Skip to main content

Prompt System

Architectural Layer: L3 Intelligence
Source Directory: src/prompt
File Count: 16 source files

Prompt registry, template rendering, versioning, and prompt references.

Module Files & Detailed Reference​

Source FileLinesKey TypesResponsibility
builtins.rs351Functions/TraitsEmbedded TOML builtin prompt contracts. Provides compile-time embeddi...
catalog.rs1420CatalogEntry, InMemoryPromptCatalog, PromptContractMetadataImmutable prompt catalog index, directory discovery, and built-in gene...
compiler.rs1130CompilationOptions, DefaultPromptCompiler, EffectivePromptPromptCompiler: Deterministic 7-layer composition and trust-aware prom...
composer.rs508ArtifactEvidenceSection, PromptComposition, PromptLayerEntryTyped 7-layer prompt composition model (Architecture Section 16). Enf...
context.rs836ArtifactEvidence, DurableStateContext, PromptBudgetTyped runtime context abstraction for PromptOS (Architecture Section 8...
contract.rs1233BodyRaw, CompatibilityRaw, EvidenceRawVersioned PromptContract models, hashing, and TOML deserialization.
error.rs79PromptErrorStrongly typed errors for prompt contract validation, rendering, and c...
mod.rs53Functions/TraitsPromptOS core module: versioned PromptContracts, MiniJinja rendering, ...
model_profile.rs360CapabilityRating, ModelProfile, StructuredOutputSupportStrongly typed model capability descriptor for PromptOS prompt adaptat...
parameter.rs20PromptParameterParameter models and limits for prompt template evaluation.
provenance.rs346PromptCompactionMetadata, PromptCompilationTrace, PromptInvocationProvenancePrompt provenance, telemetry, and observability models (Architecture S...
reachability.rs222PromptCiBucket, PromptReachabilityAuthoritative prompt reachability classification (wiring remediation v...
reference.rs246PromptReference, PromptPurposeTyped reference to a versioned prompt contract.
renderer.rs117RenderedPromptDeterministic MiniJinja template rendering under strict byte and param...
strategy.rs163PromptStrategyPromptStrategy enum and deterministic strategy resolution for PromptOS...
v2.rs565ClaimRecord, CompatibilityMetadata, EvidenceRequirementsPromptOS V2 Canonical Domain Contracts, Layered Authority, and Epistem...

src/prompt/builtins.rs​

Primary Responsibility​

Embedded TOML builtin prompt contracts. Provides compile-time embedding of the canonical prompt contracts located under prompts/. All built-in prompts are parsed and validated upon catalog initialization.

Key Functions & Signatures (8 shown)​

// pub
fn load_builtin_contracts() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_builtin_contracts_v2() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_production_contracts() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_compatibility_contracts() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_production_and_compatibility_contracts() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_test_fixture_contracts() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_deprecated_metadata_contracts() -> Result<Vec<PromptContract>, PromptError>;
// pub
fn load_all_builtin_contracts() -> Result<Vec<PromptContract>, PromptError>;

src/prompt/catalog.rs​

Primary Responsibility​

Immutable prompt catalog index, directory discovery, and built-in genesis contracts.

Types & Trait Declarations​

  • Structs (3): CatalogEntry, InMemoryPromptCatalog, PromptContractMetadata
  • Traits (1): PromptCatalog

Key Functions & Signatures (15 shown)​

// pub
fn is_protected_contract_id(id: &str) -> bool;
// pub
fn is_layer0_contract_id(id: &str) -> bool;
// pub
fn is_behavioral_contract_id(id: &str) -> bool;
// private
fn get(&self, id: &str, version: u32) -> Result<&PromptContract, PromptError>;
// private
fn contains(&self, id: &str, version: u32) -> bool;
// private
fn list(&self) -> Vec<PromptContractMetadata>;
// private
fn source_kind(&self, id: &str, version: u32) -> Option<PromptSourceKind>;
// private
fn project_guidance_for(&self, _id: &str, _version: u32) -> Vec<CatalogEntry>;
// private
fn is_overrideable(&self, id: &str, _version: u32) -> bool;
// private
fn resolve_canonical(&self, id: &str, version: u32) -> Result<&PromptContract, PromptError>;
// private
fn audit_inventory(&self) -> crate::prompt::v2::PromptInventoryAudit;
// pub
fn new() -> Self;
// pub
fn with_production_builtins() -> Self;
// pub
fn with_production_builtins_and_workspace(workspace_root: &Path) -> Self;
// pub
fn with_builtins() -> Self;

