| 1 | //! Bounded, read-only skill audit inventory. |
| 2 | //! |
| 3 | //! Separates "what is on disk" from runtime [`super::SkillRegistry`] merging. |
| 4 | //! Never executes skill bodies, never contacts the network, and never writes. |
| 5 | |
| 6 | use std::collections::{HashMap, HashSet}; |
| 7 | use std::fs::{self, File}; |
| 8 | use std::io::{Read, Take}; |
| 9 | use std::path::{Path, PathBuf}; |
| 10 | use std::time::SystemTime; |
| 11 | |
| 12 | use serde::Deserialize; |
| 13 | |
| 14 | use super::install::{INSTALLED_FROM_MARKER, TRUSTED_MARKER}; |
| 15 | use super::package_digest::{self, PackageDigestError}; |
| 16 | use super::roots::{ |
| 17 | SkillRootAccess, SkillRootCatalog, SkillRootDescriptor, SkillRootId, SkillRootKind, |
| 18 | safe_display_path, |
| 19 | }; |
| 20 | use super::system::is_exact_bundled_skill; |
| 21 | use super::{SkillRegistry, normalize_skill_name_for_lookup}; |
| 22 | |
| 23 | /// Max bytes of `SKILL.md` the auditor will read into memory. |
| 24 | pub const AUDIT_MAX_SKILL_MD_BYTES: u64 = 512 * 1024; |
| 25 | /// Max total package bytes considered for digest / integrity. |
| 26 | #[allow(dead_code)] // re-exported bound for callers / docs |
| 27 | pub const AUDIT_MAX_PACKAGE_BYTES: u64 = package_digest::PACKAGE_DIGEST_MAX_BYTES; |
| 28 | /// Max regular files included in a package digest walk. |
| 29 | #[allow(dead_code)] |
| 30 | pub const AUDIT_MAX_FILES: usize = package_digest::PACKAGE_DIGEST_MAX_FILES; |
| 31 | /// Max directory depth under a skill package (and under a root when locating packages). |
| 32 | pub const AUDIT_MAX_DEPTH: usize = package_digest::PACKAGE_DIGEST_MAX_DEPTH; |
| 33 | |
| 34 | /// Which roots the auditor visits. |
| 35 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 36 | pub enum SkillAuditMode { |
| 37 | /// CodeWhale-owned project/global roots only. |
| 38 | OwnedOnly, |
| 39 | /// Owned + compatible roots (including `.codex/skills`). Does not change runtime. |
| 40 | Compatible, |
| 41 | } |
| 42 | |
| 43 | /// Stable identity for one on-disk skill copy. |
| 44 | #[derive(Debug, Clone, PartialEq, Eq, Hash)] |
| 45 | pub struct AuditedSkillId { |
| 46 | pub root_id: SkillRootId, |
| 47 | pub relative_dir: PathBuf, |
| 48 | pub canonical_name: String, |
| 49 | } |
| 50 | |
| 51 | #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)] |
| 52 | pub enum SkillSourceKind { |
| 53 | CodeWhaleManaged, |
| 54 | CodeWhaleManual, |
| 55 | CompatibleExternal, |
| 56 | BuiltIn, |
| 57 | ReviewedPluginSnapshot, |
| 58 | RegistryCache, |
| 59 | } |
| 60 | |
| 61 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 62 | pub enum DigestUnknownReason { |
| 63 | Unreadable, |
| 64 | SymlinkPresent, |
| 65 | EscapedRoot, |
| 66 | Cycle, |
| 67 | Oversized, |
| 68 | TooManyFiles, |
| 69 | TooDeep, |
| 70 | } |
| 71 | |
| 72 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 73 | pub enum DigestState { |
| 74 | Known(String), |
| 75 | Unknown(DigestUnknownReason), |
| 76 | } |
| 77 | |
| 78 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 79 | pub enum ParserState { |
| 80 | Valid, |
| 81 | Warning(Vec<String>), |
| 82 | Broken(String), |
| 83 | Oversized, |
| 84 | } |
| 85 | |
| 86 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 87 | pub enum PrecedenceState { |
| 88 | Active, |
| 89 | ShadowedBy(AuditedSkillId), |
| 90 | InactiveSource, |
| 91 | Unknown, |
| 92 | } |
| 93 | |
| 94 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 95 | pub enum IntegrityState { |
| 96 | Healthy, |
| 97 | LocalContentDrift, |
| 98 | BrokenManagedInstall, |
| 99 | LegacyMetadataUnknown, |
| 100 | Unknown, |
| 101 | } |
| 102 | |
| 103 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 104 | #[allow(dead_code)] // NotApplicable reserved for non-filesystem rows in later stages |
| 105 | pub enum TrustState { |
| 106 | TrustedForDigest(String), |
| 107 | TrustStale, |
| 108 | LegacyAdvisory, |
| 109 | Untrusted, |
| 110 | NotApplicable, |
| 111 | Unknown, |
| 112 | } |
| 113 | |
| 114 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 115 | #[allow(dead_code)] // Partial / NeedsSetup filled when #4407 readiness cache is wired |
| 116 | pub enum ReadinessState { |
| 117 | Ready, |
| 118 | Partial, |
| 119 | NeedsSetup, |
| 120 | Unknown, |
| 121 | } |
| 122 | |
| 123 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 124 | #[allow(dead_code)] // Unknown reserved for unclassified logical sources |
| 125 | pub enum ProvenanceState { |
| 126 | Managed { |
| 127 | spec: Option<String>, |
| 128 | safe_url: Option<String>, |
| 129 | schema_version: Option<u32>, |
| 130 | }, |
| 131 | Manual, |
| 132 | External, |
| 133 | BuiltIn, |
| 134 | Plugin, |
| 135 | Cache, |
| 136 | Unknown, |
| 137 | } |
| 138 | |
| 139 | #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)] |
| 140 | pub enum SkillActionKind { |
| 141 | Install, |
| 142 | Import, |
| 143 | Update, |
| 144 | Remove, |
| 145 | Trust, |
| 146 | } |
| 147 | |
| 148 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 149 | pub enum SkillAuditWarning { |
| 150 | Message(String), |
| 151 | } |
| 152 | |
| 153 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 154 | pub struct AuditedSkill { |
| 155 | pub id: AuditedSkillId, |
| 156 | pub name: String, |
| 157 | pub description: Option<String>, |
| 158 | pub root: SkillRootDescriptor, |
| 159 | pub safe_display_path: String, |
| 160 | pub source_kind: SkillSourceKind, |
| 161 | pub parser: ParserState, |
| 162 | pub digest: DigestState, |
| 163 | pub provenance: ProvenanceState, |
| 164 | pub trust: TrustState, |
| 165 | pub readiness: ReadinessState, |
| 166 | pub precedence: PrecedenceState, |
| 167 | pub integrity: IntegrityState, |
| 168 | pub available_actions: Vec<SkillActionKind>, |
| 169 | pub warnings: Vec<SkillAuditWarning>, |
| 170 | /// Same canonical name + same digest as another copy. |
| 171 | pub exact_duplicate_of: Option<AuditedSkillId>, |
| 172 | /// Same canonical name + different digest. |
| 173 | pub conflicts_with: Vec<AuditedSkillId>, |
| 174 | /// External copy with no owned same-name skill — import candidate. |
| 175 | pub import_candidate: bool, |
| 176 | /// Package path left the declared skill root (symlink escape, etc.). |
| 177 | pub path_unsafe: bool, |
| 178 | } |
| 179 | |
| 180 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 181 | pub struct SkillAuditSnapshot { |
| 182 | pub scan_mode: SkillAuditMode, |
| 183 | pub roots: Vec<SkillRootDescriptor>, |
| 184 | pub skills: Vec<AuditedSkill>, |
| 185 | pub generated_at: SystemTime, |
| 186 | } |
| 187 | |
| 188 | /// Optional readiness cache from Issue #4407. Missing → [`ReadinessState::Unknown`]. |
| 189 | pub trait SkillReadinessProvider { |
