basemind 0.3.0

Full AI context layer over MCP — tree-sitter code-map, document RAG (PDF/Office/HTML/email + OCR + reranker), shared agent memory, on-demand web crawl, git history + blame + per-symbol diff. 300+ languages, 8 coding-agent harnesses, content-addressed Fjall + LanceDB.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109

//! In-process CLI that gives `basemind` 1:1 parity with the MCP tool surface.
//!
//! Every tool subcommand builds the matching `*Params` struct, constructs a
//! one-shot [`crate::mcp::BasemindServer`] (no background facilities), and calls
//! the identical `#[tool]` method an MCP client would dispatch — then renders the
//! returned [`rmcp::model::CallToolResult`]. Parity is by construction: the CLI
//! runs the same tool code, not a re-implementation.
//!
//! Layout:
//! - `context.rs` — one-shot server construction.
//! - `render.rs` — JSON extraction + generic human renderer.
//! - `codemap.rs` / `git.rs` / `memory.rs` / `web.rs` / `admin.rs` — subcommand groups.

pub mod admin;
pub mod codemap;
pub mod context;
pub mod git;
pub mod memory;
pub mod render;
pub mod web;

use std::io::Write;
use std::path::Path;

use anyhow::{Context, Result};
use clap::Subcommand;
use rmcp::ErrorData as McpError;
use rmcp::model::CallToolResult;

use crate::config::DocumentsCliOverrides;

/// Tool subcommand groups dispatched through the in-process server.
#[derive(Subcommand, Debug)]
pub enum ToolCmd {
    /// Code-map queries (outline, search, references, call-graph, …).
    #[command(subcommand)]
    Query(codemap::QueryCmd),
    /// Git history / blame / diff queries.
    #[command(subcommand)]
    Git(git::GitCmd),
    /// Shared agent memory + document search (needs `--features memory,documents`).
    #[command(subcommand)]
    Memory(memory::MemoryCmd),
    /// On-demand web ingestion (needs `--features crawl`).
    #[command(subcommand)]
    Web(web::WebCmd),
    /// Aggregate telemetry into a usage summary.
    Telemetry {
        /// Aggregation window: `today` (default), `1h`, `24h`, `all`.
        #[arg(long)]
        window: Option<String>,
        /// Optional exact tool-name filter.
        #[arg(long)]
        tool: Option<String>,
    },
}

/// Map a tool `Result<CallToolResult, McpError>` into an `anyhow::Result`,
/// surfacing the tool's own error message verbatim. Tools that return an
/// `is_error` result still produce `Ok(...)` — the JSON payload describes the
/// condition, so we render it rather than fail the process.
pub fn run_tool(tool: &str, result: Result<CallToolResult, McpError>) -> Result<CallToolResult> {
    result.map_err(|e| anyhow::anyhow!("{tool}: {e}"))
}

/// Dispatch a tool subcommand group. Builds one one-shot server per invocation
/// (reused across the single call) and discovers the git repo + config the same
/// way `serve` does.
///
/// `cache` commands are dispatched separately by the caller via
/// [`admin::run_cache`] because they are the offline path and need no server.
pub fn run(
    root: &Path,
    view: &str,
    documents: DocumentsCliOverrides,
    json: bool,
    cmd: ToolCmd,
) -> Result<()> {
    let runtime = tokio::runtime::Builder::new_multi_thread()
        .enable_all()
        .build()
        .context("build tokio runtime")?;

    runtime.block_on(async move {
        let server = context::build_server(root, view, documents)?;
        let stdout = std::io::stdout();
        let mut out = stdout.lock();
        match cmd {
            ToolCmd::Query(q) => codemap::run(&server, q, json, &mut out).await?,
            ToolCmd::Git(g) => git::run(&server, g, json, &mut out).await?,
            ToolCmd::Memory(m) => memory::run(&server, m, json, &mut out).await?,
            ToolCmd::Web(w) => web::run(&server, w, json, &mut out).await?,
            ToolCmd::Telemetry { window, tool } => {
                admin::run_telemetry(&server, window, tool, json, &mut out).await?
            }
        }
        out.flush().context("flush stdout")?;
        Ok(())
    })
}

/// Dispatch the offline `cache` command group (no server / flock needed).
pub fn run_cache(root: &Path, cmd: admin::CacheCmd, json: bool) -> Result<()> {
    let stdout = std::io::stdout();
    let mut out = stdout.lock();
    admin::run_cache(root, cmd, json, &mut out)?;
    out.flush().context("flush stdout")?;
    Ok(())
}