Unit Verification (6 inline tests)​

Tests defined within module: test_catalog_builtins_and_source_kind, test_override_precedence_workspace_over_project_over_builtin, test_protected_contract_override_strictly_rejected, test_behavioral_override_becomes_guidance_not_replacement, test_duplicate_in_same_scope_rejected, test_path_safety_checks

src/prompt/compiler.rs​

Primary Responsibility​

PromptCompiler: Deterministic 7-layer composition and trust-aware prompt compilation pipeline. # Responsibilities & Architectural Ownership The PromptCompiler is the canonical engine that transforms an abstract [PromptContract] and structured runtime [PromptContext] into an [EffectivePrompt]. # The Seven Canonical Layers (L0 through L6) The compiler enforces a strictly ordered, non-reorderable 7-layer composition model: 1. L0 — Runtime Safety Invariants: Immutable P0 constraints (RUNTIME_SAFETY_INVARIANTS), unprunable, prepended unconditionally before any other content. 2. L1 — Agent Role & Profile: Operating role identity and stage constraints (P1, unprunable). 3. L2 — Workflow Step Objective & Contract: Current task purpose, acceptance criteria, expected output format, and user intent wrapped in &lt;user_intent&gt; (P1, unprunable). 4. L3 — Project Charter & Durable State: .m31a/project/ project memory extracts (P2, compactable). 5. L4 — Upstream Artifact Evidence & Tool Results: Prior step outputs and tool execution logs wrapped in &lt;untrusted_evidence&gt; with closing tag escaping (P2/P3, compactable). 6. L5 — Repository Topology & Context: Repository file extracts and AST summaries (P3/P4, disposable). 7. L6 — Quality Gate & Verification Checklist: Acceptance assertions and checklist items (P1, unprunable). # Trust Handling & Delimiter Smuggling Defense External and untrusted data (user intent, tool outputs, repository files, git diffs) are encapsulated in typed XML trust envelopes (&lt;user_intent&gt; and &lt;untrusted_evidence&gt;). Any occurrences of closing tags within payload text are escaped to prevent prompt injection delimiter hijacking. External content never gains system-level instruction authority. # Deterministic Compilation & Hashing Compilation is strictly deterministic. On identical contracts, versions, parameters, and contexts, PromptCompiler produces byte-for-byte identical assembled prompts and cryptographic SHA-256 content_hash values. No random UUIDs, timestamps, or unordered collections are used in hashing.

Types & Trait Declarations​

  • Structs (3): CompilationOptions, DefaultPromptCompiler, EffectivePrompt
  • Traits (1): PromptCompiler

Key Functions & Signatures (15 shown)​

// private
fn default() -> Self;
// pub
fn with_profile(mut self, profile: ModelProfile) -> Self;
// pub
fn with_strategy(mut self, strategy: PromptStrategy) -> Self;
// pub
fn with_budget(mut self, max_bytes: usize) -> Self;
// pub
fn with_strict_parameters(mut self, strict: bool) -> Self;
// pub
fn with_source_kind(mut self, source_kind: PromptSourceKind) -> Self;
// pub
fn invocation_provenance(&self) -> Option<PromptInvocationProvenance>;
// pub
fn source_kind(&self) -> PromptSourceKind;
// private
fn compile(&self,
        contract: &PromptContract,
        context: &PromptContext,
        options: &CompilationOptio;
// private
fn compile_with_guidance(&self,
        catalog: &dyn crate::prompt::catalog::PromptCatalog,
        contract: &PromptCo;
// pub
fn new() -> Self;
// pub
fn compile_from_catalog(&self,
        catalog: &dyn crate::prompt::catalog::PromptCatalog,
        id: &str,
        ve;
// pub
fn compile_from_catalog_with_guidance(&self,
        catalog: &dyn crate::prompt::catalog::PromptCatalog,
        id: &s;
// pub
fn calculate_effective_hash(contract: &PromptContract,
        parameters: &BTreeMap<String, String>,
        layers: &[;
// pub
fn calculate_legacy_hash(contract: &PromptContract,
        parameters: &BTreeMap<String, String>,
        layers: &[Pro;