| 190 | fn readiness_for(&self, skill: &AuditedSkillId) -> Option<ReadinessState>; |
| 191 | } |
| 192 | |
| 193 | /// Scan skill roots into a full, unmerged inventory. |
| 194 | #[cfg(test)] |
| 195 | #[must_use] |
| 196 | pub fn scan( |
| 197 | workspace: &Path, |
| 198 | home: Option<&Path>, |
| 199 | mode: SkillAuditMode, |
| 200 | readiness: Option<&dyn SkillReadinessProvider>, |
| 201 | ) -> SkillAuditSnapshot { |
| 202 | scan_with_configured(workspace, home, None, mode, readiness) |
| 203 | } |
| 204 | |
| 205 | /// Scan skill roots with an optional configured `skills_dir`. |
| 206 | #[must_use] |
| 207 | pub fn scan_with_configured( |
| 208 | workspace: &Path, |
| 209 | home: Option<&Path>, |
| 210 | configured_skills_dir: Option<&Path>, |
| 211 | mode: SkillAuditMode, |
| 212 | readiness: Option<&dyn SkillReadinessProvider>, |
| 213 | ) -> SkillAuditSnapshot { |
| 214 | let catalog = SkillRootCatalog::build(workspace, home, configured_skills_dir); |
| 215 | let root_refs: Vec<SkillRootDescriptor> = match mode { |
| 216 | SkillAuditMode::OwnedOnly => catalog |
| 217 | .audit_owned_directories() |
| 218 | .into_iter() |
| 219 | .cloned() |
| 220 | .collect(), |
| 221 | SkillAuditMode::Compatible => catalog |
| 222 | .audit_compatible_directories() |
| 223 | .into_iter() |
| 224 | .cloned() |
| 225 | .collect(), |
| 226 | }; |
| 227 | |
| 228 | let mut skills = Vec::new(); |
| 229 | for root in &root_refs { |
| 230 | skills.extend(scan_root(root, workspace, home)); |
| 231 | } |
| 232 | |
| 233 | classify_cross_root(&mut skills); |
| 234 | for skill in &mut skills { |
| 235 | skill.readiness = readiness |
| 236 | .and_then(|p| p.readiness_for(&skill.id)) |
| 237 | .unwrap_or(ReadinessState::Unknown); |
| 238 | skill.available_actions = action_policy(skill); |
| 239 | } |
| 240 | |
| 241 | SkillAuditSnapshot { |
| 242 | scan_mode: mode, |
| 243 | roots: root_refs, |
| 244 | skills, |
| 245 | generated_at: SystemTime::now(), |
| 246 | } |
| 247 | } |
| 248 | |
| 249 | /// Expand an owned-only inventory to compatible roots without re-reading the |
| 250 | /// unchanged owned packages. |
| 251 | /// |
| 252 | /// The manager uses this for its interactive scan-mode toggle. Package audits |
| 253 | /// include bounded content hashing, so re-auditing every bundled owned skill |
| 254 | /// can make a simple keypress appear lost on a cold filesystem. Reusing rows by |
| 255 | /// root keeps the result ordered by catalog precedence while newly eligible |
| 256 | /// external roots are still read from disk. |
| 257 | #[must_use] |
| 258 | pub fn expand_owned_scan_to_compatible( |
| 259 | workspace: &Path, |
| 260 | home: Option<&Path>, |
| 261 | configured_skills_dir: Option<&Path>, |
| 262 | owned_skills: &[AuditedSkill], |
| 263 | readiness: Option<&dyn SkillReadinessProvider>, |
| 264 | ) -> SkillAuditSnapshot { |
| 265 | let catalog = SkillRootCatalog::build(workspace, home, configured_skills_dir); |
| 266 | let root_refs: Vec<SkillRootDescriptor> = catalog |
| 267 | .audit_compatible_directories() |
| 268 | .into_iter() |
| 269 | .cloned() |
| 270 | .collect(); |
| 271 | let reusable_root_ids: HashSet<SkillRootId> = owned_skills |
| 272 | .iter() |
| 273 | .map(|skill| skill.id.root_id.clone()) |
| 274 | .collect(); |
| 275 | |
| 276 | let mut skills = Vec::new(); |
| 277 | for root in &root_refs { |
| 278 | if reusable_root_ids.contains(&root.id) { |
| 279 | skills.extend( |
| 280 | owned_skills |
| 281 | .iter() |
| 282 | .filter(|skill| skill.id.root_id == root.id) |
| 283 | .cloned(), |
| 284 | ); |
| 285 | } else { |
| 286 | skills.extend(scan_root(root, workspace, home)); |
| 287 | } |
| 288 | } |
| 289 | |
| 290 | // The owned rows carried their previous cross-root result. Recompute it |
| 291 | // against the expanded inventory so precedence/conflict/import actions are |
| 292 | // exactly the same as a fresh compatible scan. |
| 293 | for skill in &mut skills { |
| 294 | skill.precedence = if skill.root.active_for_runtime { |
| 295 | PrecedenceState::Unknown |
| 296 | } else { |
| 297 | PrecedenceState::InactiveSource |
| 298 | }; |
| 299 | skill.exact_duplicate_of = None; |
| 300 | skill.conflicts_with.clear(); |
| 301 | skill.import_candidate = false; |
| 302 | skill.available_actions.clear(); |
| 303 | } |
| 304 | classify_cross_root(&mut skills); |
| 305 | for skill in &mut skills { |
| 306 | skill.readiness = readiness |
| 307 | .and_then(|provider| provider.readiness_for(&skill.id)) |
| 308 | .unwrap_or(ReadinessState::Unknown); |
| 309 | skill.available_actions = action_policy(skill); |
| 310 | } |
| 311 | |
| 312 | SkillAuditSnapshot { |
| 313 | scan_mode: SkillAuditMode::Compatible, |
| 314 | roots: root_refs, |
| 315 | skills, |
| 316 | generated_at: SystemTime::now(), |
| 317 | } |
| 318 | } |
| 319 | |
| 320 | /// Compute available mutations for one audited row (UI and controller share this). |
| 321 | #[must_use] |
| 322 | pub fn action_policy(skill: &AuditedSkill) -> Vec<SkillActionKind> { |
| 323 | if skill.path_unsafe { |
| 324 | return Vec::new(); |
| 325 | } |
| 326 | |
| 327 | match skill.source_kind { |
| 328 | SkillSourceKind::CodeWhaleManaged => { |
| 329 | let mut actions = Vec::new(); |
| 330 | if matches!(skill.parser, ParserState::Valid | ParserState::Warning(_)) |
| 331 | && !matches!(skill.integrity, IntegrityState::BrokenManagedInstall) |
| 332 | { |
| 333 | actions.push(SkillActionKind::Update); |
| 334 | } |
| 335 | actions.push(SkillActionKind::Remove); |
| 336 | if matches!( |
| 337 | skill.trust, |
| 338 | TrustState::Untrusted | TrustState::TrustStale | TrustState::LegacyAdvisory |
| 339 | ) && matches!(skill.digest, DigestState::Known(_)) |
| 340 | && matches!(skill.parser, ParserState::Valid | ParserState::Warning(_)) |
| 341 | { |
| 342 | actions.push(SkillActionKind::Trust); |
| 343 | } |
| 344 | actions |
| 345 | } |
| 346 | SkillSourceKind::CodeWhaleManual => Vec::new(), |
| 347 | SkillSourceKind::CompatibleExternal => { |
| 348 | // Import is offered for fresh candidates and for same-name owned |
| 349 | // peers (exact duplicate → AlreadyPresent, conflict → replace confirm). |
| 350 | // The mutation controller remains the authority on scope/conflict policy. |
| 351 | let importable = matches!(skill.parser, ParserState::Valid | ParserState::Warning(_)) |
| 352 | && matches!(skill.digest, DigestState::Known(_)) |
| 353 | && (skill.import_candidate |
| 354 | || skill.exact_duplicate_of.is_some() |
| 355 | || !skill.conflicts_with.is_empty()); |
| 356 | if importable { |
| 357 | vec![SkillActionKind::Import] |
| 358 | } else { |
| 359 | Vec::new() |
| 360 | } |
| 361 | } |
| 362 | SkillSourceKind::BuiltIn |
