⚡ Sesi Command Line Interface Reference

Sesi’s CLI is a powerful, zero-footprint environment orchestrator. It allows you to run robust scripts, evaluate inline code filelessly, query the RAG-powered interactive Co-Pilot, encrypt files, and manage sandboxing controls directly from your shell.

---

🗺️ CLI Command Reference Matrix

---

🚀 1. Parametric Script Execution

You can feed command-line arguments to Sesi scripts on launch. Sesi parses these inputs and binds them as standard string values in a globally accessible array called args.

Usage

sesi main/test_args.sesi "hello" "world"

Accessing Arguments in Sesi Scripts

// Get the number of inputs
let total = len(args)

// Access indices
if total > 0 {
  print "First argument:" args[0]
}

---

🔮 2. Fileless Inline Execution (-e / --eval)

You do not need to create or save .sesi files on your disk to use Sesi. Using the -e flag, you can execute logic on the fly.

A. Statement Termination Rules

Sesi requires statements to be terminated by either a newline or a semicolon (;).

#### Single-line Semicolon Pattern

If writing on a single terminal line, separate statements manually with ;:

sesi -e "let x = 10; let y = 20; print x + y"

#### Multiline Shell Pattern

If writing a multiline string in your shell, omit semicolons completely:

sesi -e "
let x = 10
let y = 20
print x + y
"

B. Parametric Fileless Execution

You can pass CLI arguments to fileless inline scripts too! Any positional arguments coming after the inline code string are loaded into the global args array.

#### Example: Dynamic File Character Counter

sesi -e "let path = args[0]; let content = read_file(path); print 'Character count:' len(content)" "package.json"

---

🎓 3. Interactive Co-Pilot Help (-h / --help)

Sesi’s CLI features an inline, dynamic RAG (Retrieval-Augmented Generation) co-pilot trained directly on Sesi's grammar rules, standard library, and architecture documentation.

A. General Knowledge Queries (Fileless)

Ask Sesi how to write loops, parse JSON, what functions are available, what are the best practices in Sesi, etc:

sesi -h "how do I use standard json module?"

B. Context-Grounded Code Debugging

Pass a script file along with -h to chat with the co-pilot about your code. The co-pilot will automatically ingest the file contents as grounding context:

sesi examples/main/01_hello.sesi -h "how can i improve this script?"

---

⚙️ 4. Security & Sandbox Overrides

Sesi runs in a safe-by-default, zero-trust sandbox. System shell commands (exec, spawn) are blocked by default. Use these CLI flags to configure safety parameters:

A. Disable Sandboxing (-l / --local)

Unlocks the execution of raw system commands and allows directory whitelist escapes:

sesi main/script.sesi --local

B. Directory Whitelisting (-a / --allowed-paths)

Adds custom directories to the safe path whitelist (Current Working Directory and Script Directory are whitelisted by default):

sesi main/script.sesi -a "./data,./logs"

---

📦 5. Systems Utilities

B. AST Visualization (--ast)

Renders a pretty-printed, indented hierarchical tree structure of the Abstract Syntax Tree (AST):

sesi --ast examples/main/01_hello.sesi

C. Token Stream Visualization (--tokens)

Outputs a structured, formatted table showing all lexical tokens scanned by the lexer, detailing their Line, Column, Type, and raw Lexeme value:

sesi --tokens examples/main/01_hello.sesi

D. AES-256 Script Encryption (-enc & -dec)

Secure your proprietary reasoning instructions or pipelines using password-based encryption:

# Encrypt manually with password parameter
sesi -enc my_private_script.sesi -p "my-super-secret-password"

# Decrypt manually with password parameter
sesi -dec my_private_script.sesi -p "my-super-secret-password"

#### 🔒 Zero-Configuration Environment Fallback (SESI_PASSWORD)

To avoid exposing passwords in your shell's history, you can set the SESI_PASSWORD environment variable in your .env file (or your system's shell environment).

When Sesi detects SESI_PASSWORD, you can omit the -p parameter entirely:

# Encrypt automatically using SESI_PASSWORD from env
sesi -enc my_private_script.sesi

# Decrypt automatically using SESI_PASSWORD from env
sesi -dec my_private_script.sesi

💬 6. Interactive REPL (--repl or no arguments)

Sesi includes a full interactive Read-Eval-Print Loop (REPL). It keeps your variables and scopes persistent across statement entries.

Usage

Start the REPL in your shell:

sesi --repl
# or run `sesi` with no arguments in a terminal TTY
sesi

Inside the REPL, any Expression Statement (like mathematical equations, string variables, or expressions) will auto-print its evaluated result to the terminal.

Sesi Interactive REPL (v1.5.9)
Type ".exit" or press Ctrl+C to exit.
sesi> let x = 10
sesi> let y = 20
sesi> x + y
30
sesi> .exit

---

📥 7. Interactive Console Input (input)

Sesi scripts can pause execution and prompt the user for interactive console inputs step-by-step using the built-in input() function.

Example

let name = input("What is your name? ")
let age = input("What is your age? ")

print "User info:" name "is" age "years old"

When executing this script in the terminal, Sesi will print the prompts sequentially, waiting for you to type your response and press Enter before proceeding to the next statement.

---

🛠️ 8. Package.json Script Shortcuts

If you are working inside the Sesi repository, you can leverage native package manager shortcuts defined in package.json to execute, parse, or encrypt scripts:

Usage Example

# Audit the entire workspace and save report to lint_report.md
npm run lint

# Audit a single file and print Errors & Warnings directly to stdout
npm run lint bad-syntax-file.sesi

# Evaluate inline code
npm run sesi:eval "print 'Sesi running via npm!'"

# Encrypt a private script using the .env password
npm run sesi:encrypt "my-pipeline.sesi"