src/prompt/composer.rs​

Primary Responsibility​

Typed 7-layer prompt composition model (Architecture Section 16). Enforces the canonical layered prompt hierarchy: - Layer 0: Runtime Safety Invariants (P0, immutable, unprunable) - Layer 1: Agent Role & Profile (P1, unprunable) - Layer 2: Workflow Step Objective & Output Contract (P1, unprunable) - Layer 3: Project Charter & Boundaries (P2, compactable) - Layer 4: Upstream Artifact Evidence & Tool Results (P2/P3, compactable) - Layer 5: Repository Constraints & Context (P3/P4, disposable) - Layer 6: Quality Gate Assertions (P1, unprunable)

Types & Trait Declarations​

  • Structs (3): ArtifactEvidenceSection, PromptComposition, PromptLayerEntry
  • Enums (1): PromptLayerKind

Key Functions & Signatures (15 shown)​

// pub
fn as_str(&self) -> &'static str;
// pub
fn priority(&self) -> u8;
// pub
fn is_prunable(&self) -> bool;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// pub
fn new(kind: PromptLayerKind,
        name: impl Into<String>,
        content: impl Into<String>,
        trust_level: ;
// pub
fn new(role_profile: impl Into<String>,
        step_objective: impl Into<String>,
        quality_gate_assertions: Vec<;
// pub
fn from_context(context: &PromptContext,
        role_profile: impl Into<String>,
        rendered_objective: impl Into<;
// pub
fn with_charter(mut self, charter: impl Into<String>) -> Self;
// pub
fn with_upstream_artifact(mut self,
        source_step: impl Into<String>,
        artifact_name: impl Into<String>,
  ;
// pub
fn with_repo_context(mut self, repo_context: impl Into<String>) -> Self;
// pub
fn verify_safety_invariants(&self) -> bool;
// pub
fn validate(&self) -> Result<(), PromptError>;
// pub
fn ordered_layers(&self) -> Vec<PromptLayerEntry>;
// pub
fn to_full_prompt(&self) -> String;
// pub
fn to_context_sections(&self, tokenizer: &TokenizerAdapter) -> Vec<ContextSection>;

src/prompt/context.rs​

Primary Responsibility​

Typed runtime context abstraction for PromptOS (Architecture Section 8 & Section 46). Enforces strong domain typing, explicit trust classification, and separation of ephemeral reasoning data from durable project memory.

Types & Trait Declarations​

  • Structs (12): ArtifactEvidence, DurableStateContext, PromptBudget, PromptContext, QualityGateContext, RepoContext, RepoFileContext, TaskObjectiveContext, ToolOutputEvidence, TrustedPolicyContext, TrustedSystemContext, UserInputContext
  • Enums (1): MissionStage

Key Functions & Signatures (15 shown)​

// pub
fn as_str(&self) -> &'static str;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// private
fn default() -> Self;
// pub
fn new(raw_intent: impl Into<String>) -> Self;
// pub
fn to_trust_envelope(&self) -> String;
// pub
fn new(task_objective: impl Into<String>) -> Self;
// pub
fn with_criteria(mut self, criteria: Vec<String>) -> Self;
// pub
fn with_output_format(mut self, format: impl Into<String>) -> Self;
// pub
fn load_from_workspace(workspace_root: &std::path::Path) -> Self;
// pub
fn new(source_step: impl Into<String>,
        artifact_name: impl Into<String>,
        excerpt: impl Into<String>,) ->;
// pub
fn to_trust_envelope(&self) -> String;
// pub
fn new(tool_name: impl Into<String>,
        call_id: impl Into<String>,
        output: impl Into<String>,
        is_e;
// pub
fn to_trust_envelope(&self) -> String;
// pub
fn new(path: impl Into<String>, content: impl Into<String>) -> Self;
// pub
fn to_trust_envelope(&self) -> String;

src/prompt/contract.rs​

  • Source Link: src/prompt/contract.rs on GitHub
  • Lines of Code: 1233 (45643 bytes)
  • Runtime Properties: 🛡️ Enforces Policy Gate | 🛑 Supports Cancellation Tokens

Primary Responsibility​

Versioned PromptContract models, hashing, and TOML deserialization.

