innate_core/

cli.rs

//! CLI commands — thin wrapper over KnowledgeBase.

use std::path::PathBuf;

use clap::{Parser, Subcommand};
use serde_json::json;

pub use crate::backup::BackupCommands;
pub use crate::daemon::DaemonCommands;
pub use crate::hook::HookCommands;
use crate::{RecallParams, RecordParams};

fn default_db() -> PathBuf {
    crate::paths::default_db_path()
}

#[derive(Parser)]
#[command(name = "innate", version, about = "Self-growing knowledge layer")]
pub struct Cli {
    #[arg(long, global = true, env = "INNATE_DB")]
    pub db: Option<PathBuf>,

    #[command(subcommand)]
    pub command: Commands,
}

#[derive(Subcommand)]
pub enum Commands {
    /// Search the knowledge base
    Recall {
        query: String,
        #[arg(long, default_value = "6000")]
        budget: usize,
        #[arg(long)]
        top: Option<usize>,
        #[arg(long, default_value = "text")]
        format: String,
        #[arg(long)]
        include_sparks: bool,
        /// Dependency expansion: false (default) | direct | closure
        #[arg(long, default_value = "false")]
        expand_deps: String,
        /// Allow Refiner to trim blocks that don't fit the budget
        #[arg(long)]
        allow_trim: bool,
        /// Refine mode written to usage_trace: off (default) | trim | adapt
        #[arg(long, default_value = "off")]
        refine_mode: String,
        /// Event source written to usage_trace (mcp | sdk | cli | hook | daemon | augmented)
        #[arg(long, default_value = "cli")]
        source: String,
    },
    /// Close a trace with outcome
    Record {
        trace_id: String,
        #[arg(long)]
        query: Option<String>,
        #[arg(long)]
        outcome: Option<String>,
        /// Comma-separated chunk ids. An explicit empty value means "known none".
        #[arg(long)]
        used: Option<String>,
        #[arg(long, default_value = "explicit")]
        used_attribution: String,
        /// Treat --used as partial attribution; omitted selected chunks are not penalized.
        #[arg(long)]
        used_partial: bool,
        #[arg(long)]
        output: Option<String>,
        #[arg(long)]
        output_summary: Option<String>,
        #[arg(long)]
        nomination: Option<String>,
        #[arg(long, default_value = "cli")]
        source: String,
        /// Explicit feedback: up or down (applied to --used chunks if provided)
        #[arg(long)]
        feedback: Option<String>,
        #[arg(long, default_value = "user")]
        feedback_kind: String,
        #[arg(long)]
        feedback_actor: Option<String>,
        #[arg(long)]
        feedback_reason: Option<String>,
        #[arg(long)]
        task_state: Option<String>,
        #[arg(long, default_value = "0")]
        priority: i64,
    },
    /// Add a knowledge chunk
    Add {
        content: String,
        #[arg(long, default_value = "note")]
        kind: String,
        #[arg(long)]
        trigger: Option<String>,
        #[arg(long)]
        anti_trigger: Option<String>,
        #[arg(long, default_value = "chat")]
        source: String,
        #[arg(long)]
        skill_name: Option<String>,
        /// Declare a dependency on another chunk id (repeatable).
        #[arg(long = "depends-on")]
        depends_on: Vec<String>,
        /// Dependency kind for --depends-on: hard (fail-closed) or soft (bonus).
        #[arg(long, default_value = "hard")]
        dep_kind: String,
    },
    /// Capture a spark (idea)
    Spark {
        content: String,
        #[arg(long)]
        trigger: Option<String>,
    },
    /// Distil logs + curate
    Evolve {
        #[arg(long, default_value = "manual")]
        trigger: String,
        /// Rebuild embeddings for chunks with embed_version=0 or < meta.embed_version
        #[arg(long)]
        rebuild_embeddings: bool,
    },
    /// Health check — no arg = library summary; chunk_id or trace_id = detail view
    Inspect { id: Option<String> },
    /// Approve a pending chunk
    Approve { chunk_id: String },
    /// Archive a chunk
    Archive {
        chunk_id: String,
        #[arg(long, default_value = "stale")]
        reason: String,
    },
    /// Invalidate a chunk
    Invalidate {
        chunk_id: String,
        #[arg(long, default_value = "")]
        reason: String,
    },
    /// Restore an archived chunk
    Restore { chunk_id: String },
    /// Mature a spark
    MatureSpark { spark_id: String, to: String },
    /// Promote a spark to knowledge
    PromoteSpark {
        spark_id: String,
        #[arg(long, default_value = "note")]
        to: String,
    },
    /// Drop a spark
    DropSpark {
        spark_id: String,
        #[arg(long, default_value = "")]
        reason: String,
    },
    /// Backup the database to Cloudflare R2
    Backup {
        #[command(subcommand)]
        action: BackupCommands,
    },
    /// Interactive setup wizard — configure agents to use Innate MCP server
    Install,
    /// Remove Innate from all configured agents and PATH
    Uninstall {
        /// Skip confirmation prompts
        #[arg(long, short = 'y')]
        yes: bool,
        /// Also delete knowledge data (~/.innate/). Cannot be undone.
        #[arg(long)]
        purge_data: bool,
    },
    /// Upgrade database schema to current version
    Migrate,
    /// Reclaim disk space: checkpoint the WAL and VACUUM the database
    Vacuum,
    /// Upgrade the innate binary to the latest (or specified) release
    Upgrade {
        /// Install this specific version, e.g. 0.3.0 or v0.3.0 (default: latest)
        #[arg(long, value_name = "VERSION")]
        version: Option<String>,
        /// Only report whether an upgrade is available; do not install
        #[arg(long)]
        check: bool,
    },
    /// Daemon control (Linux only)
    Daemon {
        #[command(subcommand)]
        action: DaemonCommands,
    },
    /// Start MCP stdio server
    Mcp,
    /// Start a local web UI to view and govern the knowledge base
    Web {
        /// Address to bind (localhost only by default; exposing beyond is unsafe)
        #[arg(long, default_value = "127.0.0.1")]
        bind: String,
        /// Port to listen on
        #[arg(long, default_value_t = 8788)]
        port: u16,
        /// Disable the governance auth token (NOT recommended; leaves writes unauthenticated)
        #[arg(long)]
        no_token: bool,
        /// Required to bind a non-loopback address. Exposes the knowledge base to
        /// the network; the auth token then gates reads as well as writes.
        #[arg(long)]
        allow_remote: bool,
    },
    /// Agent hook handlers (called by agent hooks; reads payload from stdin)
    Hook {
        #[command(subcommand)]
        action: HookCommands,
    },
}

