }); |
| 645 | let remote_mcp_servers = self.mcp_servers.as_ref().map_or(0, |servers| { |
| 646 | servers |
| 647 | .values() |
| 648 | .filter(|server| server.url.is_some() && server.command.is_none()) |
| 649 | .count() |
| 650 | }); |
| 651 | let mut network_hosts = self |
| 652 | .capabilities |
| 653 | .network_hosts |
| 654 | .iter() |
| 655 | .map(|host| host.to_ascii_lowercase()) |
| 656 | .collect::<Vec<_>>(); |
| 657 | if let Some(servers) = &self.mcp_servers { |
| 658 | for server in servers.values() { |
| 659 | if let Some(url) = server.url.as_deref() |
| 660 | && let Ok(url) = reqwest::Url::parse(url) |
| 661 | && let Some(host) = url.host_str() |
| 662 | { |
| 663 | network_hosts.push(host.to_ascii_lowercase()); |
| 664 | } |
| 665 | } |
| 666 | } |
| 667 | network_hosts.sort(); |
| 668 | network_hosts.dedup(); |
| 669 | |
| 670 | let mut filesystem_roots = self.capabilities.filesystem_roots.clone(); |
| 671 | filesystem_roots.sort(); |
| 672 | filesystem_roots.dedup(); |
| 673 | |
| 674 | Ok(PluginInventory { |
| 675 | skills: components.skills.len(), |
| 676 | mcp_servers: self.mcp_servers.as_ref().map_or(0, HashMap::len), |
| 677 | stdio_mcp_servers, |
| 678 | remote_mcp_servers, |
| 679 | commands: components.commands.len(), |
| 680 | agents: components.agents.len(), |
| 681 | hooks: components.hooks.len(), |
| 682 | lsp: components.lsp.len(), |
| 683 | native: components.native.len(), |
| 684 | filesystem_roots, |
| 685 | network_hosts, |
| 686 | lifecycle_mutation: self.capabilities.lifecycle_mutation, |
| 687 | }) |
| 688 | } |
| 689 | |
| 690 | #[must_use] |
| 691 | pub fn check_when(&self) -> bool { |
| 692 | let Some(when) = &self.when else { |
| 693 | return true; |
| 694 | }; |
| 695 | if let Some(os_list) = &when.os { |
| 696 | let os = std::env::consts::OS; |
| 697 | if !os_list |
| 698 | .iter() |
| 699 | .any(|candidate| candidate.eq_ignore_ascii_case(os)) |
| 700 | { |
| 701 | return false; |
| 702 | } |
| 703 | } |
| 704 | if let Some(binaries) = &when.binaries { |
| 705 | for binary in binaries { |
| 706 | if !Self::has_binary(binary) { |
| 707 | return false; |
| 708 | } |
| 709 | } |
| 710 | } |
| 711 | true |
| 712 | } |
| 713 | |
| 714 | fn has_binary(name: &str) -> bool { |
| 715 | let paths = std::env::var_os("PATH").unwrap_or_default(); |
| 716 | for path in std::env::split_paths(&paths) { |
| 717 | let candidate = path.join(name); |
| 718 | if candidate.is_file() { |
| 719 | return true; |
| 720 | } |
| 721 | #[cfg(windows)] |
| 722 | if candidate.with_extension("exe").is_file() { |
| 723 | return true; |
| 724 | } |
| 725 | } |
| 726 | false |
| 727 | } |
| 728 | } |
| 729 | |
| 730 | fn safe_toml_parse_error(error: &toml::de::Error) -> String { |
| 731 | // `Display` includes source excerpts and can echo a malformed literal |
| 732 | // secret. Byte location is enough to repair the file without copying |
| 733 | // manifest values into logs, diagnostics, or transcripts. |
| 734 | error.span().map_or_else( |
| 735 | || "failed to parse plugin.toml; check the v1 schema and field types".to_string(), |
| 736 | |span| { |
| 737 | format!( |
| 738 | "failed to parse plugin.toml near bytes {}..{}; check the v1 schema and field types", |
| 739 | span.start, span.end |
| 740 | ) |
| 741 | }, |
| 742 | ) |
| 743 | } |
| 744 | |
| 745 | fn read_manifest_bytes(path: &Path) -> Result<Vec<u8>, String> { |
| 746 | let file = open_bundle_file(path) |
| 747 | .map_err(|e| format!("failed to open plugin.toml without following links: {e}"))?; |
| 748 | let mut bytes = Vec::new(); |
| 749 | file.take(MAX_MANIFEST_BYTES + 1) |
| 750 | .read_to_end(&mut bytes) |
| 751 | .map_err(|e| format!("failed to read plugin.toml: {e}"))?; |
| 752 | if bytes.len() as u64 > MAX_MANIFEST_BYTES { |
| 753 | return Err(format!( |
| 754 | "plugin.toml exceeds the {MAX_MANIFEST_BYTES}-byte review limit" |
| 755 | )); |
| 756 | } |
| 757 | Ok(bytes) |
| 758 | } |
| 759 | |
| 760 | pub fn validate_plugin_name(name: &str) -> Result<(), String> { |
| 761 | let count = name.chars().count(); |
| 762 | let valid = count > 0 |
| 763 | && count <= MAX_PLUGIN_NAME_CHARS |
| 764 | && name |
| 765 | .chars() |
| 766 | .all(|ch| ch.is_ascii_lowercase() || ch.is_ascii_digit() || ch == '-') |
| 767 | && name |
| 768 | .chars() |
| 769 | .next() |
| 770 | .is_some_and(|ch| ch.is_ascii_lowercase() || ch.is_ascii_digit()) |
| 771 | && name |
| 772 | .chars() |
| 773 | .last() |
| 774 | .is_some_and(|ch| ch.is_ascii_lowercase() || ch.is_ascii_digit()); |
| 775 | if valid { |
| 776 | Ok(()) |
| 777 | } else { |
| 778 | Err(format!( |
| 779 | "plugin name `{name}` must be 1-{MAX_PLUGIN_NAME_CHARS} lowercase ASCII letters, digits, or internal hyphens" |
| 780 | )) |
| 781 | } |
| 782 | } |
| 783 | |
| 784 | fn validate_component_name(kind: &str, name: &str) -> Result<(), String> { |
| 785 | let count = name.chars().count(); |
| 786 | let valid = count > 0 |
| 787 | && count <= MAX_PLUGIN_NAME_CHARS |
| 788 | && name |
| 789 | .chars() |
| 790 | .all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '-' | '_')) |
| 791 | && !name.starts_with(['-', '_']) |
| 792 | && !name.ends_with(['-', '_']); |
| 793 | if valid { |
| 794 | Ok(()) |
| 795 | } else { |
| 796 | Err(format!("{kind} name `{name}` is invalid")) |
| 797 | } |
| 798 | } |
| 799 | |
| 800 | fn validate_optional_text( |
| 801 | field: &str, |
| 802 | value: Option<&str>, |
| 803 | max_chars: usize, |
| 804 | ) -> Result<(), String> { |
| 805 | if let Some(value) = value { |
| 806 | validate_text(field, value, max_chars)?; |
| 807 | } |
| 808 | Ok(()) |
| 809 | } |
| 810 | |
| 811 | fn validate_mcp_argv_has_no_literal_credentials( |
| 812 | server_name: &str, |
| 813 | arguments: &[String], |
| 814 | ) -> Result<(), String> { |
| 815 | for (index, argument) in arguments.iter().enumerate() { |
| 816 | let (key, assigned_value) = argument |
| 817 | .split_once('=') |
| 818 | .map_or((argument.as_str(), None), |(key, value)| (key, Some(value))); |
| 819 | if credential_argument_key(key) |
| 820 | && (assigned_value.is_some_and(|value| !value.is_empty()) |
| 821 | || (assigned_value.is_none() && arguments.get(index + 1).is_some())) |
| 822 | { |
| 823 | return Err(format!( |
| 824 | "MCP server `{server_name}` argument #{} embeds a credential-bearing value; pass credentials through a reviewed environment mapping instead", |
| 825 | index + 1 |
| 826 | )); |