Types & Trait Declarations​

  • Structs (16): BodyRaw, CompatibilityRaw, EvidenceRaw, FailureBehaviorRaw, InputParamOptionalRaw, InputParamRequiredRaw, InputsRaw, OutputRaw, PromptContract, PromptFileRaw, PromptMetadataRaw, ReasoningRaw, RoleVisitor, SecurityRaw, TemplateRaw, VerificationRaw

Key Functions & Signatures (15 shown)​

// pub
fn deserialize_flexible_role(deserializer: D) -> Result<AgentRole, D::Error>
where
    D: serde::Deserializer<'de>,;
// private
fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result;
// private
fn visit_str(self, value: &str) -> Result<AgentRole, E>
        where
            E: serde::de::Error,;
// pub
fn deserialize_optional_flexible_role(deserializer: D,) -> Result<Option<AgentRole>, D::Error>
where
    D: serde::Deser;
// pub
fn parse_role_flexible(s: &str) -> Result<AgentRole, String>;
// pub
fn new(id: impl Into<String>,
        version: u32,
        role: AgentRole,
        description: impl Into<String>,
   ;
// pub
fn new_v2(id: impl Into<String>,
        version: u32,
        role: AgentRole,
        kind: PromptKind,
        stage:;
// pub
fn is_v2(&self) -> bool;
// pub
fn is_deprecated(&self) -> bool;
// pub
fn canonical_id(&self) -> &str;
// pub
fn requires_falsification(&self) -> bool;
// pub
fn requires_evidence(&self) -> bool;
// pub
fn calculate_hash(id: &str,
        version: u32,
        role: AgentRole,
        description: &str,
        input_para;
// pub
fn calculate_hash_v2(id: &str,
        version: u32,
        kind: PromptKind,
        role: AgentRole,
        descript;
// pub
fn from_toml_str(toml_str: &str) -> Result<Self, PromptError>;

src/prompt/error.rs​

Primary Responsibility​

Strongly typed errors for prompt contract validation, rendering, and catalog operations.

Types & Trait Declarations​

  • Enums (1): PromptError

src/prompt/mod.rs​

Primary Responsibility​

PromptOS core module: versioned PromptContracts, MiniJinja rendering, composition, and catalog.

src/prompt/model_profile.rs​

Primary Responsibility​

Strongly typed model capability descriptor for PromptOS prompt adaptation (Section 11, Artifacts F & G). # Core Invariant: "The model proposes. The runtime decides." ModelProfile represents trusted runtime facts about a model's operational characteristics, physical context boundaries, and tool-calling fidelity. A ModelProfile is a capability descriptor, never an authority or privilege mechanism. Models cannot self-report, elevate, or negotiate their own profile attributes.

Types & Trait Declarations​

  • Structs (2): CapabilityRating, ModelProfile
  • Enums (1): StructuredOutputSupport

Key Functions & Signatures (15 shown)​

// pub
fn new(val: u8) -> Result<Self, PromptError>;
// pub
fn clamped(val: u8) -> Self;
// pub
fn as_u8(&self) -> u8;
// pub
fn score(&self) -> u8;
// private
fn try_from(value: u8) -> Result<Self, Self::Error>;
// private
fn from(rating: CapabilityRating) -> Self;
// private
fn eq(&self, other: &u8) -> bool;
// private
fn partial_cmp(&self, other: &u8) -> Option<std::cmp::Ordering>;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// private
fn serialize(&self, serializer: S) -> Result<S::Ok, S::Error>
    where
        S: serde::Serializer,;
// private
fn deserialize(deserializer: D) -> Result<Self, D::Error>
    where
        D: serde::Deserializer<'de>,;
// pub
fn as_str(&self) -> &'static str;
// pub
fn requires_emulated_guidance(&self) -> bool;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// pub
fn conservative_default() -> Self;

src/prompt/parameter.rs​

Primary Responsibility​

Parameter models and limits for prompt template evaluation.