| 363 | | SkillSourceKind::ReviewedPluginSnapshot |
| 364 | | SkillSourceKind::RegistryCache => Vec::new(), |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | // ── per-root scan ──────────────────────────────────────────────────────────── |
| 369 | |
| 370 | fn scan_root( |
| 371 | root: &SkillRootDescriptor, |
| 372 | workspace: &Path, |
| 373 | home: Option<&Path>, |
| 374 | ) -> Vec<AuditedSkill> { |
| 375 | let mut out = Vec::new(); |
| 376 | let Ok(canonical_root) = fs::canonicalize(&root.path) else { |
| 377 | return out; |
| 378 | }; |
| 379 | let mut visited = HashSet::new(); |
| 380 | let mut packages = Vec::new(); |
| 381 | find_skill_packages(&root.path, &canonical_root, 0, &mut visited, &mut packages); |
| 382 | |
| 383 | for package_dir in packages { |
| 384 | out.push(audit_package( |
| 385 | root, |
| 386 | &package_dir, |
| 387 | &canonical_root, |
| 388 | workspace, |
| 389 | home, |
| 390 | )); |
| 391 | } |
| 392 | out |
| 393 | } |
| 394 | |
| 395 | fn find_skill_packages( |
| 396 | dir: &Path, |
| 397 | canonical_root: &Path, |
| 398 | depth: usize, |
| 399 | visited: &mut HashSet<PathBuf>, |
| 400 | out: &mut Vec<PathBuf>, |
| 401 | ) { |
| 402 | if depth > AUDIT_MAX_DEPTH { |
| 403 | return; |
| 404 | } |
| 405 | let Ok(meta) = fs::symlink_metadata(dir) else { |
| 406 | return; |
| 407 | }; |
| 408 | if meta.file_type().is_symlink() { |
| 409 | let Ok(canonical) = fs::canonicalize(dir) else { |
| 410 | return; |
| 411 | }; |
| 412 | if !canonical.starts_with(canonical_root) || !canonical.is_dir() { |
| 413 | return; |
| 414 | } |
| 415 | if !visited.insert(canonical) { |
| 416 | return; |
| 417 | } |
| 418 | } else if meta.is_dir() { |
| 419 | let Ok(canonical) = fs::canonicalize(dir) else { |
| 420 | return; |
| 421 | }; |
| 422 | if !visited.insert(canonical) { |
| 423 | return; |
| 424 | } |
| 425 | } else { |
| 426 | return; |
| 427 | } |
| 428 | |
| 429 | let skill_md = dir.join("SKILL.md"); |
| 430 | if skill_md.is_file() || fs::symlink_metadata(&skill_md).is_ok() { |
| 431 | out.push(dir.to_path_buf()); |
| 432 | return; // do not descend into a skill package |
| 433 | } |
| 434 | |
| 435 | let Ok(entries) = fs::read_dir(dir) else { |
| 436 | return; |
| 437 | }; |
| 438 | for entry in entries.flatten() { |
| 439 | let path = entry.path(); |
| 440 | if path |
| 441 | .file_name() |
| 442 | .and_then(|s| s.to_str()) |
| 443 | .is_some_and(|name| name.starts_with('.')) |
| 444 | { |
| 445 | continue; |
| 446 | } |
| 447 | let Ok(meta) = fs::symlink_metadata(&path) else { |
| 448 | continue; |
| 449 | }; |
| 450 | if meta.is_dir() || meta.file_type().is_symlink() { |
| 451 | find_skill_packages(&path, canonical_root, depth + 1, visited, out); |
| 452 | } |
| 453 | } |
| 454 | } |
| 455 | |
| 456 | fn audit_package( |
| 457 | root: &SkillRootDescriptor, |
| 458 | package_dir: &Path, |
| 459 | canonical_root: &Path, |
| 460 | workspace: &Path, |
| 461 | home: Option<&Path>, |
| 462 | ) -> AuditedSkill { |
| 463 | let relative_dir = package_dir |
| 464 | .strip_prefix(&root.path) |
| 465 | .map(Path::to_path_buf) |
| 466 | .unwrap_or_else(|_| { |
| 467 | package_dir |
| 468 | .file_name() |
| 469 | .map(PathBuf::from) |
| 470 | .unwrap_or_else(|| PathBuf::from(".")) |
| 471 | }); |
| 472 | |
| 473 | let mut warnings = Vec::new(); |
| 474 | let skill_md = package_dir.join("SKILL.md"); |
| 475 | let (parser, name, description, skill_md_content) = parse_skill_md_bounded(&skill_md); |
| 476 | |
| 477 | let package = analyze_package(package_dir, canonical_root); |
| 478 | let path_unsafe = package.path_unsafe; |
| 479 | if path_unsafe { |
| 480 | warnings.push(SkillAuditWarning::Message( |
| 481 | "package contains a symlink that escapes the skill root or cycles".into(), |
| 482 | )); |
| 483 | } |
| 484 | for w in package.warnings { |
| 485 | warnings.push(SkillAuditWarning::Message(w)); |
| 486 | } |
| 487 | |
| 488 | let canonical_name = name |
| 489 | .as_deref() |
| 490 | .map(normalize_skill_name_for_lookup) |
| 491 | .unwrap_or_else(|| { |
| 492 | normalize_skill_name_for_lookup( |
| 493 | &relative_dir |
| 494 | .file_name() |
| 495 | .map(|s| s.to_string_lossy().into_owned()) |
| 496 | .unwrap_or_else(|| "skill".into()), |
| 497 | ) |
| 498 | }); |
| 499 | |
| 500 | let marker = read_installed_from(package_dir); |
| 501 | let trust = read_trust_state(package_dir, &package.digest); |
| 502 | let (source_kind, provenance, integrity) = classify_source( |
| 503 | root, |
| 504 | &canonical_name, |
| 505 | skill_md_content.as_deref(), |
| 506 | &marker, |
| 507 | &package.digest, |
| 508 | path_unsafe, |
| 509 | ); |
| 510 | |
| 511 | let display = format!( |
| 512 | "{}/{}", |
| 513 | safe_display_path(&root.path, Some(workspace), home), |
| 514 | relative_dir.display() |
| 515 | ) |
| 516 | .replace('\\', "/"); |
| 517 | |
| 518 | AuditedSkill { |
| 519 | id: AuditedSkillId { |
| 520 | root_id: root.id.clone(), |
| 521 | relative_dir, |
| 522 | canonical_name: canonical_name.clone(), |
| 523 | }, |
| 524 | name: canonical_name, |
| 525 | description, |
| 526 | root: root.clone(), |
| 527 | safe_display_path: display, |
| 528 | source_kind, |
| 529 | parser, |
| 530 | digest: package.digest, |
| 531 | provenance, |
| 532 | trust, |
| 533 | readiness: ReadinessState::Unknown, |
| 534 | precedence: if root.active_for_runtime { |
| 535 | PrecedenceState::Unknown // filled in classify_cross_root |
| 536 | } else { |
| 537 | PrecedenceState::InactiveSource |
| 538 | }, |
| 539 | integrity, |
| 540 | available_actions: Vec::new(), |
| 541 | warnings, |
| 542 | exact_duplicate_of: None, |
| 543 | conflicts_with: Vec::new(), |
| 544 | import_candidate: false, |
| 545 | path_unsafe, |
| 546 | } |
| 547 | } |
| 548 | |
| 549 | fn parse_skill_md_bounded( |
| 550 | path: &Path, |
| 551 | ) -> (ParserState, Option<String>, Option<String>, Option<String>) { |
| 552 | let meta = match fs::symlink_metadata(path) { |
| 553 | Ok(m) => m, |
| 554 | Err(err) => { |
| 555 | return ( |
| 556 | ParserState::Broken(format!("cannot stat SKILL.md: {err}")), |
| 557 | None, |
| 558 | None, |
| 559 | None, |
| 560 | ); |
| 561 | } |
| 562 | }; |
| 563 | if meta.file_type().is_symlink() { |
| 564 | return ( |
| 565 | ParserState::Broken("SKILL.md is a symlink".into()), |
| 566 | None, |
| 567 | None, |
| 568 | None, |
| 569 | ); |
| 570 | } |
| 571 | if meta.len() > AUDIT_MAX_SKILL_MD_BYTES { |
| 572 | return (ParserState::Oversized, None, None, None); |
| 573 | } |
| 574 | |
| 575 | let file = match File::open(path) { |
| 576 | Ok(f) => f, |
| 577 | Err(err) => { |
| 578 | return ( |
| 579 | ParserState::Broken(format!("cannot open SKILL.md: {err}")), |
| 580 | None, |
| 581 | None, |
