Struct PromptContext 

pub struct PromptContext<'a> {

    pub workspace_dir: &'a Path,
    pub agent_workspace_dir: &'a Path,
    pub model_name: &'a str,
    pub tools: &'a [Box<dyn Tool>],
    pub hidden_builtin_names: Option<&'a HashSet<String>>,
    pub skills: &'a [Skill],
    pub skills_prompt_mode: SkillsPromptInjectionMode,
    pub identity_config: Option<&'a IdentityConfig>,
    pub interaction: Option<&'a InteractionContext>,
    pub dispatcher_instructions: &'a str,
    pub sends_native_tool_specs: bool,
    pub security_summary: Option<String>,
    pub autonomy_level: AutonomyLevel,
    pub inject_memory: bool,
    pub shell_profile: Option<ShellProfile>,
}

Fields

workspace_dir: &'a Path

agent_workspace_dir: &'a Path

model_name: &'a str

tools: &'a [Box<dyn Tool>]

hidden_builtin_names: Option<&'a HashSet<String>>

Render-time projection of unselected built-in schemas. Execution still resolves the complete policy-admitted registry.

skills: &'a [Skill]

skills_prompt_mode: SkillsPromptInjectionMode

identity_config: Option<&'a IdentityConfig>

interaction: Option<&'a InteractionContext>

dispatcher_instructions: &'a str

sends_native_tool_specs: bool

True when the provider request carries native tool specs. In that mode the prompt must not duplicate the same tool catalog in prose.

security_summary: Option<String>

Pre-rendered security policy summary for inclusion in the Safety prompt section. When present, the LLM sees the concrete constraints (allowed commands, forbidden paths, autonomy level) so it can plan tool calls without trial-and-error. See

autonomy_level: AutonomyLevel

Autonomy level the prompt describes. Full autonomy omits “ask before acting” instructions so the model executes tools directly without simulating approval. Policy-aware builders separately supply any always_ask exceptions without expanding this public context struct.

inject_memory: bool

False for isolated / ACP sessions created with exclude_memory: true. Mirrors system_prompt::build_system_prompt_with_mode_and_autonomy’s inject_memory flag so both prompt builders enforce one policy: when it is false, MEMORY.md is not injected into the provider-visible system prompt.

shell_profile: Option<ShellProfile>

The shell the runtime adapter will spawn, or None for a shell-less runtime (which omits the Shell: field and the dialect guidance). Resolved from RuntimeAdapter::shell_profile so the reported shell cannot drift from the executed one.