| 827 | } |
| 828 | if looks_like_literal_credential(argument) { |
| 829 | return Err(format!( |
| 830 | "MCP server `{server_name}` argument #{} looks like a literal credential; pass credentials through a reviewed environment mapping instead", |
| 831 | index + 1 |
| 832 | )); |
| 833 | } |
| 834 | } |
| 835 | Ok(()) |
| 836 | } |
| 837 | |
| 838 | fn credential_argument_key(value: &str) -> bool { |
| 839 | let key = value |
| 840 | .trim_start_matches('-') |
| 841 | .replace('_', "-") |
| 842 | .to_ascii_lowercase(); |
| 843 | [ |
| 844 | "token", |
| 845 | "api-key", |
| 846 | "apikey", |
| 847 | "password", |
| 848 | "passwd", |
| 849 | "secret", |
| 850 | "client-secret", |
| 851 | "authorization", |
| 852 | "auth-token", |
| 853 | "access-key", |
| 854 | "private-key", |
| 855 | "credential", |
| 856 | "credentials", |
| 857 | ] |
| 858 | .iter() |
| 859 | .any(|sensitive| key == *sensitive || key.ends_with(&format!("-{sensitive}"))) |
| 860 | } |
| 861 | |
| 862 | fn looks_like_literal_credential(value: &str) -> bool { |
| 863 | let trimmed = value.trim(); |
| 864 | trimmed.starts_with("sk-") |
| 865 | || trimmed.starts_with("ghp_") |
| 866 | || trimmed.starts_with("github_pat_") |
| 867 | || trimmed.starts_with("xoxb-") |
| 868 | || trimmed.starts_with("xoxp-") |
| 869 | || (trimmed.starts_with("AKIA") && trimmed.len() >= 16) |
| 870 | } |
| 871 | |
| 872 | fn validate_text(field: &str, value: &str, max_chars: usize) -> Result<(), String> { |
| 873 | let trimmed = value.trim(); |
| 874 | if trimmed.is_empty() || trimmed.chars().count() > max_chars { |
| 875 | return Err(format!( |
| 876 | "{field} must contain 1-{max_chars} non-whitespace characters" |
| 877 | )); |
| 878 | } |
| 879 | if value.chars().any(char::is_control) { |
| 880 | return Err(format!("{field} may not contain control characters")); |
| 881 | } |
| 882 | if value.chars().any(is_bidi_control) { |
| 883 | return Err(format!( |
| 884 | "{field} may not contain bidirectional formatting characters" |
| 885 | )); |
| 886 | } |
| 887 | Ok(()) |
| 888 | } |
| 889 | |
| 890 | fn is_bidi_control(ch: char) -> bool { |
| 891 | matches!( |
| 892 | ch, |
| 893 | '\u{061c}' | '\u{200e}' | '\u{200f}' | '\u{202a}'..='\u{202e}' | '\u{2066}'..='\u{2069}' |
| 894 | ) |
| 895 | } |
| 896 | |
| 897 | fn validate_unique_texts(field: &str, values: &[String], max_chars: usize) -> Result<(), String> { |
| 898 | if values.len() > MAX_COMPONENT_PATHS { |
| 899 | return Err(format!( |
| 900 | "too many {field} values; maximum is {MAX_COMPONENT_PATHS}" |
| 901 | )); |
| 902 | } |
| 903 | let mut seen = BTreeSet::new(); |
| 904 | for value in values { |
| 905 | validate_text(field, value, max_chars)?; |
| 906 | let normalized = value.to_ascii_lowercase(); |
| 907 | if !seen.insert(normalized) { |
| 908 | return Err(format!("duplicate {field} value `{value}`")); |
| 909 | } |
| 910 | } |
| 911 | Ok(()) |
| 912 | } |
| 913 | |
| 914 | fn normalize_network_host(host: &str) -> Result<String, String> { |
| 915 | if host.contains("://") || host.contains('/') || host.contains('\\') { |
| 916 | return Err(format!( |
| 917 | "network host `{host}` must be a host name, not a URL or path" |
| 918 | )); |
| 919 | } |
| 920 | let parsed = reqwest::Url::parse(&format!("https://{host}")) |
| 921 | .map_err(|e| format!("network host `{host}` is invalid: {e}"))?; |
| 922 | if parsed.port().is_some() |
| 923 | || !parsed.username().is_empty() |
| 924 | || parsed.password().is_some() |
| 925 | || parsed.path() != "/" |
| 926 | || parsed.query().is_some() |
| 927 | || parsed.fragment().is_some() |
| 928 | { |
| 929 | return Err(format!( |
| 930 | "network host `{host}` must contain only a normalized host name" |
| 931 | )); |
| 932 | } |
| 933 | parsed |
| 934 | .host_str() |
| 935 | .map(|host| host.to_ascii_lowercase()) |
| 936 | .ok_or_else(|| format!("network host `{host}` is invalid")) |
| 937 | } |
| 938 | |
| 939 | fn validate_bare_executable(command: &str) -> Result<(), String> { |
| 940 | if command.len() <= 128 |
| 941 | && command |
| 942 | .chars() |
| 943 | .all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '-' | '_' | '.' | '+')) |
| 944 | && !matches!(command, "." | "..") |
| 945 | { |
| 946 | Ok(()) |
| 947 | } else { |
| 948 | Err("MCP command must be a bare executable name or a contained plugin path".to_string()) |
| 949 | } |
| 950 | } |
| 951 | |
| 952 | fn validate_environment_name(field: &str, value: &str) -> Result<(), String> { |
| 953 | validate_text(field, value, 128)?; |
| 954 | if value |
| 955 | .chars() |
| 956 | .next() |
| 957 | .is_some_and(|ch| ch.is_ascii_alphabetic() || ch == '_') |
| 958 | && value |
| 959 | .chars() |
| 960 | .all(|ch| ch.is_ascii_alphanumeric() || ch == '_') |
| 961 | { |
| 962 | Ok(()) |
| 963 | } else { |
| 964 | Err(format!( |
| 965 | "{field} must be an ASCII environment variable name" |
| 966 | )) |
| 967 | } |
| 968 | } |
| 969 | |
| 970 | fn exact_environment_placeholder(value: &str) -> Option<&str> { |
| 971 | value.strip_prefix("${")?.strip_suffix('}') |
| 972 | } |
| 973 | |
| 974 | fn validate_http_header_name(name: &str) -> Result<(), String> { |
| 975 | validate_text("MCP HTTP header name", name, 128)?; |
| 976 | reqwest::header::HeaderName::from_bytes(name.as_bytes()) |
| 977 | .map(|_| ()) |
| 978 | .map_err(|_| "MCP HTTP header name is invalid".to_string()) |
| 979 | } |
| 980 | |
| 981 | fn validate_safe_oauth_resource(resource: &str) -> Result<(), String> { |
| 982 | validate_text("OAuth resource", resource, 2_048)?; |
| 983 | let parsed = reqwest::Url::parse(resource) |
| 984 | .map_err(|_| "OAuth resource must be an absolute HTTPS URL".to_string())?; |
| 985 | if parsed.scheme() != "https" |
| 986 | || parsed.host_str().is_none() |
| 987 | || !parsed.username().is_empty() |
| 988 | || parsed.password().is_some() |
| 989 | || parsed.query().is_some() |
| 990 | || parsed.fragment().is_some() |
| 991 | { |
| 992 | return Err( |
| 993 | "OAuth resource must be an HTTPS URL without credentials, query, or fragment" |
| 994 | .to_string(), |
| 995 | ); |
| 996 | } |
| 997 | Ok(()) |
| 998 | } |
| 999 | |
| 1000 | fn validate_nested_mcp_schema(content: &str) -> Result<(), String> { |
| 1001 | const SERVER_FIELDS: &[&str] = &[ |
| 1002 | "command", |
| 1003 | "args", |
| 1004 | "env", |
| 1005 | "cwd", |
| 1006 | "url", |
| 1007 | "transport", |
| 1008 | "connect_timeout", |
| 1009 | "execute_timeout", |