| 582 | None, |
| 583 | ); |
| 584 | } |
| 585 | }; |
| 586 | let mut limited: Take<File> = file.take(AUDIT_MAX_SKILL_MD_BYTES + 1); |
| 587 | let mut buf = Vec::new(); |
| 588 | if let Err(err) = limited.read_to_end(&mut buf) { |
| 589 | return ( |
| 590 | ParserState::Broken(format!("cannot read SKILL.md: {err}")), |
| 591 | None, |
| 592 | None, |
| 593 | None, |
| 594 | ); |
| 595 | } |
| 596 | if buf.len() as u64 > AUDIT_MAX_SKILL_MD_BYTES { |
| 597 | return (ParserState::Oversized, None, None, None); |
| 598 | } |
| 599 | let content = match String::from_utf8(buf) { |
| 600 | Ok(s) => s, |
| 601 | Err(_) => { |
| 602 | return ( |
| 603 | ParserState::Broken("SKILL.md is not valid UTF-8".into()), |
| 604 | None, |
| 605 | None, |
| 606 | None, |
| 607 | ); |
| 608 | } |
| 609 | }; |
| 610 | |
| 611 | match SkillRegistry::parse_skill(path, &content) { |
| 612 | Ok(skill) => { |
| 613 | let desc = if skill.description.is_empty() { |
| 614 | None |
| 615 | } else { |
| 616 | Some(truncate_desc(&skill.description)) |
| 617 | }; |
| 618 | let mut warnings = Vec::new(); |
| 619 | if skill.description.is_empty() { |
| 620 | warnings.push("missing description".into()); |
| 621 | } |
| 622 | let parser = if warnings.is_empty() { |
| 623 | ParserState::Valid |
| 624 | } else { |
| 625 | ParserState::Warning(warnings) |
| 626 | }; |
| 627 | (parser, Some(skill.name), desc, Some(content)) |
| 628 | } |
| 629 | Err(reason) => (ParserState::Broken(reason), None, None, Some(content)), |
| 630 | } |
| 631 | } |
| 632 | |
| 633 | fn truncate_desc(s: &str) -> String { |
| 634 | const MAX: usize = 280; |
| 635 | let count = s.chars().count(); |
| 636 | if count <= MAX { |
| 637 | s.to_string() |
| 638 | } else { |
| 639 | let truncated: String = s.chars().take(MAX.saturating_sub(1)).collect(); |
| 640 | format!("{truncated}…") |
| 641 | } |
| 642 | } |
| 643 | |
| 644 | struct PackageAnalysis { |
| 645 | digest: DigestState, |
| 646 | path_unsafe: bool, |
| 647 | warnings: Vec<String>, |
| 648 | } |
| 649 | |
| 650 | /// Compute the bounded package content digest used by audit and mutation. |
| 651 | #[allow(dead_code)] // public wrapper; mutation uses package_digest directly |
| 652 | pub fn compute_package_digest(package_dir: &Path) -> Result<String, DigestUnknownReason> { |
| 653 | package_digest::compute_package_digest(package_dir).map_err(digest_error_to_unknown) |
| 654 | } |
| 655 | |
| 656 | fn digest_error_to_unknown(err: PackageDigestError) -> DigestUnknownReason { |
| 657 | match err { |
| 658 | PackageDigestError::Unreadable => DigestUnknownReason::Unreadable, |
| 659 | PackageDigestError::SymlinkPresent => DigestUnknownReason::SymlinkPresent, |
| 660 | PackageDigestError::EscapedRoot => DigestUnknownReason::EscapedRoot, |
| 661 | PackageDigestError::Cycle => DigestUnknownReason::Cycle, |
| 662 | PackageDigestError::Oversized => DigestUnknownReason::Oversized, |
| 663 | PackageDigestError::TooManyFiles => DigestUnknownReason::TooManyFiles, |
| 664 | PackageDigestError::TooDeep => DigestUnknownReason::TooDeep, |
| 665 | } |
| 666 | } |
| 667 | |
| 668 | fn analyze_package(package_dir: &Path, _canonical_root: &Path) -> PackageAnalysis { |
| 669 | match package_digest::compute_package_digest(package_dir) { |
| 670 | Ok(digest) => PackageAnalysis { |
| 671 | digest: DigestState::Known(digest), |
| 672 | path_unsafe: false, |
| 673 | warnings: Vec::new(), |
| 674 | }, |
| 675 | Err(err) => { |
| 676 | let reason = digest_error_to_unknown(err.clone()); |
| 677 | let path_unsafe = matches!( |
| 678 | err, |
| 679 | PackageDigestError::SymlinkPresent |
| 680 | | PackageDigestError::EscapedRoot |
| 681 | | PackageDigestError::Cycle |
| 682 | ); |
| 683 | PackageAnalysis { |
| 684 | digest: DigestState::Unknown(reason), |
| 685 | path_unsafe, |
| 686 | warnings: vec![err.to_string()], |
| 687 | } |
| 688 | } |
| 689 | } |
| 690 | } |
| 691 | |
| 692 | // ── markers ────────────────────────────────────────────────────────────────── |
| 693 | |
| 694 | #[derive(Debug, Clone, Deserialize)] |
| 695 | #[allow(dead_code)] // optional v2 fields reserved for mutation receipts in #4651 stage 3 |
| 696 | struct InstalledFromFile { |
| 697 | #[serde(default)] |
| 698 | schema_version: Option<u32>, |
| 699 | #[serde(default)] |
| 700 | spec: Option<String>, |
| 701 | #[serde(default)] |
| 702 | url: Option<String>, |
| 703 | /// v1 field name. |
| 704 | #[serde(default)] |
| 705 | checksum: Option<String>, |
| 706 | #[serde(default)] |
| 707 | source_checksum: Option<String>, |
| 708 | #[serde(default)] |
| 709 | content_digest: Option<String>, |
| 710 | #[serde(default)] |
| 711 | installed_name: Option<String>, |
| 712 | } |
| 713 | |
| 714 | #[derive(Debug, Clone)] |
| 715 | enum MarkerParse { |
| 716 | Absent, |
| 717 | V1(InstalledFromFile), |
| 718 | V2(InstalledFromFile), |
| 719 | #[allow(dead_code)] // reason retained for future audit warning surfacing |
| 720 | Broken(String), |
| 721 | } |
| 722 | |
| 723 | fn read_installed_from(package_dir: &Path) -> MarkerParse { |
| 724 | let path = package_dir.join(INSTALLED_FROM_MARKER); |
| 725 | let meta = match fs::symlink_metadata(&path) { |
| 726 | Ok(m) => m, |
| 727 | Err(_) => return MarkerParse::Absent, |
| 728 | }; |
| 729 | if meta.file_type().is_symlink() { |
| 730 | return MarkerParse::Broken("symlink .installed-from".into()); |
| 731 | } |
| 732 | if !meta.is_file() { |
| 733 | return MarkerParse::Broken(".installed-from is not a regular file".into()); |
| 734 | } |
| 735 | let Ok(body) = fs::read_to_string(&path) else { |
| 736 | return MarkerParse::Broken("unreadable .installed-from".into()); |
| 737 | }; |
| 738 | let Ok(parsed) = serde_json::from_str::<InstalledFromFile>(&body) else { |
| 739 | return MarkerParse::Broken("malformed .installed-from".into()); |
| 740 | }; |
| 741 | match parsed.schema_version { |
| 742 | Some(v) if v >= 2 => MarkerParse::V2(parsed), |
| 743 | Some(_) | None => MarkerParse::V1(parsed), |
| 744 | } |
| 745 | } |
| 746 | |
| 747 | #[derive(Debug, Deserialize)] |
| 748 | struct TrustFileV2 { |
| 749 | #[serde(default)] |
| 750 | schema_version: Option<u32>, |
| 751 | #[serde(default)] |
| 752 | content_digest: Option<String>, |
| 753 | } |
| 754 | |
| 755 | fn read_trust_state(package_dir: &Path, digest: &DigestState) -> TrustState { |
| 756 | let path = package_dir.join(TRUSTED_MARKER); |
| 757 | let meta = match fs::symlink_metadata(&path) { |
| 758 | Ok(m) => m, |
| 759 | Err(_) => return TrustState::Untrusted, |
| 760 | }; |
| 761 | if meta.file_type().is_symlink() { |
| 762 | // Do not follow symlink trust markers. |
| 763 | return TrustState::Unknown; |
| 764 | } |
| 765 | if !meta.is_file() { |
