indra 0.7.0

Run, watch and orchestrate AI coding agents (Claude Code, Codex, Gemini, Ollama) across machines: a web console, workflows with conditions and loops, boards, an MCP server and graph memory, in one binary.
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
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162

//! indra - one CLI for AI agent harnesses.
//!
//! Named for Indra's net: an endless net with a jewel at every knot, each
//! reflecting all the others. `indra mem` keeps that net of memories, and
//! `indra serve` is a console over the agents of a harness.

mod completions;
mod deck;
mod findings;
mod mem;
mod paths;
mod publish;
mod review;
mod site;
mod skills;
mod update;

use clap::{Parser, Subcommand};
use std::process::ExitCode;

#[derive(Parser)]
#[command(
    name = "indra",
    version,
    about = "One CLI for AI agent harnesses: graph memory and a web console to view, monitor and run agents",
    long_about = "One CLI for AI agent harnesses.\n\n\
        indra mem     graph memory: Markdown notes with typed [[links]], ranked recall, Obsidian-compatible\n\
        indra serve   web console: agents, skills, prompts, graph maps, memory, runs, launch and watch agents\n\
        indra ls      the console's picture in the terminal\n\
        indra nodes   the machines agents run on, local and over SSH\n\
        indra run     run a multi-step, multi-node workflow and follow it\n\n\
        The security harness is the default profile; any Claude Code project or custom harness works too.",
    after_help = "Start here:\n  \
        indra serve --open           open the console in your browser\n  \
        indra claude -m qwen3-coder  Claude Code on a local model (Ollama)\n  \
        indra update                 get the newest version (--auto on: by itself)\n  \
        indra completions --install  Tab completion in your shell"
)]
struct Cli {
    #[command(subcommand)]
    cmd: Cmd,
}

#[derive(Subcommand)]
enum Cmd {
    /// Graph memory: add, link, recall, tree, export... (`indra mem --help`)
    #[command(
        alias = "m",
        long_about = "Graph memory for people and AI agents.\n\n\
            Notes are plain Markdown with [[wikilinks]] and typed relations (`mitigated-by:: [[note]]`). \
            The vault is an ordinary folder that also opens in Obsidian.\n\n\
            Vault location: --vault, else $INDRA_VAULT, else the nearest parent folder holding `.indra/`, \
            else a per-user default (see `indra mem where`)."
    )]
    Mem(mem::cli::MemArgs),
    /// Start the web console (prints a URL with an access token)
    Serve(deck::cli::ServeArgs),
    /// Summarise harnesses, agents, skills, prompts and runs in the terminal
    Ls(deck::cli::LsArgs),
    /// Validate the console configuration and report what was found
    Check(deck::cli::CheckArgs),
    /// Write a commented starter indra.toml for custom harnesses and runners
    Init(deck::cli::InitArgs),
    /// Check the machines agents run on: this one and every [[node]] over SSH
    Nodes(deck::cli::NodesArgs),
    /// Run a workflow from indra.toml and follow it here (omit the name to list them)
    Run(deck::cli::RunArgs),
    /// The remote side of a node; the hub runs `indra node ...` over SSH
    Node(deck::cli::NodeArgs),
    /// Secrets for agents' profiles: set (value from stdin), list, rm
    Secret(deck::cli::SecretArgs),
    /// List the agents a running console knows about (`indra attach` steers one)
    Agents(deck::cli::AgentsArgs),
    /// Follow an agent of a running console and steer it from this terminal
    Attach(deck::cli::AttachArgs),
    /// Claude Code in this terminal on a model served by Ollama (or only its subagents there)
    Claude(deck::claude::ClaudeArgs),
    /// Publish plans for Jira, Confluence, GitHub Issues or Notion from a review run
    Publish(publish::PublishArgs),
    /// PR review support: the review cache, local metrics, posting to GitHub
    Review(review::ReviewArgs),
    /// Sync the skill mirror, or install the skills for an agent
    Skills(skills::SkillsArgs),
    /// Update indra to the newest release (checks its SHA-256 first); --auto on keeps it up to date
    Update(update::UpdateArgs),
    /// Tab completion for your shell: print the script, or --install it
    Completions(completions::CompletionsArgs),
    /// MCP server for AI agents to design workflows (`claude mcp add indra -- indra mcp`)
    Mcp(deck::cli::McpArgs),
    /// Accounts agents run with: what each used today and this month, and its limits
    Accounts(deck::cli::AccountsArgs),
    /// Console users and roles: add, list, passwd, role, rm
    User(deck::users::UserArgs),
    /// Maintainer tools: build the website
    #[command(hide = true)]
    Dev {
        #[command(subcommand)]
        cmd: DevCmd,
    },
}

#[derive(Subcommand)]
enum DevCmd {
    /// Build the website (landing page + docs) into _site
    Site(site::SiteArgs),
    /// Find em and en dashes in the repository (exit 1 if any), or --fix them
    Dashes {
        /// The repository (default: this folder)
        #[arg(long)]
        repo: Option<std::path::PathBuf>,
        #[arg(long)]
        fix: bool,
    },
}

fn main() -> ExitCode {
    let cli = Cli::parse();
    // Say what the daily release check found, and start today's. Not for
    // update and completions themselves, nor the remote side of a node
    // (its output is read by the hub).
    if !matches!(
        cli.cmd,
        Cmd::Update(_) | Cmd::Completions(_) | Cmd::Node(_) | Cmd::Mcp(_) | Cmd::Dev { .. }
    ) {
        update::on_start();
    }
    let res = match cli.cmd {
        Cmd::Mem(a) => mem::cli::run(a),
        Cmd::Serve(a) => deck::cli::run_serve(a),
        Cmd::Ls(a) => deck::cli::run_ls(a),
        Cmd::Check(a) => deck::cli::run_check(a),
        Cmd::Init(a) => deck::cli::init(a),
        Cmd::Nodes(a) => deck::cli::run_nodes(a),
        Cmd::Run(a) => deck::cli::run_workflow(a),
        Cmd::Node(a) => deck::cli::run_node(a),
        Cmd::Secret(a) => deck::cli::run_secret(a),
        Cmd::Agents(a) => deck::cli::run_agents(a),
        Cmd::Attach(a) => deck::cli::run_attach(a),
        Cmd::Claude(a) => deck::claude::run(a),
        Cmd::Update(a) => update::run(a),
        Cmd::Completions(a) => completions::run(a),
        Cmd::Mcp(a) => deck::cli::run_mcp(a),
        Cmd::Accounts(a) => deck::cli::run_accounts(a),
        Cmd::User(a) => deck::users::run(a),
        Cmd::Publish(a) => publish::run(a),
        Cmd::Review(a) => review::run(a),
        Cmd::Skills(a) => skills::run(a),
        Cmd::Dev {
            cmd: DevCmd::Site(a),
        } => site::run(a),
        Cmd::Dev {
            cmd: DevCmd::Dashes { repo, fix },
        } => site::dashes(repo, fix),
    };
    match res {
        Ok(code) => code,
        Err(e) => {
            eprintln!("indra: {}", e);
            ExitCode::FAILURE
        }
    }
}