Types & Trait Declarations​

  • Structs (1): PromptParameter

src/prompt/provenance.rs​

Primary Responsibility​

Prompt provenance, telemetry, and observability models (Architecture Section 24 & Stage 7). Enforces complete auditability and reproducibility for every model-facing prompt execution without introducing new runtime authority. # Core Distinctions 1. Contract Hash: SHA-256 hash of the unrendered template in [crate::prompt::PromptContract]. 2. Effective Prompt Hash: Cryptographic composite SHA-256 hash of the fully compiled, strategy-adapted, parameter-bound, and trust-delimited prompt. 3. Context Digest: Deterministic cryptographic SHA-256 digest of the structured runtime context manifest. # Privacy & Security Guarantees - Zero raw secrets or credentials recorded in provenance metadata. - Raw prompt bodies are never stored in telemetry records. - Zero chain-of-thought or hidden reasoning prompts recorded.

Types & Trait Declarations​

  • Structs (5): PromptCompactionMetadata, PromptCompilationTrace, PromptInvocationProvenance, PromptLayerMetadata, PromptProvenance
  • Enums (1): PromptSourceKind

Key Functions & Signatures (15 shown)​

// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// pub
fn new(prompt_id: impl Into<String>,
        prompt_version: u32,
        prompt_content_hash: impl Into<String>,
      ;
// pub
fn with_source_kind(mut self, source_kind: PromptSourceKind) -> Self;
// pub
fn with_mission_id(mut self, mission_id: impl Into<String>) -> Self;
// pub
fn with_task_id(mut self, task_id: impl Into<String>) -> Self;
// pub
fn with_agent_id(mut self, agent_id: impl Into<String>) -> Self;
// pub
fn sanitized(&self, redactor: &SecretRedactor) -> Self;
// pub
fn new(kind: PromptLayerKind,
        name: impl Into<String>,
        byte_size: usize,
        trust_level: TrustLevel;
// pub
fn new(budget_max_bytes: usize,
        protected_bytes: usize,
        total_compiled_bytes: usize,
        layers_reta;
// pub
fn has_dropped_layers(&self) -> bool;
// pub
fn to_invocation_provenance(&self) -> PromptInvocationProvenance;
// pub
fn source_kind(&self) -> PromptSourceKind;
// pub
fn has_dropped_layers(&self) -> bool;
// pub
fn retained_layers(&self) -> &[PromptLayerKind];
// pub
fn dropped_layers(&self) -> &[PromptLayerKind];

src/prompt/reachability.rs​

Primary Responsibility​

Authoritative prompt reachability classification (wiring remediation v0.1.1). Every shipped prompt asset has EXACTLY ONE classification describing how (or whether) the production runtime can reach it. Embedding a TOML via include_str! is NOT consumption: a prompt is production-reachable only when its identifier can flow into an actual model invocation through the production runtime spine (PromptReference → task/role binding → PromptCatalog → PromptCompiler → EffectivePrompt → ModelCaller). The CI gate ([ci_bucket]) maps each classification into one of four release buckets. A shipped prompt with no classification fails CI.

Types & Trait Declarations​

  • Enums (2): PromptCiBucket, PromptReachability

Key Functions & Signatures (5 shown)​

// pub
fn as_str(&self) -> &'static str;
// pub
fn as_str(&self) -> &'static str;
// pub
fn classify_builtin(id: &str, version: u32) -> Option<(PromptReachability, PromptCiBucket)>;
// pub
fn canonical_version(id: &str) -> Option<u32>;
// pub
fn builtin_reachability_matrix(catalog: &dyn crate::prompt::PromptCatalog,) -> (Vec<ReachabilityRow>, Vec<UnclassifiedRo;

src/prompt/reference.rs​

Primary Responsibility​

Typed reference to a versioned prompt contract.