| 766 | return TrustState::Unknown; |
| 767 | } |
| 768 | let Ok(body) = fs::read_to_string(&path) else { |
| 769 | return TrustState::Unknown; |
| 770 | }; |
| 771 | if let Ok(parsed) = serde_json::from_str::<TrustFileV2>(&body) |
| 772 | && parsed.schema_version == Some(2) |
| 773 | { |
| 774 | let Some(trusted_digest) = parsed.content_digest else { |
| 775 | return TrustState::Unknown; |
| 776 | }; |
| 777 | return match digest { |
| 778 | DigestState::Known(current) if current == &trusted_digest => { |
| 779 | TrustState::TrustedForDigest(trusted_digest) |
| 780 | } |
| 781 | DigestState::Known(_) => TrustState::TrustStale, |
| 782 | DigestState::Unknown(_) => TrustState::Unknown, |
| 783 | }; |
| 784 | } |
| 785 | TrustState::LegacyAdvisory |
| 786 | } |
| 787 | |
| 788 | fn sanitize_url_for_display(url: &str) -> String { |
| 789 | // Strip userinfo, query, and fragment before UI / receipts. |
| 790 | let without_fragment = url.split('#').next().unwrap_or(url); |
| 791 | let without_query = without_fragment |
| 792 | .split('?') |
| 793 | .next() |
| 794 | .unwrap_or(without_fragment); |
| 795 | if let Some(scheme_end) = without_query.find("://") { |
| 796 | let scheme = &without_query[..scheme_end]; |
| 797 | let rest = &without_query[scheme_end + 3..]; |
| 798 | if let Some(at) = rest.find('@') { |
| 799 | return format!("{scheme}://{}", &rest[at + 1..]); |
| 800 | } |
| 801 | } |
| 802 | without_query.to_string() |
| 803 | } |
| 804 | |
| 805 | fn classify_source( |
| 806 | root: &SkillRootDescriptor, |
| 807 | canonical_name: &str, |
| 808 | skill_md_content: Option<&str>, |
| 809 | marker: &MarkerParse, |
| 810 | digest: &DigestState, |
| 811 | _path_unsafe: bool, |
| 812 | ) -> (SkillSourceKind, ProvenanceState, IntegrityState) { |
| 813 | if matches!(root.kind, SkillRootKind::RegistryCache) { |
| 814 | return ( |
| 815 | SkillSourceKind::RegistryCache, |
| 816 | ProvenanceState::Cache, |
| 817 | IntegrityState::Unknown, |
| 818 | ); |
| 819 | } |
| 820 | if matches!(root.kind, SkillRootKind::ReviewedPluginSnapshot) { |
| 821 | return ( |
| 822 | SkillSourceKind::ReviewedPluginSnapshot, |
| 823 | ProvenanceState::Plugin, |
| 824 | IntegrityState::Unknown, |
| 825 | ); |
| 826 | } |
| 827 | |
| 828 | if root.access != SkillRootAccess::WritableOwned { |
| 829 | return ( |
| 830 | SkillSourceKind::CompatibleExternal, |
| 831 | ProvenanceState::External, |
| 832 | IntegrityState::Unknown, |
| 833 | ); |
| 834 | } |
| 835 | |
| 836 | // Managed markers win over bundled-name heuristics so a registry install |
| 837 | // that reuses a bundled command name (e.g. `pdf`) stays Update/Remove/Trust |
| 838 | // capable. Exact shipped body without a marker is still BuiltIn. |
| 839 | match marker { |
| 840 | MarkerParse::V1(_) | MarkerParse::V2(_) | MarkerParse::Broken(_) => {} |
| 841 | MarkerParse::Absent => { |
| 842 | if let Some(content) = skill_md_content |
| 843 | && is_exact_bundled_skill(canonical_name, content) |
| 844 | { |
| 845 | return ( |
| 846 | SkillSourceKind::BuiltIn, |
| 847 | ProvenanceState::BuiltIn, |
| 848 | IntegrityState::Healthy, |
| 849 | ); |
| 850 | } |
| 851 | } |
| 852 | } |
| 853 | |
| 854 | match marker { |
| 855 | MarkerParse::Absent => ( |
| 856 | SkillSourceKind::CodeWhaleManual, |
| 857 | ProvenanceState::Manual, |
| 858 | IntegrityState::Unknown, |
| 859 | ), |
| 860 | MarkerParse::Broken(_) => ( |
| 861 | SkillSourceKind::CodeWhaleManaged, |
| 862 | ProvenanceState::Managed { |
| 863 | spec: None, |
| 864 | safe_url: None, |
| 865 | schema_version: None, |
| 866 | }, |
| 867 | IntegrityState::BrokenManagedInstall, |
| 868 | ), |
| 869 | MarkerParse::V1(m) => ( |
| 870 | SkillSourceKind::CodeWhaleManaged, |
| 871 | ProvenanceState::Managed { |
| 872 | spec: m.spec.clone(), |
| 873 | safe_url: m.url.as_deref().map(sanitize_url_for_display), |
| 874 | schema_version: m.schema_version.or(Some(1)), |
| 875 | }, |
| 876 | IntegrityState::LegacyMetadataUnknown, |
| 877 | ), |
| 878 | MarkerParse::V2(m) => { |
| 879 | let integrity = match (&m.content_digest, digest) { |
| 880 | (Some(expected), DigestState::Known(actual)) if expected == actual => { |
| 881 | IntegrityState::Healthy |
| 882 | } |
| 883 | (Some(_), DigestState::Known(_)) => IntegrityState::LocalContentDrift, |
| 884 | (None, _) => IntegrityState::Unknown, |
| 885 | (_, DigestState::Unknown(_)) => IntegrityState::Unknown, |
| 886 | }; |
| 887 | ( |
| 888 | SkillSourceKind::CodeWhaleManaged, |
| 889 | ProvenanceState::Managed { |
| 890 | spec: m.spec.clone(), |
| 891 | safe_url: m.url.as_deref().map(sanitize_url_for_display), |
| 892 | schema_version: m.schema_version, |
| 893 | }, |
| 894 | integrity, |
| 895 | ) |
| 896 | } |
| 897 | } |
| 898 | } |
| 899 | |
| 900 | // ── cross-root classification ──────────────────────────────────────────────── |
| 901 | |
| 902 | fn classify_cross_root(skills: &mut [AuditedSkill]) { |
| 903 | // Group by canonical name preserving first-seen order (catalog precedence). |
| 904 | let mut by_name: HashMap<String, Vec<usize>> = HashMap::new(); |
| 905 | for (idx, skill) in skills.iter().enumerate() { |
| 906 | by_name |
| 907 | .entry(skill.id.canonical_name.clone()) |
| 908 | .or_default() |
| 909 | .push(idx); |
| 910 | } |
| 911 | |
| 912 | let owned_names: HashSet<String> = skills |
| 913 | .iter() |
| 914 | .filter(|s| s.root.is_writable_owned()) |
| 915 | .map(|s| s.id.canonical_name.clone()) |
| 916 | .collect(); |
| 917 | |
| 918 | for indices in by_name.values() { |
| 919 | if indices.is_empty() { |
| 920 | continue; |
| 921 | } |
| 922 | |
| 923 | // Runtime-active winners: among copies whose root is active_for_runtime, |
| 924 | // the earliest in catalog order wins. Audit-only roots stay InactiveSource. |
| 925 | let runtime_indices: Vec<usize> = indices |
| 926 | .iter() |
| 927 | .copied() |
| 928 | .filter(|&i| skills[i].root.active_for_runtime) |
| 929 | .collect(); |
| 930 | // Already in scan order which follows catalog precedence. |
| 931 | if let Some(&winner) = runtime_indices.first() { |
| 932 | let winner_id = skills[winner].id.clone(); |
| 933 | for &idx in &runtime_indices { |
| 934 | if idx == winner { |
| 935 | if !matches!(skills[idx].precedence, PrecedenceState::InactiveSource) { |
| 936 | skills[idx].precedence = PrecedenceState::Active; |
| 937 | } |
| 938 | } else { |
| 939 | skills[idx].precedence = PrecedenceState::ShadowedBy(winner_id.clone()); |
| 940 | } |
| 941 | } |
| 942 | } |
| 943 | |
| 944 | // Duplicate / conflict among all copies (including inactive). |
| 945 | let digests: Vec<(usize, Option<String>)> = indices |
| 946 | .iter() |
| 947 | .map(|&i| { |
| 948 | let d = match &skills[i].digest { |
| 949 | DigestState::Known(s) => Some(s.clone()), |