| 1010 | "read_timeout", |
| 1011 | "enabled", |
| 1012 | "required", |
| 1013 | "enabled_tools", |
| 1014 | "disabled_tools", |
| 1015 | "headers", |
| 1016 | "env_headers", |
| 1017 | "env_http_headers", |
| 1018 | "bearer_token_env_var", |
| 1019 | "scopes", |
| 1020 | "oauth", |
| 1021 | "oauth_resource", |
| 1022 | ]; |
| 1023 | let value: toml::Value = |
| 1024 | toml::from_str(content).map_err(|error| safe_toml_parse_error(&error))?; |
| 1025 | let Some(servers) = value.get("mcp_servers") else { |
| 1026 | return Ok(()); |
| 1027 | }; |
| 1028 | let servers = servers |
| 1029 | .as_table() |
| 1030 | .ok_or_else(|| "mcp_servers must be a table".to_string())?; |
| 1031 | for server in servers.values() { |
| 1032 | let server = server |
| 1033 | .as_table() |
| 1034 | .ok_or_else(|| "each MCP server must be a table".to_string())?; |
| 1035 | if server |
| 1036 | .keys() |
| 1037 | .any(|field| !SERVER_FIELDS.contains(&field.as_str())) |
| 1038 | { |
| 1039 | return Err("plugin MCP server contains an unsupported field".to_string()); |
| 1040 | } |
| 1041 | if let Some(oauth) = server.get("oauth") { |
| 1042 | let oauth = oauth |
| 1043 | .as_table() |
| 1044 | .ok_or_else(|| "plugin MCP oauth must be a table".to_string())?; |
| 1045 | if oauth.keys().any(|field| field != "client_id") { |
| 1046 | return Err("plugin MCP oauth contains an unsupported field".to_string()); |
| 1047 | } |
| 1048 | } |
| 1049 | } |
| 1050 | Ok(()) |
| 1051 | } |
| 1052 | |
| 1053 | fn resolve_spec( |
| 1054 | root: &Path, |
| 1055 | kind: &str, |
| 1056 | spec: Option<&PluginPathSpec>, |
| 1057 | default: Option<&str>, |
| 1058 | ) -> Result<Vec<PathBuf>, String> { |
| 1059 | let Some(spec) = spec else { |
| 1060 | return Ok(Vec::new()); |
| 1061 | }; |
| 1062 | spec.declared_paths(default)? |
| 1063 | .iter() |
| 1064 | .map(|path| resolve_contained_path(root, path, kind)) |
| 1065 | .collect() |
| 1066 | } |
| 1067 | |
| 1068 | fn resolve_contained_path(root: &Path, raw: &str, kind: &str) -> Result<PathBuf, String> { |
| 1069 | validate_text(&format!("{kind} path"), raw, 1_024)?; |
| 1070 | if Path::new(raw).is_absolute() || looks_windows_absolute(raw) { |
| 1071 | return Err(format!("{kind} path must be relative: `{raw}`")); |
| 1072 | } |
| 1073 | if Path::new(raw).components().any(|component| { |
| 1074 | matches!( |
| 1075 | component, |
| 1076 | Component::ParentDir | Component::RootDir | Component::Prefix(_) |
| 1077 | ) |
| 1078 | }) { |
| 1079 | return Err(format!("{kind} path escapes the plugin root: `{raw}`")); |
| 1080 | } |
| 1081 | let joined = root.join(raw); |
| 1082 | reject_symlink_components(root, &joined, kind)?; |
| 1083 | let canonical = joined |
| 1084 | .canonicalize() |
| 1085 | .map_err(|e| format!("{kind} path `{raw}` cannot be resolved: {e}"))?; |
| 1086 | if !canonical.starts_with(root) { |
| 1087 | return Err(format!("{kind} path escapes the plugin root: `{raw}`")); |
| 1088 | } |
| 1089 | Ok(canonical) |
| 1090 | } |
| 1091 | |
| 1092 | fn reject_symlink_components(root: &Path, target: &Path, kind: &str) -> Result<(), String> { |
| 1093 | let relative = target |
| 1094 | .strip_prefix(root) |
| 1095 | .map_err(|_| format!("{kind} path is outside the plugin root"))?; |
| 1096 | let mut cursor = root.to_path_buf(); |
| 1097 | for component in relative.components() { |
| 1098 | cursor.push(component.as_os_str()); |
| 1099 | let metadata = fs::symlink_metadata(&cursor) |
| 1100 | .map_err(|e| format!("failed to inspect {kind} path {}: {e}", cursor.display()))?; |
| 1101 | if metadata_is_link_or_reparse(&metadata) { |
| 1102 | return Err(format!( |
| 1103 | "{kind} path may not traverse symbolic link {}", |
| 1104 | cursor.display() |
| 1105 | )); |
| 1106 | } |
| 1107 | } |
| 1108 | Ok(()) |
| 1109 | } |
| 1110 | |
| 1111 | fn looks_windows_absolute(raw: &str) -> bool { |
| 1112 | let bytes = raw.as_bytes(); |
| 1113 | bytes |
| 1114 | .first() |
| 1115 | .is_some_and(|byte| matches!(*byte, b'\\' | b'/')) |
| 1116 | || (bytes.len() >= 3 |
| 1117 | && bytes[0].is_ascii_alphabetic() |
| 1118 | && bytes[1] == b':' |
| 1119 | && matches!(bytes[2], b'\\' | b'/')) |
| 1120 | } |
| 1121 | |
| 1122 | fn hash_bundle( |
| 1123 | root: &Path, |
| 1124 | manifest_bytes: &[u8], |
| 1125 | ) -> Result<(String, BTreeMap<PathBuf, String>), String> { |
| 1126 | let mut hasher = Sha256::new(); |
| 1127 | // v2 length-frames every variable-length field. The v1 delimiter-only |
| 1128 | // stream was structurally ambiguous across file-record boundaries. |
| 1129 | // Changing the domain invalidates every ambiguous v1 receipt. |
| 1130 | hasher.update(b"codewhale-plugin-content-v2\0plugin.toml\0"); |
| 1131 | hasher.update((manifest_bytes.len() as u64).to_le_bytes()); |
| 1132 | hasher.update(manifest_bytes); |
| 1133 | let mut budget = HashBudget::default(); |
| 1134 | // Hash the complete bundle, not only declared component roots. Local MCP |
| 1135 | // entrypoints and companion assets are security-relevant even when they do |
| 1136 | // not have a separate component table. |
| 1137 | hash_path(root, root, &mut hasher, &mut budget)?; |
| 1138 | Ok((hex_digest(hasher.finalize()), budget.file_hashes)) |
| 1139 | } |
| 1140 | |
| 1141 | #[derive(Default)] |
| 1142 | struct HashBudget { |
| 1143 | files: usize, |
| 1144 | bytes: u64, |
| 1145 | file_hashes: BTreeMap<PathBuf, String>, |
| 1146 | } |
| 1147 | |
| 1148 | fn hash_path( |
| 1149 | root: &Path, |
| 1150 | path: &Path, |
| 1151 | hasher: &mut Sha256, |
| 1152 | budget: &mut HashBudget, |
| 1153 | ) -> Result<(), String> { |
| 1154 | let metadata = fs::symlink_metadata(path) |
| 1155 | .map_err(|e| format!("failed to inspect component {}: {e}", path.display()))?; |
| 1156 | if metadata_is_link_or_reparse(&metadata) { |
| 1157 | return Err(format!( |
| 1158 | "component trees may not contain symbolic link {}", |
| 1159 | path.display() |
| 1160 | )); |
| 1161 | } |
| 1162 | let relative = path |
| 1163 | .strip_prefix(root) |
| 1164 | .map_err(|_| format!("component {} is outside the plugin root", path.display()))?; |
| 1165 | hash_permissions(&metadata, hasher); |
| 1166 | if metadata.is_dir() { |
| 1167 | #[cfg(windows)] |
| 1168 | let directory_guard = open_bundle_directory(path) |
| 1169 | .map_err(|e| format!("failed to open component directory safely: {e}"))?; |
| 1170 | #[cfg(windows)] |
| 1171 | ensure_windows_path_still_opened(path, &directory_guard)?; |