pub fn run() -> anyhow::Result<()> {
    let cli = Cli::parse();
    // Create the ~/.innate subdirectory layout and migrate any legacy flat files
    // before any path is resolved.
    crate::paths::ensure_layout();
    let db_path = cli.db.unwrap_or_else(default_db);

    if let Commands::Mcp = &cli.command {
        return crate::mcp::run_server(db_path);
    }

    if let Commands::Install = &cli.command {
        return crate::install::run_install();
    }

    if let Commands::Uninstall { yes, purge_data } = &cli.command {
        return crate::install::run_uninstall(*yes, *purge_data);
    }

    if let Commands::Migrate = &cli.command {
        let applied = crate::migrate::run_migrations(&db_path)?;
        if applied.is_empty() {
            println!("already at 4.14 — nothing to do");
        } else {
            for step in &applied {
                println!("  applied: {step}");
            }
            println!("migration complete");
        }
        return Ok(());
    }

    if let Commands::Daemon { action } = &cli.command {
        return crate::daemon::run_command(action, &db_path);
    }

    if let Commands::Backup { action } = &cli.command {
        return crate::backup::run_command(action, &db_path);
    }

    if let Commands::Upgrade { version, check } = &cli.command {
        return crate::upgrade::run_upgrade(version.as_deref(), &db_path, *check);
    }

    if let Commands::Hook { action } = &cli.command {
        return crate::hook::run_command(action);
    }

