workspacescan

type AIOptions struct {
	Bin     string        // CLI binary path; empty → "claude" via PATH
	Timeout time.Duration // per-invocation bound; <=0 → aiDefaultTimeout
}

type GitRemotes struct {
	GitHub     []string `json:"github,omitempty"`
	OtherHosts []string `json:"other_hosts,omitempty"`
}

type InstallerPin struct {
	Version, Platform, SHA256 string
}

type LeakFinding struct {
	Path string `json:"path"`
	Kind string `json:"kind"`
	Line int    `json:"line,omitempty"`
	// Source names the attached source once multiple sources merge into one
	// profile; empty otherwise. A separate field rather than a Path prefix,
	// because Path is path-classified downstream (testdata|__tests__|fixtures)
	// and folding a locator in would misclassify leaks under a source whose
	// own path looks like a fixture.
	Source string `json:"source,omitempty"`
}

type ManifestHit struct {
	Path   string `json:"path"`   // slash-separated, relative to the scan root
	Marker string `json:"marker"` // canonical marker id, e.g. "package-lock.json"
}

type Pins struct {
	BaseDigest string
	Features   map[string]string // feature ref -> digest
	Installers map[string]InstallerPin
}

type ScanFacts struct {
	ManifestsFound      []ManifestHit        `json:"manifests_found,omitempty"`
	GitRemotes          GitRemotes           `json:"git_remotes,omitempty"`
	HasDevcontainer     bool                 `json:"has_devcontainer,omitempty"`
	HasDockerfile       bool                 `json:"has_dockerfile,omitempty"`
	UnrecognizedSamples []UnrecognizedSample `json:"unrecognized_samples,omitempty"`
	Truncated           bool                 `json:"truncated,omitempty"`

	// Content-lane evidence: names/keys/hosts only, via anchored capture
	// groups — no file value ever lands here. UNTRUSTED until DeriveProfile
	// re-validates and caps it.
	SecretRequirements []SecretNeed  `json:"secret_requirements,omitempty"`
	ServicesFound      []string      `json:"services_found,omitempty"`
	SuggestedEgress    []string      `json:"suggested_egress,omitempty"`
	SecretFilesPresent []string      `json:"secret_files_present,omitempty"`
	BuildMemoryMiB     int           `json:"build_memory_mib,omitempty"`
	LeakFindings       []LeakFinding `json:"leak_findings,omitempty"`
	// Raw setup-command SIGNALS (not commands): which conventional script/target
	// keys exist. DeriveProfile synthesizes fixed-template SetupCommands from
	// these plus detected package managers — file content never becomes a command.
	ScriptKeys  []string `json:"script_keys,omitempty"`  // package.json scripts: build|test|lint present
	MakeTargets []string `json:"make_targets,omitempty"` // Makefile targets: build|test|install|lint present
	// BuildInputHashes maps a build-input file's path to a hex sha256 of its
	// content (devcontainer.json / Dockerfile) — a digest, not content, so safe.
	BuildInputHashes map[string]string `json:"build_input_hashes,omitempty"`
}

type SecretNeed struct {
	Name     string `json:"name"`
	Kind     string `json:"kind,omitempty"`
	Optional bool   `json:"optional,omitempty"`
}

type SetupCommand struct {
	Stage   string `json:"stage"`   // install | build | test | lint
	Command string `json:"command"` // fixed-template command, never file content
	Source  string `json:"source"`  // what implied it, e.g. "convention:go", "package.json:build"
}

type UnrecognizedSample struct {
	Path    string `json:"path"`
	Content string `json:"content"`
}

type WorkspaceProfile struct {
	Languages       []string `json:"languages,omitempty"`
	PackageManagers []string `json:"package_managers,omitempty"`
	// ToolchainNeeds below reads these two: dispatch env derives from what
	// the scan actually detected, never a platform-wide guess.
	EgressDomains   []string   `json:"egress_domains,omitempty"`
	Tools           []string   `json:"tools,omitempty"`
	GitRemotes      GitRemotes `json:"git_remotes,omitempty"`
	HasDevcontainer bool       `json:"has_devcontainer,omitempty"`
	HasDockerfile   bool       `json:"has_dockerfile,omitempty"`

	// Advisory "needs" fields (content lane, validated by DeriveProfile)
	// inform the operator and never gate a launch or create a grant.
	// SECURITY: SuggestedEgress is content-derived and is NEVER auto-unioned
	// into a run's allowlist — only EgressDomains (filename-keyed) grants
	// that; an operator must promote hosts into ApprovedEgress explicitly.
	// These fields ride in ProfileHash, so a rescan after this changes
	// forces one harmless no-op image rebuild; hash only the image-affecting
	// subset instead if that churn ever bites.
	RequiredSecrets    []SecretNeed `json:"required_secrets,omitempty"`
	ServicesNeeded     []string     `json:"services_needed,omitempty"`
	SuggestedEgress    []string     `json:"suggested_egress,omitempty"`
	SecretFilesPresent []string     `json:"secret_files_present,omitempty"`
	// BuildMemoryMiB is the largest build-heap ceiling detected (JVM -Xmx /
	// Node --max-old-space-size). Advisory sizing hint only, never
	// auto-applied to a run's ResourceLimits.
	BuildMemoryMiB int `json:"build_memory_mib,omitempty"`
	// LeakFindings are content-free reports of suspected committed secret
	// values (path + detector kind + line, never the value). Advisory warning.
	LeakFindings []LeakFinding `json:"leak_findings,omitempty"`
	// SetupCommands are the conventional install/build/test/lint steps
	// implied, synthesized from fixed templates only. Advisory:
	// operator-approved, sandbox-only execution.
	SetupCommands []SetupCommand `json:"setup_commands,omitempty"`
	// ContextHash digests the build-input files' content (devcontainer.json /
	// Dockerfile); it rides ProfileHash so the built-image cache busts when a
	// build input changes even if the detected profile is unchanged. Empty
	// when no build-input files are present.
	ContextHash string `json:"context_hash,omitempty"`
	// Confidence is one of ConfidenceHigh/Medium/Low.
	Confidence  string `json:"confidence"`
	NeedsReview bool   `json:"needs_review,omitempty"`
	// Source is one of SourceDeterministic/SourceAIAssisted.
	Source string `json:"source"`
}