| 1172 | hasher.update(b"D\0"); |
| 1173 | super::path_identity::hash_os_path(hasher, b"bundle-relative-directory", relative); |
| 1174 | let mut entries = fs::read_dir(path) |
| 1175 | .map_err(|e| format!("failed to read component directory {}: {e}", path.display()))? |
| 1176 | .collect::<Result<Vec<_>, _>>() |
| 1177 | .map_err(|e| format!("failed to read component directory {}: {e}", path.display()))?; |
| 1178 | entries.sort_by_key(fs::DirEntry::file_name); |
| 1179 | for entry in entries { |
| 1180 | hash_path(root, &entry.path(), hasher, budget)?; |
| 1181 | } |
| 1182 | hasher.update(b"E\0"); |
| 1183 | #[cfg(windows)] |
| 1184 | ensure_windows_path_still_opened(path, &directory_guard)?; |
| 1185 | } else if metadata.is_file() { |
| 1186 | budget.files += 1; |
| 1187 | if budget.files > MAX_HASHED_FILES { |
| 1188 | return Err(format!( |
| 1189 | "plugin bundle content exceeds the v0.9.1 review limit ({MAX_HASHED_FILES} files / {MAX_HASHED_BYTES} bytes)" |
| 1190 | )); |
| 1191 | } |
| 1192 | hasher.update(b"F\0"); |
| 1193 | super::path_identity::hash_os_path(hasher, b"bundle-relative-file", relative); |
| 1194 | let mut file = open_bundle_file(path) |
| 1195 | .map_err(|e| format!("failed to read component file {}: {e}", path.display()))?; |
| 1196 | #[cfg(windows)] |
| 1197 | ensure_windows_path_still_opened(path, &file)?; |
| 1198 | let expected_len = file |
| 1199 | .metadata() |
| 1200 | .map_err(|e| format!("failed to inspect component file {}: {e}", path.display()))? |
| 1201 | .len(); |
| 1202 | hasher.update(expected_len.to_le_bytes()); |
| 1203 | let mut file_hasher = Sha256::new(); |
| 1204 | file_hasher.update(b"codewhale-plugin-file-bytes-v1\0"); |
| 1205 | // Keep the read buffer off the stack. `hash_path` is recursive and the |
| 1206 | // fixed-size array inflated every directory frame, which could exhaust |
| 1207 | // a Tokio worker stack while revalidating a nested plugin bundle. |
| 1208 | let mut buffer = vec![0_u8; 64 * 1024]; |
| 1209 | let mut actual_len = 0_u64; |
| 1210 | loop { |
| 1211 | let read = file |
| 1212 | .read(&mut buffer) |
| 1213 | .map_err(|e| format!("failed to read component file {}: {e}", path.display()))?; |
| 1214 | if read == 0 { |
| 1215 | break; |
| 1216 | } |
| 1217 | actual_len = actual_len.saturating_add(read as u64); |
| 1218 | budget.bytes = budget.bytes.saturating_add(read as u64); |
| 1219 | if budget.bytes > MAX_HASHED_BYTES { |
| 1220 | return Err(format!( |
| 1221 | "plugin bundle content exceeds the v0.9.1 review limit ({MAX_HASHED_FILES} files / {MAX_HASHED_BYTES} bytes)" |
| 1222 | )); |
| 1223 | } |
| 1224 | hasher.update(&buffer[..read]); |
| 1225 | file_hasher.update(&buffer[..read]); |
| 1226 | } |
| 1227 | if actual_len != expected_len { |
| 1228 | return Err(format!( |
| 1229 | "component file {} changed length while being reviewed", |
| 1230 | path.display() |
| 1231 | )); |
| 1232 | } |
| 1233 | budget |
| 1234 | .file_hashes |
| 1235 | .insert(relative.to_path_buf(), hex_digest(file_hasher.finalize())); |
| 1236 | #[cfg(windows)] |
| 1237 | ensure_windows_path_still_opened(path, &file)?; |
| 1238 | } else { |
| 1239 | return Err(format!( |
| 1240 | "component {} is neither a regular file nor directory", |
| 1241 | path.display() |
| 1242 | )); |
| 1243 | } |
| 1244 | Ok(()) |
| 1245 | } |
| 1246 | |
| 1247 | #[cfg(unix)] |
| 1248 | fn hash_permissions(metadata: &fs::Metadata, hasher: &mut Sha256) { |
| 1249 | use std::os::unix::fs::PermissionsExt; |
| 1250 | |
| 1251 | // Runtime snapshots deliberately remove group/other access and write bits. |
| 1252 | // Bind identity only to whether a regular file is executable, so the |
| 1253 | // owner-only staged representation has the same reviewed content hash. |
| 1254 | hasher.update(b"unix-executable\0"); |
| 1255 | hasher.update([u8::from( |
| 1256 | metadata.is_file() && metadata.permissions().mode() & 0o111 != 0, |
| 1257 | )]); |
| 1258 | } |
| 1259 | |
| 1260 | #[cfg(not(unix))] |
| 1261 | fn hash_permissions(metadata: &fs::Metadata, hasher: &mut Sha256) { |
| 1262 | let _ = metadata; |
| 1263 | // Windows staging marks files read-only as a defense-in-depth hardening |
| 1264 | // step; that representation change is not plugin content identity. |
| 1265 | hasher.update(b"portable-mode\0"); |
| 1266 | } |
| 1267 | |
| 1268 | #[cfg(unix)] |
| 1269 | pub(crate) fn open_bundle_file(path: &Path) -> std::io::Result<fs::File> { |
| 1270 | use std::os::unix::fs::OpenOptionsExt; |
| 1271 | |
| 1272 | fs::OpenOptions::new() |
| 1273 | .read(true) |
| 1274 | .custom_flags(libc::O_NOFOLLOW) |
| 1275 | .open(path) |
| 1276 | } |
| 1277 | |
| 1278 | #[cfg(windows)] |
| 1279 | pub(crate) fn open_bundle_file(path: &Path) -> std::io::Result<fs::File> { |
| 1280 | use std::os::windows::fs::OpenOptionsExt as _; |
| 1281 | |
| 1282 | const FILE_ATTRIBUTE_REPARSE_POINT: u32 = 0x0000_0400; |
| 1283 | let file = fs::OpenOptions::new() |
| 1284 | .read(true) |
| 1285 | .share_mode(0x0000_0001) // deny concurrent writes and replacement |
| 1286 | .custom_flags(0x0020_0000) // FILE_FLAG_OPEN_REPARSE_POINT |
| 1287 | .open(path)?; |
| 1288 | let metadata = file.metadata()?; |
| 1289 | let identity = windows_file_identity(&file)?; |
| 1290 | if !metadata.is_file() |
| 1291 | || identity.attributes & FILE_ATTRIBUTE_REPARSE_POINT != 0 |
| 1292 | || identity.links != 1 |
| 1293 | { |
| 1294 | return Err(std::io::Error::new( |
| 1295 | std::io::ErrorKind::InvalidData, |
| 1296 | "plugin file is a reparse point, hard link, or non-regular file", |
| 1297 | )); |
| 1298 | } |
| 1299 | Ok(file) |
| 1300 | } |
| 1301 | |
| 1302 | #[cfg(all(not(unix), not(windows)))] |
| 1303 | pub(crate) fn open_bundle_file(path: &Path) -> std::io::Result<fs::File> { |
| 1304 | fs::File::open(path) |
| 1305 | } |
| 1306 | |
| 1307 | #[cfg(windows)] |
| 1308 | fn open_bundle_directory(path: &Path) -> std::io::Result<fs::File> { |
| 1309 | use std::os::windows::fs::OpenOptionsExt as _; |
| 1310 | |
| 1311 | const FILE_ATTRIBUTE_REPARSE_POINT: u32 = 0x0000_0400; |
| 1312 | let file = fs::OpenOptions::new() |
| 1313 | .read(true) |
| 1314 | .share_mode(0x0000_0001) |
| 1315 | .custom_flags(0x0220_0000) // BACKUP_SEMANTICS | OPEN_REPARSE_POINT |
| 1316 | .open(path)?; |
| 1317 | let metadata = file.metadata()?; |
| 1318 | let identity = windows_file_identity(&file)?; |