    let kb = crate::open_kb(&db_path)?;

    match cli.command {
        Commands::Recall {
            query,
            budget,
            top,
            format,
            include_sparks,
            expand_deps,
            allow_trim,
            refine_mode,
            source,
        } => {
            let result = kb.recall(RecallParams {
                query: &query,
                budget,
                trace: true,
                include_sparks,
                top,
                source: &source,
                expand_deps: &expand_deps,
                allow_trim,
                refine_mode: &refine_mode,
            })?;
            match format.as_str() {
                "json" => println!(
                    "{}",
                    serde_json::to_string_pretty(&json!({
                        "trace_id": result.trace_id,
                        "knowledge": result.knowledge,
                        "sparks": result.sparks,
                        "empty": result.empty,
                    }))?
                ),
                "prompt" => {
                    for chunk in &result.knowledge {
                        let content = chunk.get("content").and_then(|v| v.as_str()).unwrap_or("");
                        println!("{content}\n---");
                    }
                    // metadata at end (§九 CLI contract)
                    println!("<!-- innate_trace_id: {} -->", result.trace_id);
                    println!(
                        "<!-- innate_selected: {} -->",
                        result
                            .knowledge
                            .iter()
                            .filter_map(|c| c.get("id").and_then(|v| v.as_str()))
                            .collect::<Vec<_>>()
                            .join(",")
                    );
                }
                _ => {
                    for chunk in &result.knowledge {
                        let id = chunk.get("id").and_then(|v| v.as_str()).unwrap_or("?");
                        let content = chunk.get("content").and_then(|v| v.as_str()).unwrap_or("");
                        let conf = chunk
                            .get("confidence")
                            .and_then(|v| v.as_f64())
                            .unwrap_or(0.5);
                        println!("[{id}] (conf={conf:.2})\n{content}\n");
                    }
                    if result.empty {
                        println!("(no results)");
                    }
                }
            }
        }
        Commands::Record {
            trace_id,
            query,
            outcome,
            used,
            used_attribution,
            used_partial,
            output,
            output_summary,
            nomination,
            source,
            feedback,
            feedback_kind,
            feedback_actor,
            feedback_reason,
            task_state,
            priority,
        } => {
            let used_ids = used.as_deref().map(|raw| {
                raw.split(',')
                    .map(str::trim)
                    .filter(|id| !id.is_empty())
                    .map(str::to_string)
                    .collect::<Vec<_>>()
            });
            let used_ref = used_ids.as_deref();
            // Per §二·五B: trace-level "up" applies only to explicitly used chunks.
            let (fb_up, fb_down): (Option<Vec<String>>, Option<Vec<String>>) =
                match feedback.as_deref() {
                    Some("up") if used_ids.as_ref().is_some_and(|ids| !ids.is_empty()) => {
                        (used_ids.clone(), None)
                    }
                    Some("down") if used_ids.as_ref().is_some_and(|ids| !ids.is_empty()) => {
                        (None, used_ids.clone())
                    }
                    Some("up") => (None, None), // no used chunks — ignore per design
                    Some("down") => (None, None),
                    _ => (None, None),
                };
            let fb_up_ref = fb_up.as_deref();
            let fb_down_ref = fb_down.as_deref();
            kb.record(RecordParams {
                trace_id: &trace_id,
                query: query.as_deref(),
                output: output.as_deref(),
                output_summary: output_summary.as_deref(),
                outcome: outcome.as_deref(),
                used: used_ref,
                used_attribution: &used_attribution,
                used_complete: Some(!used_partial),
                feedback_up: fb_up_ref,
                feedback_down: fb_down_ref,
                feedback_kind: &feedback_kind,
                feedback_actor: feedback_actor.as_deref(),
                feedback_reason: feedback_reason.as_deref(),
                nomination: nomination.as_deref(),
                priority,
                task_state: task_state.as_deref(),