| 950 | DigestState::Unknown(_) => None, |
| 951 | }; |
| 952 | (i, d) |
| 953 | }) |
| 954 | .collect(); |
| 955 | |
| 956 | for &(i, ref di) in &digests { |
| 957 | for &(j, ref dj) in &digests { |
| 958 | if i >= j { |
| 959 | continue; |
| 960 | } |
| 961 | match (di, dj) { |
| 962 | (Some(a), Some(b)) if a == b => { |
| 963 | let other = skills[j].id.clone(); |
| 964 | if skills[i].exact_duplicate_of.is_none() { |
| 965 | skills[i].exact_duplicate_of = Some(other); |
| 966 | } else { |
| 967 | let other = skills[i].id.clone(); |
| 968 | if skills[j].exact_duplicate_of.is_none() { |
| 969 | skills[j].exact_duplicate_of = Some(other); |
| 970 | } |
| 971 | } |
| 972 | } |
| 973 | (Some(_), Some(_)) => { |
| 974 | let id_j = skills[j].id.clone(); |
| 975 | let id_i = skills[i].id.clone(); |
| 976 | skills[i].conflicts_with.push(id_j); |
| 977 | skills[j].conflicts_with.push(id_i); |
| 978 | } |
| 979 | _ => {} |
| 980 | } |
| 981 | } |
| 982 | } |
| 983 | } |
| 984 | |
| 985 | for skill in skills.iter_mut() { |
| 986 | if skill.source_kind == SkillSourceKind::CompatibleExternal |
| 987 | && !owned_names.contains(&skill.id.canonical_name) |
| 988 | && matches!(skill.parser, ParserState::Valid | ParserState::Warning(_)) |
| 989 | && matches!(skill.digest, DigestState::Known(_)) |
| 990 | && !skill.path_unsafe |
| 991 | { |
| 992 | skill.import_candidate = true; |
| 993 | } |
| 994 | } |
| 995 | } |
| 996 | |
| 997 | #[cfg(test)] |
| 998 | mod tests { |
| 999 | use super::*; |
| 1000 | use tempfile::TempDir; |
| 1001 | |
| 1002 | fn write_skill(dir: &Path, name: &str, description: &str, body: &str) { |
| 1003 | let skill_dir = dir.join(name); |
| 1004 | fs::create_dir_all(&skill_dir).unwrap(); |
| 1005 | fs::write( |
| 1006 | skill_dir.join("SKILL.md"), |
| 1007 | format!("---\nname: {name}\ndescription: {description}\n---\n{body}\n"), |
| 1008 | ) |
| 1009 | .unwrap(); |
| 1010 | } |
| 1011 | |
| 1012 | #[test] |
| 1013 | fn owned_only_skips_compatible_and_codex() { |
| 1014 | let tmp = TempDir::new().unwrap(); |
| 1015 | let workspace = tmp.path().join("ws"); |
| 1016 | let home = tmp.path().join("home"); |
| 1017 | write_skill( |
| 1018 | &workspace.join(".codewhale").join("skills"), |
| 1019 | "owned", |
| 1020 | "owned skill", |
| 1021 | "body", |
| 1022 | ); |
| 1023 | write_skill( |
| 1024 | &workspace.join(".claude").join("skills"), |
| 1025 | "claude", |
| 1026 | "claude skill", |
| 1027 | "body", |
| 1028 | ); |
| 1029 | write_skill( |
| 1030 | &workspace.join(".codex").join("skills"), |
| 1031 | "codex", |
| 1032 | "codex skill", |
| 1033 | "body", |
| 1034 | ); |
| 1035 | |
| 1036 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1037 | let names: Vec<_> = snap.skills.iter().map(|s| s.name.as_str()).collect(); |
| 1038 | assert_eq!(names, vec!["owned"]); |
| 1039 | assert!(!snap.roots.iter().any(|r| matches!( |
| 1040 | r.kind, |
| 1041 | SkillRootKind::CompatibleProject(_) | SkillRootKind::CompatibleGlobal(_) |
| 1042 | ))); |
| 1043 | } |
| 1044 | |
| 1045 | #[test] |
| 1046 | fn compatible_includes_codex_without_activating_runtime_precedence() { |
| 1047 | let tmp = TempDir::new().unwrap(); |
| 1048 | let workspace = tmp.path().join("ws"); |
| 1049 | let home = tmp.path().join("home"); |
| 1050 | write_skill( |
| 1051 | &workspace.join(".codewhale").join("skills"), |
| 1052 | "shared", |
| 1053 | "owned", |
| 1054 | "owned-body", |
| 1055 | ); |
| 1056 | write_skill( |
| 1057 | &workspace.join(".codex").join("skills"), |
| 1058 | "shared", |
| 1059 | "codex", |
| 1060 | "codex-body", |
| 1061 | ); |
| 1062 | |
| 1063 | let snap = scan(&workspace, Some(&home), SkillAuditMode::Compatible, None); |
| 1064 | assert_eq!(snap.skills.len(), 2); |
| 1065 | let codex = snap |
| 1066 | .skills |
| 1067 | .iter() |
| 1068 | .find(|s| { |
| 1069 | matches!( |
| 1070 | s.root.kind, |
| 1071 | SkillRootKind::CompatibleProject(super::super::roots::CompatibleHarness::Codex) |
| 1072 | ) |
| 1073 | }) |
| 1074 | .expect("codex copy"); |
| 1075 | assert_eq!(codex.precedence, PrecedenceState::InactiveSource); |
| 1076 | assert!(!codex.root.active_for_runtime); |
| 1077 | |
| 1078 | let owned = snap |
| 1079 | .skills |
| 1080 | .iter() |
| 1081 | .find(|s| s.root.kind == SkillRootKind::CodeWhaleProject) |
| 1082 | .expect("owned"); |
| 1083 | assert_eq!(owned.precedence, PrecedenceState::Active); |
| 1084 | } |
| 1085 | |
| 1086 | #[test] |
| 1087 | fn expanding_owned_scan_matches_fresh_compatible_scan() { |
| 1088 | let tmp = TempDir::new().unwrap(); |
| 1089 | let workspace = tmp.path().join("ws"); |
| 1090 | let home = tmp.path().join("home"); |
| 1091 | write_skill( |
| 1092 | &workspace.join(".codewhale").join("skills"), |
| 1093 | "shared", |
| 1094 | "owned", |
| 1095 | "owned-body", |
| 1096 | ); |
| 1097 | write_skill( |
| 1098 | &workspace.join(".agents").join("skills"), |
| 1099 | "shared", |
| 1100 | "external conflict", |
| 1101 | "external-body", |
| 1102 | ); |
| 1103 | write_skill( |
| 1104 | &workspace.join(".codex").join("skills"), |
| 1105 | "candidate", |
| 1106 | "import candidate", |
| 1107 | "candidate-body", |
| 1108 | ); |
| 1109 | |
| 1110 | let owned = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1111 | let expanded = |
| 1112 | expand_owned_scan_to_compatible(&workspace, Some(&home), None, &owned.skills, None); |
| 1113 | let fresh = scan(&workspace, Some(&home), SkillAuditMode::Compatible, None); |
| 1114 | |
| 1115 | assert_eq!(expanded.scan_mode, SkillAuditMode::Compatible); |
| 1116 | assert_eq!(expanded.roots, fresh.roots); |
| 1117 | assert_eq!(expanded.skills, fresh.skills); |
| 1118 | } |
| 1119 | |
| 1120 | #[test] |
| 1121 | fn detects_shadow_duplicate_and_conflict() { |
| 1122 | let tmp = TempDir::new().unwrap(); |
| 1123 | let workspace = tmp.path().join("ws"); |
| 1124 | let home = tmp.path().join("home"); |
| 1125 | |
| 1126 | // Identical package content → exact duplicate (after shadowing). |
| 1127 | let identical = "---\nname: shared\ndescription: same\n---\nbody\n"; |
| 1128 | fs::create_dir_all(workspace.join(".agents").join("skills").join("shared")).unwrap(); |
| 1129 | fs::write( |
| 1130 | workspace |
| 1131 | .join(".agents") |
| 1132 | .join("skills") |
| 1133 | .join("shared") |
| 1134 | .join("SKILL.md"), |
| 1135 | identical, |
| 1136 | ) |
| 1137 | .unwrap(); |
| 1138 | fs::create_dir_all(workspace.join(".claude").join("skills").join("shared")).unwrap(); |
| 1139 | fs::write( |
| 1140 | workspace |
| 1141 | .join(".claude") |
| 1142 | .join("skills") |
| 1143 | .join("shared") |
| 1144 | .join("SKILL.md"), |
| 1145 | identical, |
| 1146 | ) |
| 1147 | .unwrap(); |
| 1148 | // Different content → conflict with the active copy. |