| 1319 | if !metadata.is_dir() || identity.attributes & FILE_ATTRIBUTE_REPARSE_POINT != 0 { |
| 1320 | return Err(std::io::Error::new( |
| 1321 | std::io::ErrorKind::InvalidData, |
| 1322 | "plugin directory is a reparse point or non-directory", |
| 1323 | )); |
| 1324 | } |
| 1325 | Ok(file) |
| 1326 | } |
| 1327 | |
| 1328 | /// Reopen a reviewed path only to compare its current handle identity with a |
| 1329 | /// retained authority handle. Desired access is zero and sharing is permissive |
| 1330 | /// because the retained handle remains responsible for denying writes and |
| 1331 | /// replacement throughout the comparison. |
| 1332 | #[cfg(windows)] |
| 1333 | pub(crate) fn open_bundle_identity_probe( |
| 1334 | path: &Path, |
| 1335 | expect_directory: bool, |
| 1336 | ) -> std::io::Result<fs::File> { |
| 1337 | use std::os::windows::fs::OpenOptionsExt as _; |
| 1338 | |
| 1339 | const FILE_ATTRIBUTE_REPARSE_POINT: u32 = 0x0000_0400; |
| 1340 | const FILE_FLAG_BACKUP_SEMANTICS: u32 = 0x0200_0000; |
| 1341 | const FILE_FLAG_OPEN_REPARSE_POINT: u32 = 0x0020_0000; |
| 1342 | const FILE_SHARE_READ_WRITE_DELETE: u32 = 0x0000_0007; |
| 1343 | let flags = FILE_FLAG_OPEN_REPARSE_POINT |
| 1344 | | if expect_directory { |
| 1345 | FILE_FLAG_BACKUP_SEMANTICS |
| 1346 | } else { |
| 1347 | 0 |
| 1348 | }; |
| 1349 | let file = fs::OpenOptions::new() |
| 1350 | .access_mode(0) |
| 1351 | .share_mode(FILE_SHARE_READ_WRITE_DELETE) |
| 1352 | .custom_flags(flags) |
| 1353 | .open(path)?; |
| 1354 | let metadata = file.metadata()?; |
| 1355 | let identity = windows_file_identity(&file)?; |
| 1356 | let expected_kind = if expect_directory { |
| 1357 | metadata.is_dir() |
| 1358 | } else { |
| 1359 | metadata.is_file() && identity.links == 1 |
| 1360 | }; |
| 1361 | if !expected_kind || identity.attributes & FILE_ATTRIBUTE_REPARSE_POINT != 0 { |
| 1362 | return Err(std::io::Error::new( |
| 1363 | std::io::ErrorKind::InvalidData, |
| 1364 | "plugin identity probe found a reparse point, hard link, or unexpected object", |
| 1365 | )); |
| 1366 | } |
| 1367 | Ok(file) |
| 1368 | } |
| 1369 | |
| 1370 | #[cfg(windows)] |
| 1371 | fn ensure_windows_path_still_opened(path: &Path, opened: &fs::File) -> Result<(), String> { |
| 1372 | let after = fs::symlink_metadata(path) |
| 1373 | .map_err(|e| format!("failed to re-inspect plugin path after handle open: {e}"))?; |
| 1374 | if metadata_is_link_or_reparse(&after) { |
| 1375 | return Err("plugin path changed into a reparse point during validation".to_string()); |
| 1376 | } |
| 1377 | let expect_directory = if after.is_dir() { |
| 1378 | true |
| 1379 | } else if after.is_file() { |
| 1380 | false |
| 1381 | } else { |
| 1382 | return Err("plugin path changed into an unsupported object during validation".to_string()); |
| 1383 | }; |
| 1384 | let current = open_bundle_identity_probe(path, expect_directory) |
| 1385 | .map_err(|e| format!("failed to reopen plugin path for identity validation: {e}"))?; |
| 1386 | let opened = windows_file_identity(opened) |
| 1387 | .map_err(|e| format!("failed to identify retained plugin handle: {e}"))?; |
| 1388 | let current = windows_file_identity(¤t) |
| 1389 | .map_err(|e| format!("failed to identify current plugin path: {e}"))?; |
| 1390 | if opened.volume != current.volume || opened.index != current.index { |
| 1391 | return Err("plugin path identity changed between handle open and validation".to_string()); |
| 1392 | } |
| 1393 | Ok(()) |
| 1394 | } |
| 1395 | |
| 1396 | fn hash_inventory(inventory: &PluginInventory) -> String { |
| 1397 | let mut normalized = BTreeMap::new(); |
| 1398 | normalized.insert("skills", inventory.skills.to_string()); |
| 1399 | normalized.insert("mcp", inventory.mcp_servers.to_string()); |
| 1400 | normalized.insert("mcp-stdio", inventory.stdio_mcp_servers.to_string()); |
| 1401 | normalized.insert("mcp-remote", inventory.remote_mcp_servers.to_string()); |
| 1402 | normalized.insert("commands", inventory.commands.to_string()); |
| 1403 | normalized.insert("agents", inventory.agents.to_string()); |
| 1404 | normalized.insert("hooks", inventory.hooks.to_string()); |
| 1405 | normalized.insert("lsp", inventory.lsp.to_string()); |
| 1406 | normalized.insert("native", inventory.native.to_string()); |
| 1407 | normalized.insert("filesystem", inventory.filesystem_roots.join("\n")); |
| 1408 | normalized.insert("network", inventory.network_hosts.join("\n")); |
| 1409 | normalized.insert("lifecycle", inventory.lifecycle_mutation.to_string()); |
| 1410 | let mut hasher = Sha256::new(); |
| 1411 | hasher.update(b"codewhale-plugin-capabilities-v1\0"); |
| 1412 | for (key, value) in normalized { |
| 1413 | hasher.update(key.as_bytes()); |
| 1414 | hasher.update(b"\0"); |
| 1415 | hasher.update(value.as_bytes()); |
| 1416 | hasher.update(b"\0"); |
| 1417 | } |
| 1418 | hex_digest(hasher.finalize()) |
| 1419 | } |
| 1420 | |
| 1421 | fn hex_digest(bytes: impl AsRef<[u8]>) -> String { |
| 1422 | let bytes = bytes.as_ref(); |
| 1423 | let mut output = String::with_capacity(bytes.len() * 2); |
| 1424 | for byte in bytes { |
| 1425 | use std::fmt::Write as _; |
| 1426 | let _ = write!(output, "{byte:02x}"); |
| 1427 | } |
| 1428 | output |
| 1429 | } |
| 1430 | |
| 1431 | #[cfg(test)] |
| 1432 | mod tests { |
| 1433 | use super::*; |
| 1434 | |
| 1435 | fn write_manifest(root: &Path, extra: &str) -> PathBuf { |
| 1436 | fs::create_dir_all(root.join("skills/example")).unwrap(); |
| 1437 | fs::write( |
| 1438 | root.join("skills/example/SKILL.md"), |
| 1439 | "---\nname: example\ndescription: example\n---\nbody\n", |
| 1440 | ) |
| 1441 | .unwrap(); |
| 1442 | let path = root.join("plugin.toml"); |
| 1443 | fs::write( |
| 1444 | &path, |
| 1445 | format!( |
| 1446 | "schema_version = 1\n[plugin]\nname = \"example-plugin\"\nversion = \"1.2.3\"\n[skills]\npath = \"skills\"\n{extra}" |
| 1447 | ), |
| 1448 | ) |
| 1449 | .unwrap(); |
| 1450 | path |
| 1451 | } |
| 1452 | |
| 1453 | #[test] |
| 1454 | fn validates_versioned_manifest_and_hashes_declared_content() { |
| 1455 | let tmp = tempfile::tempdir().unwrap(); |
| 1456 | let path = write_manifest(tmp.path(), ""); |
| 1457 | let first = PluginManifest::validate_from_path(&path).unwrap(); |
| 1458 | assert_eq!(first.inventory.skills, 1); |
| 1459 | assert!(first.warnings.is_empty()); |
| 1460 | |
| 1461 | fs::write( |
| 1462 | tmp.path().join("skills/example/SKILL.md"), |