Types & Trait Declarations​

  • Structs (1): PromptReference
  • Enums (1): PromptPurpose

Key Functions & Signatures (10 shown)​

// pub
fn infer_from_id(id: &str) -> Self;
// pub
fn as_str(&self) -> &'static str;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// pub
fn new(id: impl Into<String>, version: u32) -> Self;
// pub
fn with_purpose(id: impl Into<String>, version: u32, purpose: PromptPurpose) -> Self;
// pub
fn with_explicit_purpose(mut self, purpose: PromptPurpose) -> Self;
// pub
fn for_role(role: crate::state_machine::agent::AgentRole) -> Self;
// pub
fn parse(s: &str) -> Result<Self, PromptError>;
// pub
fn parse_lenient(s: &str) -> Result<Self, PromptError>;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;

src/prompt/renderer.rs​

Primary Responsibility​

Deterministic MiniJinja template rendering under strict byte and parameter bounds.

Types & Trait Declarations​

  • Structs (1): RenderedPrompt

Key Functions & Signatures (1 shown)​

// pub
fn render_prompt(contract: &PromptContract,
    parameters: &BTreeMap<String, String>,
    strict_parameters: bool,) -> ;

src/prompt/strategy.rs​

Primary Responsibility​

PromptStrategy enum and deterministic strategy resolution for PromptOS (Section 10 & 11). # Core Invariant: "The model proposes. The runtime decides." A single logical prompt contract is executed across different models and operational contexts by applying a runtime-selected [PromptStrategy]. # DeepReasoning Invariant DeepReasoning injects explicit architectural directives (task decomposition, invariant validation, and pre-execution verification checklists). DeepReasoning MUST NOT request or instruct the model to disclose private chain-of-thought or hidden reasoning.

Types & Trait Declarations​

  • Enums (1): PromptStrategy

Key Functions & Signatures (7 shown)​

// pub
fn as_str(&self) -> &'static str;
// pub
fn example_budget(&self) -> usize;
// pub
fn instruction_density(&self) -> &'static str;
// pub
fn structural_directives(&self) -> &'static str;
// pub
fn few_shot_examples(&self) -> Option<&'static str>;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// pub
fn resolve_strategy(_role: AgentRole,
    stage: MissionStage,
    task_complexity_high: bool,
    in_recovery: bool,
  ;

src/prompt/v2.rs​

Primary Responsibility​

PromptOS V2 Canonical Domain Contracts, Layered Authority, and Epistemic Governance. Provides the foundational models and contract invariants for PromptOS V2: 1. Canonical Taxonomy: Deterministic, role- and stage-aware prompt categorization. 2. Explicit Layered Authority: Non-reorderable 7-layer hierarchy from Kernel (L0) to Dynamic Mission (L6). 3. Structured Response Kinds: Strongly typed model response expectations replacing raw string outputs. 4. Evidence & Verification Semantics: Mandatory counterexample search, falsification, and empirical evidence validation. 5. Epistemic Discipline: Categorization into FACT, INFERENCE, ASSUMPTION, PROPOSAL, OBSERVATION, VERIFICATION, UNKNOWN. 6. Production Reachability Awareness: Tracking defined, instantiated, referenced, reachable, production-reachable, exercised, and verified code paths.

Types & Trait Declarations​

  • Structs (8): ClaimRecord, CompatibilityMetadata, EvidenceRequirements, FailurePolicy, OutputContract, PromptInventoryAudit, ReasoningPolicy, VerificationRequirements
  • Enums (9): AuthorityLevel, ClaimVerificationStatus, EpistemicStatus, FailureBehavior, ProductionReachability, PromptKind, PromptResponseKind, ReasoningDepth, ReasoningMode

Key Functions & Signatures (15 shown)​

// pub
fn parse_kind(s: &str) -> Result<Self, PromptError>;
// pub
fn as_str(&self) -> &'static str;
// private
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result;
// pub
fn rank(&self) -> u8;
// pub
fn can_override(&self, target: AuthorityLevel) -> bool;
// pub
fn from_type_str(s: &str) -> Self;
// pub
fn as_str(&self) -> &'static str;
// pub
fn new(response_kind: PromptResponseKind,
        format_type: impl Into<String>,
        schema_uri: Option<String>,
  ;
// pub
fn code_proposal() -> Self;
// pub
fn verification_verdict() -> Self;
// pub
fn failure_diagnosis() -> Self;
// private
fn default() -> Self;
// pub
fn for_implementer() -> Self;
// pub
fn for_verifier() -> Self;
// pub
fn strict_adversarial() -> Self;