| 1149 | write_skill( |
| 1150 | &workspace.join(".cursor").join("skills"), |
| 1151 | "shared", |
| 1152 | "cursor conflict", |
| 1153 | "different-body", |
| 1154 | ); |
| 1155 | |
| 1156 | let snap = scan(&workspace, Some(&home), SkillAuditMode::Compatible, None); |
| 1157 | let shared: Vec<_> = snap.skills.iter().filter(|s| s.name == "shared").collect(); |
| 1158 | assert_eq!(shared.len(), 3); |
| 1159 | assert!( |
| 1160 | shared |
| 1161 | .iter() |
| 1162 | .any(|s| matches!(s.precedence, PrecedenceState::Active)) |
| 1163 | ); |
| 1164 | assert!( |
| 1165 | shared |
| 1166 | .iter() |
| 1167 | .any(|s| matches!(s.precedence, PrecedenceState::ShadowedBy(_))) |
| 1168 | ); |
| 1169 | assert!(shared.iter().any(|s| s.exact_duplicate_of.is_some())); |
| 1170 | assert!(shared.iter().any(|s| !s.conflicts_with.is_empty())); |
| 1171 | } |
| 1172 | |
| 1173 | #[test] |
| 1174 | fn external_without_owned_peer_is_import_candidate() { |
| 1175 | let tmp = TempDir::new().unwrap(); |
| 1176 | let workspace = tmp.path().join("ws"); |
| 1177 | let home = tmp.path().join("home"); |
| 1178 | fs::create_dir_all(workspace.join(".codewhale").join("skills")).unwrap(); |
| 1179 | write_skill( |
| 1180 | &workspace.join(".claude").join("skills"), |
| 1181 | "from-claude", |
| 1182 | "desc", |
| 1183 | "body", |
| 1184 | ); |
| 1185 | |
| 1186 | let snap = scan(&workspace, Some(&home), SkillAuditMode::Compatible, None); |
| 1187 | let skill = snap |
| 1188 | .skills |
| 1189 | .iter() |
| 1190 | .find(|s| s.name == "from-claude") |
| 1191 | .expect("skill"); |
| 1192 | assert!(skill.import_candidate); |
| 1193 | assert_eq!(skill.available_actions, vec![SkillActionKind::Import]); |
| 1194 | assert_eq!(skill.source_kind, SkillSourceKind::CompatibleExternal); |
| 1195 | } |
| 1196 | |
| 1197 | #[test] |
| 1198 | fn external_conflicting_with_owned_still_offers_import() { |
| 1199 | let tmp = TempDir::new().unwrap(); |
| 1200 | let workspace = tmp.path().join("ws"); |
| 1201 | let home = tmp.path().join("home"); |
| 1202 | write_skill( |
| 1203 | &workspace.join(".codewhale").join("skills"), |
| 1204 | "shared", |
| 1205 | "desc", |
| 1206 | "owned-body", |
| 1207 | ); |
| 1208 | write_skill( |
| 1209 | &workspace.join(".claude").join("skills"), |
| 1210 | "shared", |
| 1211 | "desc", |
| 1212 | "external-body", |
| 1213 | ); |
| 1214 | |
| 1215 | let snap = scan(&workspace, Some(&home), SkillAuditMode::Compatible, None); |
| 1216 | let external = snap |
| 1217 | .skills |
| 1218 | .iter() |
| 1219 | .find(|s| s.name == "shared" && s.source_kind == SkillSourceKind::CompatibleExternal) |
| 1220 | .expect("external"); |
| 1221 | assert!(!external.import_candidate); |
| 1222 | assert!(!external.conflicts_with.is_empty()); |
| 1223 | assert_eq!(external.available_actions, vec![SkillActionKind::Import]); |
| 1224 | } |
| 1225 | |
| 1226 | #[test] |
| 1227 | fn v1_marker_is_legacy_integrity_and_managed_actions() { |
| 1228 | let tmp = TempDir::new().unwrap(); |
| 1229 | let workspace = tmp.path().join("ws"); |
| 1230 | let home = tmp.path().join("home"); |
| 1231 | let root = workspace.join(".codewhale").join("skills"); |
| 1232 | write_skill(&root, "managed", "desc", "body"); |
| 1233 | fs::write( |
| 1234 | root.join("managed").join(INSTALLED_FROM_MARKER), |
| 1235 | r#"{"spec":"github:o/r","url":"https://user:pass@example.com/x?token=1#frag","checksum":"abc"}"#, |
| 1236 | ) |
| 1237 | .unwrap(); |
| 1238 | |
| 1239 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1240 | let skill = &snap.skills[0]; |
| 1241 | assert_eq!(skill.source_kind, SkillSourceKind::CodeWhaleManaged); |
| 1242 | assert_eq!(skill.integrity, IntegrityState::LegacyMetadataUnknown); |
| 1243 | assert!(skill.available_actions.contains(&SkillActionKind::Update)); |
| 1244 | assert!(skill.available_actions.contains(&SkillActionKind::Remove)); |
| 1245 | if let ProvenanceState::Managed { safe_url, .. } = &skill.provenance { |
| 1246 | let url = safe_url.as_deref().unwrap(); |
| 1247 | assert!(!url.contains("user:pass")); |
| 1248 | assert!(!url.contains("token")); |
| 1249 | assert!(!url.contains("frag")); |
| 1250 | } else { |
| 1251 | panic!("expected managed provenance"); |
| 1252 | } |
| 1253 | } |
| 1254 | |
| 1255 | #[test] |
| 1256 | fn v2_marker_detects_healthy_and_drift() { |
| 1257 | let tmp = TempDir::new().unwrap(); |
| 1258 | let workspace = tmp.path().join("ws"); |
| 1259 | let home = tmp.path().join("home"); |
| 1260 | let root = workspace.join(".codewhale").join("skills"); |
| 1261 | write_skill(&root, "managed", "desc", "body"); |
| 1262 | |
| 1263 | // First scan to learn digest, then write matching v2 marker. |
| 1264 | let preliminary = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1265 | let DigestState::Known(digest) = &preliminary.skills[0].digest else { |
| 1266 | panic!("expected known digest"); |
| 1267 | }; |
| 1268 | fs::write( |
| 1269 | root.join("managed").join(INSTALLED_FROM_MARKER), |
| 1270 | format!(r#"{{"schema_version":2,"spec":"github:o/r","content_digest":"{digest}"}}"#), |
| 1271 | ) |
| 1272 | .unwrap(); |
| 1273 | let healthy = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1274 | assert_eq!(healthy.skills[0].integrity, IntegrityState::Healthy); |
| 1275 | |
| 1276 | fs::write( |
| 1277 | root.join("managed").join(INSTALLED_FROM_MARKER), |
| 1278 | r#"{"schema_version":2,"spec":"github:o/r","content_digest":"deadbeef"}"#, |
| 1279 | ) |
| 1280 | .unwrap(); |
| 1281 | let drift = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1282 | assert_eq!(drift.skills[0].integrity, IntegrityState::LocalContentDrift); |
| 1283 | } |
| 1284 | |
| 1285 | #[test] |
| 1286 | fn legacy_trust_and_digest_bound_trust() { |
| 1287 | let tmp = TempDir::new().unwrap(); |
| 1288 | let workspace = tmp.path().join("ws"); |
| 1289 | let home = tmp.path().join("home"); |
| 1290 | let root = workspace.join(".codewhale").join("skills"); |
| 1291 | write_skill(&root, "managed", "desc", "body"); |
| 1292 | fs::write( |
| 1293 | root.join("managed").join(INSTALLED_FROM_MARKER), |
| 1294 | r#"{"spec":"github:o/r","checksum":"x"}"#, |
| 1295 | ) |
| 1296 | .unwrap(); |
| 1297 | fs::write(root.join("managed").join(TRUSTED_MARKER), "trusted\n").unwrap(); |
| 1298 | |
| 1299 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1300 | assert_eq!(snap.skills[0].trust, TrustState::LegacyAdvisory); |
| 1301 | |
| 1302 | let DigestState::Known(digest) = &snap.skills[0].digest else { |
| 1303 | panic!("digest"); |
| 1304 | }; |
| 1305 | fs::write( |
| 1306 | root.join("managed").join(TRUSTED_MARKER), |
| 1307 | format!(r#"{{"schema_version":2,"content_digest":"{digest}"}}"#), |
| 1308 | ) |