| 1463 | "---\nname: example\ndescription: changed\n---\nbody\n", |
| 1464 | ) |
| 1465 | .unwrap(); |
| 1466 | let second = PluginManifest::validate_from_path(&path).unwrap(); |
| 1467 | assert_ne!(first.content_hash, second.content_hash); |
| 1468 | assert_eq!(first.capability_hash, second.capability_hash); |
| 1469 | } |
| 1470 | |
| 1471 | #[test] |
| 1472 | fn validates_deep_bundle_on_small_stack() { |
| 1473 | let tmp = tempfile::tempdir().unwrap(); |
| 1474 | let manifest = write_manifest(tmp.path(), ""); |
| 1475 | let mut nested = tmp.path().join("nested"); |
| 1476 | for level in 0..8 { |
| 1477 | nested = nested.join(format!("level-{level}")); |
| 1478 | } |
| 1479 | fs::create_dir_all(&nested).unwrap(); |
| 1480 | fs::write(nested.join("payload.txt"), "nested bundle payload").unwrap(); |
| 1481 | |
| 1482 | let worker = std::thread::Builder::new() |
| 1483 | .name("plugin-small-stack-validation".to_string()) |
| 1484 | .stack_size(512 * 1024) |
| 1485 | .spawn(move || PluginManifest::validate_from_path(&manifest)) |
| 1486 | .unwrap(); |
| 1487 | let validated = worker |
| 1488 | .join() |
| 1489 | .expect("nested plugin validation must not overflow a small stack") |
| 1490 | .expect("nested plugin bundle must validate"); |
| 1491 | assert_eq!(validated.inventory.skills, 1); |
| 1492 | } |
| 1493 | |
| 1494 | #[test] |
| 1495 | fn bundle_hash_is_deterministic_and_covers_undeclared_companion_files() { |
| 1496 | let left = tempfile::tempdir().unwrap(); |
| 1497 | let right = tempfile::tempdir().unwrap(); |
| 1498 | let left_manifest = write_manifest(left.path(), ""); |
| 1499 | let right_manifest = write_manifest(right.path(), ""); |
| 1500 | fs::write(left.path().join("z.txt"), "z").unwrap(); |
| 1501 | fs::write(left.path().join("a.txt"), "a").unwrap(); |
| 1502 | fs::write(right.path().join("a.txt"), "a").unwrap(); |
| 1503 | fs::write(right.path().join("z.txt"), "z").unwrap(); |
| 1504 | |
| 1505 | let left_hash = PluginManifest::validate_from_path(&left_manifest).unwrap(); |
| 1506 | let right_hash = PluginManifest::validate_from_path(&right_manifest).unwrap(); |
| 1507 | assert_eq!(left_hash.content_hash, right_hash.content_hash); |
| 1508 | assert_eq!(left_hash.capability_hash, right_hash.capability_hash); |
| 1509 | |
| 1510 | fs::write(right.path().join("z.txt"), "changed").unwrap(); |
| 1511 | let changed = PluginManifest::validate_from_path(&right_manifest).unwrap(); |
| 1512 | assert_ne!(right_hash.content_hash, changed.content_hash); |
| 1513 | assert_eq!(right_hash.capability_hash, changed.capability_hash); |
| 1514 | } |
| 1515 | |
| 1516 | #[cfg(unix)] |
| 1517 | #[test] |
| 1518 | fn bundle_hash_frames_adversarial_binary_records() { |
| 1519 | let left = tempfile::tempdir().unwrap(); |
| 1520 | let right = tempfile::tempdir().unwrap(); |
| 1521 | let manifest = b"schema_version = 1\n[plugin]\nname = \"framing\"\nversion = \"1.0.0\"\n"; |
| 1522 | for root in [left.path(), right.path()] { |
| 1523 | fs::write(root.join("plugin.toml"), manifest).unwrap(); |
| 1524 | } |
| 1525 | |
| 1526 | fs::write(left.path().join("a.bin"), b"alpha").unwrap(); |
| 1527 | fs::write(left.path().join("b.bin"), b"omega").unwrap(); |
| 1528 | |
| 1529 | let mut adversarial = b"alpha\0unix-executable\0\0F\0codewhale-os-path-v1\0".to_vec(); |
| 1530 | adversarial.extend_from_slice(&(b"bundle-relative-file".len() as u64).to_le_bytes()); |
| 1531 | adversarial.extend_from_slice(b"bundle-relative-file"); |
| 1532 | adversarial.extend_from_slice(b"unix-bytes\0"); |
| 1533 | adversarial.extend_from_slice(&(b"b.bin".len() as u64).to_le_bytes()); |
| 1534 | adversarial.extend_from_slice(b"b.bin"); |
| 1535 | adversarial.extend_from_slice(b"omega"); |
| 1536 | fs::write(right.path().join("a.bin"), adversarial).unwrap(); |
| 1537 | |
| 1538 | let left = PluginManifest::validate_from_path(&left.path().join("plugin.toml")).unwrap(); |
| 1539 | let right = PluginManifest::validate_from_path(&right.path().join("plugin.toml")).unwrap(); |
| 1540 | assert_ne!(left.content_hash, right.content_hash); |
| 1541 | assert_eq!(left.capability_hash, right.capability_hash); |
| 1542 | } |
| 1543 | |
| 1544 | // Darwin rejects these malformed bytes at the filesystem boundary. The |
| 1545 | // platform-independent native-path framing is covered in path_identity; |
| 1546 | // run this full bundle-walk regression where Unix permits the entries. |
| 1547 | #[cfg(all(unix, not(target_os = "macos")))] |
| 1548 | #[test] |
| 1549 | fn bundle_hash_distinguishes_lossy_colliding_native_file_names() { |
| 1550 | use std::ffi::OsString; |
| 1551 | use std::os::unix::ffi::OsStringExt as _; |
| 1552 | |
| 1553 | let left = tempfile::tempdir().unwrap(); |
| 1554 | let right = tempfile::tempdir().unwrap(); |
| 1555 | let left_manifest = write_manifest(left.path(), ""); |
| 1556 | let right_manifest = write_manifest(right.path(), ""); |
| 1557 | let left_name = OsString::from_vec(vec![b'a', 0xff]); |
| 1558 | let right_name = OsString::from_vec(vec![b'a', 0xfe]); |
| 1559 | assert_eq!(left_name.to_string_lossy(), right_name.to_string_lossy()); |
| 1560 | fs::write(left.path().join(left_name), "same bytes").unwrap(); |
| 1561 | fs::write(right.path().join(right_name), "same bytes").unwrap(); |
| 1562 | |
| 1563 | let left = PluginManifest::validate_from_path(&left_manifest).unwrap(); |
| 1564 | let right = PluginManifest::validate_from_path(&right_manifest).unwrap(); |
| 1565 | assert_ne!(left.content_hash, right.content_hash); |
| 1566 | } |
| 1567 | |
| 1568 | #[test] |
| 1569 | fn legacy_manifest_is_accepted_with_migration_warning() { |
| 1570 | let tmp = tempfile::tempdir().unwrap(); |
| 1571 | fs::write( |
| 1572 | tmp.path().join("plugin.toml"), |
| 1573 | "[plugin]\nname = \"legacy\"\n", |
| 1574 | ) |
| 1575 | .unwrap(); |
| 1576 | let validated = |
| 1577 | PluginManifest::validate_from_path(&tmp.path().join("plugin.toml")).unwrap(); |
| 1578 | assert_eq!(validated.manifest.schema_version, 0); |
| 1579 | assert_eq!(validated.manifest.plugin.version, "0.0.0"); |
| 1580 | assert_eq!(validated.warnings.len(), 2); |
| 1581 | } |
| 1582 | |
| 1583 | #[test] |
| 1584 | fn rejects_unknown_fields_invalid_names_and_versions() { |
| 1585 | let invalid = [ |
| 1586 | "schema_version = 1\nunknown = true\n[plugin]\nname = \"ok\"\nversion = \"1.0.0\"\n", |
| 1587 | "schema_version = 1\n[plugin]\nname = \"Bad_Name\"\nversion = \"1.0.0\"\n", |