                source: &source,
            })?;
            println!("recorded");
        }
        Commands::Add {
            content,
            kind,
            trigger,
            anti_trigger,
            source,
            skill_name,
            depends_on,
            dep_kind,
        } => {
            // If kind=skill and content is a readable file path, load its content.
            let content = if kind == "skill" {
                let p = std::path::Path::new(&content);
                if p.exists() && p.is_file() {
                    std::fs::read_to_string(p).map_err(|e| {
                        anyhow::anyhow!("Failed to read skill file {}: {e}", p.display())
                    })?
                } else {
                    content
                }
            } else {
                content
            };
            let deps: Vec<(String, String)> = depends_on
                .iter()
                .map(|d| (d.clone(), dep_kind.clone()))
                .collect();
            let id = kb.add_with_deps(
                &content,
                &kind,
                trigger.as_deref(),
                anti_trigger.as_deref(),
                &source,
                skill_name.as_deref(),
                &deps,
            )?;
            println!("{id}");
        }
        Commands::Spark { content, trigger } => {
            let id = kb.spark(&content, trigger.as_deref(), None)?;
            println!("{id}");
        }
        Commands::Evolve {
            trigger,
            rebuild_embeddings,
        } => {
            if rebuild_embeddings {
                let rebuilt = kb.rebuild_embeddings()?;
                let report = kb.evolve(&trigger)?;
                println!(
                    "{}",
                    serde_json::to_string_pretty(&json!({
                        "rebuilt_embeddings": rebuilt,
                        "evolve": report
                    }))?
                );
            } else {
                let report = kb.evolve(&trigger)?;
                println!("{}", serde_json::to_string_pretty(&report)?);
            }
        }
        Commands::Inspect { id } => match id.as_deref() {
            None => {
                let info = kb.inspect()?;
                println!("{}", serde_json::to_string_pretty(&info)?);
            }
            Some(id) => {
                let detail = kb.inspect_id(id)?;
                println!("{}", serde_json::to_string_pretty(&detail)?);
            }
        },
        Commands::Approve { chunk_id } => {
            kb.approve(&chunk_id)?;
            println!("approved");
        }
        Commands::Archive { chunk_id, reason } => {
            kb.archive(&chunk_id, &reason)?;
            println!("archived");
        }
        Commands::Invalidate { chunk_id, reason } => {
            kb.invalidate(&chunk_id, &reason)?;
            println!("invalidated");
        }
        Commands::Restore { chunk_id } => {
            kb.restore(&chunk_id)?;
            println!("restored");
        }
        Commands::MatureSpark { spark_id, to } => {
            kb.mature_spark(&spark_id, &to)?;
            println!("matured");
        }
        Commands::PromoteSpark { spark_id, to } => {
            let id = kb.promote_spark(&spark_id, &to)?;
            println!("{id}");
        }
        Commands::DropSpark { spark_id, reason } => {
            kb.drop_spark(&spark_id, &reason)?;
            println!("dropped");
        }
        Commands::Vacuum => {
            let (before, after) = kb.storage.vacuum()?;
            let mb = |b: i64| b as f64 / 1_048_576.0;
            println!(
                "vacuumed: {:.2} MB → {:.2} MB (reclaimed {:.2} MB)",
                mb(before),
                mb(after),
                mb(before - after)
            );
        }
        Commands::Web {
            bind,
            port,
            no_token,
            allow_remote,
        } => {
            let loopback = crate::web::is_loopback(&bind);
            if !loopback && !allow_remote {
                anyhow::bail!(
                    "refusing to bind non-loopback address {bind} without --allow-remote \
                     (this exposes the knowledge base to the network)"
                );
            }
            if !loopback && no_token {
                anyhow::bail!(
                    "--no-token cannot be combined with a non-loopback bind: a network-exposed \
                     server must keep the auth token to gate reads and writes"
                );
            }
            crate::web::serve(kb, &bind, port, !no_token)?;
        }
        Commands::Mcp
        | Commands::Install
        | Commands::Uninstall { .. }
        | Commands::Migrate
        | Commands::Upgrade { .. }
        | Commands::Daemon { .. }
        | Commands::Backup { .. }
        | Commands::Hook { .. } => unreachable!(),
    }
    Ok(())
}