| 1309 | .unwrap(); |
| 1310 | let trusted = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1311 | assert!(matches!( |
| 1312 | trusted.skills[0].trust, |
| 1313 | TrustState::TrustedForDigest(_) |
| 1314 | )); |
| 1315 | |
| 1316 | fs::write( |
| 1317 | root.join("managed").join(TRUSTED_MARKER), |
| 1318 | r#"{"schema_version":2,"content_digest":"stale"}"#, |
| 1319 | ) |
| 1320 | .unwrap(); |
| 1321 | let stale = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1322 | assert_eq!(stale.skills[0].trust, TrustState::TrustStale); |
| 1323 | } |
| 1324 | |
| 1325 | #[test] |
| 1326 | fn oversized_skill_md_is_fail_closed() { |
| 1327 | let tmp = TempDir::new().unwrap(); |
| 1328 | let workspace = tmp.path().join("ws"); |
| 1329 | let home = tmp.path().join("home"); |
| 1330 | let skill_dir = workspace.join(".codewhale").join("skills").join("big"); |
| 1331 | fs::create_dir_all(&skill_dir).unwrap(); |
| 1332 | let huge = format!( |
| 1333 | "---\nname: big\ndescription: x\n---\n{}", |
| 1334 | "x".repeat(AUDIT_MAX_SKILL_MD_BYTES as usize + 64) |
| 1335 | ); |
| 1336 | fs::write(skill_dir.join("SKILL.md"), huge).unwrap(); |
| 1337 | |
| 1338 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1339 | assert_eq!(snap.skills[0].parser, ParserState::Oversized); |
| 1340 | assert!(snap.skills[0].available_actions.is_empty()); |
| 1341 | } |
| 1342 | |
| 1343 | #[test] |
| 1344 | fn readiness_missing_stays_unknown() { |
| 1345 | let tmp = TempDir::new().unwrap(); |
| 1346 | let workspace = tmp.path().join("ws"); |
| 1347 | let home = tmp.path().join("home"); |
| 1348 | write_skill(&workspace.join(".codewhale").join("skills"), "a", "d", "b"); |
| 1349 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1350 | assert_eq!(snap.skills[0].readiness, ReadinessState::Unknown); |
| 1351 | } |
| 1352 | |
| 1353 | #[test] |
| 1354 | fn bundled_name_alone_is_not_built_in() { |
| 1355 | let tmp = TempDir::new().unwrap(); |
| 1356 | let workspace = tmp.path().join("ws"); |
| 1357 | let home = tmp.path().join("home"); |
| 1358 | // `pdf` is a bundled name, but custom body must not classify as BuiltIn. |
| 1359 | write_skill( |
| 1360 | &workspace.join(".codewhale").join("skills"), |
| 1361 | "pdf", |
| 1362 | "user override", |
| 1363 | "not-the-bundled-body", |
| 1364 | ); |
| 1365 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1366 | assert_eq!(snap.skills[0].source_kind, SkillSourceKind::CodeWhaleManual); |
| 1367 | assert!(snap.skills[0].available_actions.is_empty()); |
| 1368 | } |
| 1369 | |
| 1370 | #[test] |
| 1371 | fn managed_marker_wins_over_bundled_name() { |
| 1372 | let tmp = TempDir::new().unwrap(); |
| 1373 | let workspace = tmp.path().join("ws"); |
| 1374 | let home = tmp.path().join("home"); |
| 1375 | let root = workspace.join(".codewhale").join("skills"); |
| 1376 | // Bundled command name + different body + install marker → managed. |
| 1377 | write_skill(&root, "pdf", "registry pdf", "community-body"); |
| 1378 | fs::write( |
| 1379 | root.join("pdf").join(INSTALLED_FROM_MARKER), |
| 1380 | r#"{"spec":"github:o/pdf-skill","checksum":"abc"}"#, |
| 1381 | ) |
| 1382 | .unwrap(); |
| 1383 | |
| 1384 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1385 | assert_eq!( |
| 1386 | snap.skills[0].source_kind, |
| 1387 | SkillSourceKind::CodeWhaleManaged |
| 1388 | ); |
| 1389 | assert!( |
| 1390 | snap.skills[0] |
| 1391 | .available_actions |
| 1392 | .contains(&SkillActionKind::Update) |
| 1393 | ); |
| 1394 | assert!( |
| 1395 | snap.skills[0] |
| 1396 | .available_actions |
| 1397 | .contains(&SkillActionKind::Remove) |
| 1398 | ); |
| 1399 | assert!( |
| 1400 | snap.skills[0] |
| 1401 | .available_actions |
| 1402 | .contains(&SkillActionKind::Trust) |
| 1403 | ); |
| 1404 | } |
| 1405 | |
| 1406 | #[test] |
| 1407 | fn exact_bundled_content_is_built_in() { |
| 1408 | let tmp = TempDir::new().unwrap(); |
| 1409 | let workspace = tmp.path().join("ws"); |
| 1410 | let home = tmp.path().join("home"); |
| 1411 | let root = workspace.join(".codewhale").join("skills"); |
| 1412 | fs::create_dir_all(root.join("pdf")).unwrap(); |
| 1413 | fs::write( |
| 1414 | root.join("pdf").join("SKILL.md"), |
| 1415 | include_str!("../../assets/skills/pdf/SKILL.md"), |
| 1416 | ) |
| 1417 | .unwrap(); |
| 1418 | |
| 1419 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1420 | assert_eq!(snap.skills[0].source_kind, SkillSourceKind::BuiltIn); |
| 1421 | assert!(snap.skills[0].available_actions.is_empty()); |
| 1422 | } |
| 1423 | |
| 1424 | #[cfg(unix)] |
| 1425 | #[test] |
| 1426 | fn symlink_installed_from_marker_is_fail_closed() { |
| 1427 | let tmp = TempDir::new().unwrap(); |
| 1428 | let workspace = tmp.path().join("ws"); |
| 1429 | let home = tmp.path().join("home"); |
| 1430 | let root = workspace.join(".codewhale").join("skills"); |
| 1431 | write_skill(&root, "managed", "desc", "body"); |
| 1432 | let outside = tmp.path().join("outside-marker.json"); |
| 1433 | fs::write(&outside, r#"{"spec":"github:o/r","checksum":"x"}"#).unwrap(); |
| 1434 | std::os::unix::fs::symlink(&outside, root.join("managed").join(INSTALLED_FROM_MARKER)) |
| 1435 | .unwrap(); |
| 1436 | |
| 1437 | let snap = scan(&workspace, Some(&home), SkillAuditMode::OwnedOnly, None); |
| 1438 | assert!(snap.skills[0].path_unsafe); |
| 1439 | assert!(matches!( |
| 1440 | snap.skills[0].digest, |
| 1441 | DigestState::Unknown(DigestUnknownReason::SymlinkPresent) |
| 1442 | )); |
| 1443 | assert_eq!( |
| 1444 | snap.skills[0].integrity, |
| 1445 | IntegrityState::BrokenManagedInstall |
| 1446 | ); |
| 1447 | assert!(snap.skills[0].available_actions.is_empty()); |
| 1448 | } |
| 1449 | |
| 1450 | #[test] |
| 1451 | fn sanitize_url_strips_secrets() { |
| 1452 | assert_eq!( |
| 1453 | sanitize_url_for_display("https://user:pw@host/path?token=1#x"), |
| 1454 | "https://host/path" |
| 1455 | ); |
| 1456 | } |
| 1457 | |
| 1458 | struct AlwaysReady; |
| 1459 | impl SkillReadinessProvider for AlwaysReady { |
| 1460 | fn readiness_for(&self, _: &AuditedSkillId) -> Option<ReadinessState> { |
| 1461 | Some(ReadinessState::Ready) |
| 1462 | } |
| 1463 | } |
| 1464 | |
| 1465 | #[test] |
| 1466 | fn readiness_provider_is_consulted_when_present() { |
| 1467 | let tmp = TempDir::new().unwrap(); |
| 1468 | let workspace = tmp.path().join("ws"); |
| 1469 | let home = tmp.path().join("home"); |
| 1470 | write_skill(&workspace.join(".codewhale").join("skills"), "a", "d", "b"); |
| 1471 | let snap = scan( |
| 1472 | &workspace, |
| 1473 | Some(&home), |
| 1474 | SkillAuditMode::OwnedOnly, |
| 1475 | Some(&AlwaysReady), |
| 1476 | ); |
| 1477 | assert_eq!(snap.skills[0].readiness, ReadinessState::Ready); |
| 1478 | } |
| 1479 | } |
| 1480 |