| 1588 | "schema_version = 1\n[plugin]\nname = \"ok\"\nversion = \"latest\"\n", |
| 1589 | "schema_version = 1\n[plugin]\nname = \"ok\"\n", |
| 1590 | ]; |
| 1591 | for source in invalid { |
| 1592 | let tmp = tempfile::tempdir().unwrap(); |
| 1593 | let path = tmp.path().join("plugin.toml"); |
| 1594 | fs::write(&path, source).unwrap(); |
| 1595 | assert!(PluginManifest::validate_from_path(&path).is_err()); |
| 1596 | } |
| 1597 | } |
| 1598 | |
| 1599 | #[test] |
| 1600 | fn parse_diagnostics_do_not_echo_manifest_values() { |
| 1601 | let tmp = tempfile::tempdir().unwrap(); |
| 1602 | let path = tmp.path().join("plugin.toml"); |
| 1603 | fs::write( |
| 1604 | &path, |
| 1605 | "schema_version = 1\n[plugin]\nname = \"safe\"\nversion = \"1.0.0\"\ndescription = [\"sk-sensitive-value\"]\n", |
| 1606 | ) |
| 1607 | .unwrap(); |
| 1608 | let error = PluginManifest::validate_from_path(&path).unwrap_err(); |
| 1609 | assert!(!error.contains("sk-sensitive-value")); |
| 1610 | } |
| 1611 | |
| 1612 | #[test] |
| 1613 | fn rejects_parent_absolute_and_windows_absolute_component_paths() { |
| 1614 | for bad in [ |
| 1615 | "../escape", |
| 1616 | "/tmp/escape", |
| 1617 | r"C:\\escape", |
| 1618 | r"\\\\server\\share", |
| 1619 | ] { |
| 1620 | let tmp = tempfile::tempdir().unwrap(); |
| 1621 | let path = write_manifest(tmp.path(), &format!("\n[commands]\npath = {bad:?}\n")); |
| 1622 | assert!( |
| 1623 | PluginManifest::validate_from_path(&path).is_err(), |
| 1624 | "accepted {bad}" |
| 1625 | ); |
| 1626 | } |
| 1627 | } |
| 1628 | |
| 1629 | #[cfg(unix)] |
| 1630 | #[test] |
| 1631 | fn rejects_symlinked_component_and_nested_symlink() { |
| 1632 | use std::os::unix::fs::symlink; |
| 1633 | |
| 1634 | let outside = tempfile::tempdir().unwrap(); |
| 1635 | fs::write(outside.path().join("SKILL.md"), "# outside").unwrap(); |
| 1636 | |
| 1637 | let tmp = tempfile::tempdir().unwrap(); |
| 1638 | let path = write_manifest(tmp.path(), ""); |
| 1639 | fs::remove_dir_all(tmp.path().join("skills")).unwrap(); |
| 1640 | symlink(outside.path(), tmp.path().join("skills")).unwrap(); |
| 1641 | assert!(PluginManifest::validate_from_path(&path).is_err()); |
| 1642 | |
| 1643 | fs::remove_file(tmp.path().join("skills")).unwrap(); |
| 1644 | fs::create_dir_all(tmp.path().join("skills/example")).unwrap(); |
| 1645 | fs::write(tmp.path().join("skills/example/SKILL.md"), "# safe").unwrap(); |
| 1646 | symlink( |
| 1647 | outside.path().join("SKILL.md"), |
| 1648 | tmp.path().join("skills/example/linked.md"), |
| 1649 | ) |
| 1650 | .unwrap(); |
| 1651 | assert!(PluginManifest::validate_from_path(&path).is_err()); |
| 1652 | } |
| 1653 | |
| 1654 | #[cfg(unix)] |
| 1655 | #[test] |
| 1656 | fn rejects_symlinked_manifest() { |
| 1657 | use std::os::unix::fs::symlink; |
| 1658 | |
| 1659 | let tmp = tempfile::tempdir().unwrap(); |
| 1660 | let real = tmp.path().join("real.toml"); |
| 1661 | fs::write( |
| 1662 | &real, |
| 1663 | "schema_version = 1\n[plugin]\nname = \"linked\"\nversion = \"1.0.0\"\n", |
| 1664 | ) |
| 1665 | .unwrap(); |
| 1666 | let linked = tmp.path().join("plugin.toml"); |
| 1667 | symlink(&real, &linked).unwrap(); |
| 1668 | |
| 1669 | assert!(PluginManifest::validate_from_path(&linked).is_err()); |
| 1670 | } |
| 1671 | |
| 1672 | #[cfg(unix)] |
| 1673 | #[test] |
| 1674 | fn rejects_symlinked_bundle_root() { |
| 1675 | use std::os::unix::fs::symlink; |
| 1676 | |
| 1677 | let tmp = tempfile::tempdir().unwrap(); |
| 1678 | let real_root = tmp.path().join("real"); |
| 1679 | fs::create_dir(&real_root).unwrap(); |
| 1680 | fs::write( |
| 1681 | real_root.join("plugin.toml"), |
| 1682 | "schema_version = 1\n[plugin]\nname = \"linked-root\"\nversion = \"1.0.0\"\n", |
| 1683 | ) |
| 1684 | .unwrap(); |
| 1685 | let linked_root = tmp.path().join("linked"); |
| 1686 | symlink(&real_root, &linked_root).unwrap(); |
| 1687 | |
| 1688 | assert!(PluginManifest::validate_from_path(&linked_root.join("plugin.toml")).is_err()); |
| 1689 | } |
| 1690 | |
| 1691 | #[test] |
| 1692 | fn rejects_absolute_mcp_arguments_and_embedded_url_credentials() { |
| 1693 | let absolute = tempfile::tempdir().unwrap(); |
| 1694 | let absolute_path = write_manifest( |
| 1695 | absolute.path(), |
| 1696 | "\n[mcp_servers.local]\ncommand = \"node\"\nargs = [\"/tmp/server.js\"]\n", |
| 1697 | ); |
| 1698 | assert!(PluginManifest::validate_from_path(&absolute_path).is_err()); |
| 1699 | |
| 1700 | let credentialed = tempfile::tempdir().unwrap(); |
| 1701 | let credentialed_path = write_manifest( |
| 1702 | credentialed.path(), |
| 1703 | "\n[mcp_servers.remote]\nurl = \"https://user:secret@example.invalid/mcp\"\n", |
| 1704 | ); |
| 1705 | assert!(PluginManifest::validate_from_path(&credentialed_path).is_err()); |
| 1706 | } |
| 1707 | |
| 1708 | #[test] |
| 1709 | fn windows_rooted_path_detection_is_host_independent() { |
| 1710 | assert!(looks_windows_absolute(r"C:\plugins\server.js")); |
| 1711 | assert!(looks_windows_absolute(r"C:/plugins/server.js")); |
| 1712 | assert!(looks_windows_absolute(r"\plugins\server.js")); |
| 1713 | assert!(looks_windows_absolute("/plugins/server.js")); |
| 1714 | assert!(!looks_windows_absolute("plugins/server.js")); |
| 1715 | assert!(!looks_windows_absolute("server.js")); |
| 1716 | } |
| 1717 | |
| 1718 | #[cfg(windows)] |
| 1719 | #[test] |
| 1720 | fn retained_bundle_handle_allows_identity_revalidation_but_denies_writes() { |
| 1721 | use std::os::windows::fs::OpenOptionsExt as _; |
| 1722 | |
| 1723 | let directory = tempfile::tempdir().unwrap(); |
| 1724 | let path = directory.path().join("reviewed.bin"); |
| 1725 | fs::write(&path, b"reviewed").unwrap(); |
| 1726 | let retained = open_bundle_file(&path).unwrap(); |
| 1727 | |
| 1728 | ensure_windows_path_still_opened(&path, &retained).unwrap(); |
| 1729 | let denied = fs::OpenOptions::new() |
| 1730 | .write(true) |
| 1731 | .share_mode(0x0000_0007) |
| 1732 | .open(&path); |
| 1733 | assert!(denied.is_err(), "retained authority must deny mutation"); |
| 1734 | |
| 1735 | drop(retained); |
| 1736 | fs::OpenOptions::new().write(true).open(&path).unwrap(); |
| 1737 | } |
| 1738 | |
| 1739 | #[test] |
| 1740 | fn unsupported_capabilities_are_inventoried() { |
| 1741 | let tmp = tempfile::tempdir().unwrap(); |
| 1742 | fs::create_dir_all(tmp.path().join("hooks")).unwrap(); |
| 1743 | let path = write_manifest( |
| 1744 | tmp.path(), |
| 1745 | "\n[hooks]\npath = \"hooks\"\n[capabilities]\nfilesystem_roots = [\"workspace\"]\nlifecycle_mutation = true\n", |
| 1746 | ); |
| 1747 | let validated = PluginManifest::validate_from_path(&path).unwrap(); |
| 1748 | assert!(validated.inventory.has_unsupported_capabilities()); |
| 1749 | assert!(validated.inventory.unsupported_labels().contains(&"hooks")); |
| 1750 | assert!( |
| 1751 | validated |
| 1752 | .inventory |
| 1753 | .unsupported_labels() |
| 1754 | .contains(&"filesystem-roots") |
| 1755 | ); |
| 1756 | } |
| 1757 | |
| 1758 | #[test] |
| 1759 | fn plugin_mcp_schema_and_transport_combinations_fail_closed() { |
| 1760 | let invalid = [ |
| 1761 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp\"\nunknown_nested = true\n[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n", |
| 1762 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp\"\nargs = [\"secret\"]\n[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n", |
| 1763 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp?token=secret\"\n[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n", |
| 1764 | "\n[mcp_servers.remote]\nurl = \"http://example.invalid/mcp\"\n[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n", |
| 1765 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp\"\n[capabilities]\nnetwork_hosts = [\"other.invalid\"]\n", |
| 1766 | "\n[mcp_servers.local]\ncommand = \"node\"\ntransport = \"sse\"\n", |
| 1767 | "\n[mcp_servers.local]\ncommand = \"node\"\nconnect_timeout = 0\n", |
| 1768 | "\n[mcp_servers.local]\ncommand = \"node\"\nenabled_tools = [\"same\"]\ndisabled_tools = [\"same\"]\n", |
| 1769 | "\n[mcp_servers.local]\ncommand = \"node\"\n[mcp_servers.local.env]\nTOKEN = \"literal-secret\"\n", |
| 1770 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp\"\n[mcp_servers.remote.headers]\nAuthorization = \"literal-secret\"\n[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n", |
| 1771 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp\"\n[mcp_servers.remote.oauth]\nclient_id = \"public\"\nsecret = \"must-not-parse\"\n[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n", |
| 1772 | ]; |
| 1773 | for extra in invalid { |
| 1774 | let tmp = tempfile::tempdir().unwrap(); |
| 1775 | let path = write_manifest(tmp.path(), extra); |
| 1776 | assert!( |
| 1777 | PluginManifest::validate_from_path(&path).is_err(), |
| 1778 | "accepted invalid plugin MCP manifest: {extra}" |
| 1779 | ); |
| 1780 | } |
| 1781 | } |
| 1782 | |
| 1783 | #[test] |
| 1784 | fn plugin_mcp_remote_allowlist_and_env_provenance_are_exact() { |
| 1785 | let tmp = tempfile::tempdir().unwrap(); |
| 1786 | let path = write_manifest( |
| 1787 | tmp.path(), |
| 1788 | r#" |
| 1789 | [mcp_servers.remote] |
| 1790 | url = "https://Example.Invalid:8443/mcp/v1" |
| 1791 | transport = "sse" |
| 1792 | connect_timeout = 30 |
| 1793 | execute_timeout = 120 |
| 1794 | read_timeout = 180 |
| 1795 | required = true |
| 1796 | enabled_tools = ["read"] |
| 1797 | disabled_tools = ["write"] |
| 1798 | bearer_token_env_var = "PLUGIN_BEARER" |
| 1799 | |
| 1800 | [mcp_servers.remote.env_headers] |
| 1801 | X_Api_Key = "PLUGIN_API_KEY" |
| 1802 | |
| 1803 | [capabilities] |
| 1804 | network_hosts = ["example.invalid"] |
| 1805 | "#, |
| 1806 | ); |
| 1807 | let validated = PluginManifest::validate_from_path(&path).unwrap(); |
| 1808 | assert_eq!( |
| 1809 | validated.inventory.network_hosts, |
| 1810 | vec!["example.invalid".to_string()] |
| 1811 | ); |
| 1812 | assert_eq!(validated.inventory.remote_mcp_servers, 1); |
| 1813 | } |
| 1814 | |
| 1815 | #[test] |
| 1816 | fn plugin_mcp_oauth_fields_are_rejected_for_v091() { |
| 1817 | for oauth_fields in [ |
| 1818 | "scopes = [\"tools.read\"]\n", |
| 1819 | "oauth_resource = \"https://resource.invalid/mcp\"\n", |
| 1820 | "[mcp_servers.remote.oauth]\nclient_id = \"public-client-id\"\n", |
| 1821 | ] { |
| 1822 | let tmp = tempfile::tempdir().unwrap(); |
| 1823 | let path = write_manifest( |
| 1824 | tmp.path(), |
| 1825 | &format!( |
| 1826 | "\n[mcp_servers.remote]\nurl = \"https://example.invalid/mcp\"\n{oauth_fields}[capabilities]\nnetwork_hosts = [\"example.invalid\"]\n" |
| 1827 | ), |
| 1828 | ); |
| 1829 | let error = PluginManifest::validate_from_path(&path) |
| 1830 | .expect_err("plugin OAuth authority must remain disabled in v0.9.1"); |
| 1831 | assert!(error.contains("plugin OAuth is disabled in v0.9.1")); |
| 1832 | } |
| 1833 | } |
| 1834 | |
| 1835 | #[test] |
| 1836 | fn reviewed_stdio_argv_rejects_literal_credentials_but_accepts_exact_safe_values() { |
| 1837 | for args in [ |
| 1838 | r#"["server.js", "--token", "literal-secret"]"#, |
| 1839 | r#"["server.js", "--api-key=literal-secret"]"#, |
| 1840 | r#"["server.js", "sk-live-literal"]"#, |
| 1841 | ] { |
| 1842 | let tmp = tempfile::tempdir().unwrap(); |
| 1843 | fs::write(tmp.path().join("server.js"), "// entrypoint\n").unwrap(); |
| 1844 | let path = write_manifest( |
| 1845 | tmp.path(), |
| 1846 | &format!("\n[mcp_servers.local]\ncommand = \"node\"\nargs = {args}\n"), |
| 1847 | ); |
| 1848 | let error = PluginManifest::validate_from_path(&path) |
| 1849 | .expect_err("credential-bearing argv must fail closed"); |
| 1850 | assert!(error.contains("credential"), "{error}"); |
| 1851 | } |
| 1852 | |
| 1853 | let safe = tempfile::tempdir().unwrap(); |
| 1854 | fs::write(safe.path().join("server.js"), "// entrypoint\n").unwrap(); |
| 1855 | let path = write_manifest( |
| 1856 | safe.path(), |
| 1857 | r#" |
| 1858 | [mcp_servers.local] |
| 1859 | command = "node" |
| 1860 | args = ["server.js", "--mode=worker", "-e", "console.log('ready')"] |
| 1861 | "#, |
| 1862 | ); |
| 1863 | PluginManifest::validate_from_path(&path) |
| 1864 | .expect("safe interpreter argv should remain reviewable exactly"); |
| 1865 | } |
| 1866 | |
| 1867 | #[test] |
| 1868 | fn manifest_text_rejects_controls_and_bidirectional_spoofing() { |
| 1869 | for unsafe_text in ["line\nbreak", "safe\u{202e}lmot.nigulp"] { |
| 1870 | let tmp = tempfile::tempdir().unwrap(); |
| 1871 | let path = tmp.path().join("plugin.toml"); |
| 1872 | fs::write( |
| 1873 | &path, |
| 1874 | format!( |
| 1875 | "schema_version = 1\n[plugin]\nname = \"safe\"\nversion = \"1.0.0\"\nauthor = {unsafe_text:?}\n" |
| 1876 | ), |
| 1877 | ) |
| 1878 | .unwrap(); |
| 1879 | assert!(PluginManifest::validate_from_path(&path).is_err()); |
| 1880 | } |
| 1881 | } |
| 1882 | } |
| 1883 |