.png){kind=icon}
01. Autonomous Engineering Loops
Scout plans, executes, tests, and verifies multi-file projects from scratch or refactors existing modules through high-agency ReAct loops.
FrontTerrain CLI v7.0.7
16 Autonomous ReAct Tools
The Autonomous AI Principal Engineer & Codebase Co-Pilot.
FrontTerrain (scout) executes autonomous multi-step engineering loops, builds complete projects from scratch, edits & refactors codebases, controls external apps, self-heals compiler errors, and provides interactive agent sessions in your terminal.
$
npm install -g ft-scout@latest
scout agent "Build a responsive web application"
✦ Scout Agent Session Initialized (v7.0.7)
ℹ Indexing workspace structure & git telemetry...
> Executing ReAct loop [Turn 1/12]
⚙ [Tool] list_dir path="src/"
⚙ [Tool] write_file path="src/app.ts"
⚙ [Tool] run_command cmd="npm test"
✔ Compiler self-healing passed (0 errors)
★ Goal completed successfully. Rollback backup saved.
Core Capabilities
High-agency engineering built for high-velocity teams.
FT-Scout v7.0.7 combines autonomous ReAct tool execution with deep codebase due-diligence and terminal productivity.
02. App Control & Desktop Automation
Seamlessly launch, focus, and navigate external applications, web browsers, terminals, and code editors with built-in desktop control tools.
03. Automated Verification & Self-Healing
Validates TypeScript, JS, Python, Rust, Go, JSON, and HTML. Auto-diagnoses build errors and applies fixes before completing goals.
04. Atomic Multi-Snippet Edits
Batch apply non-contiguous code replacements in a single turn with multi_edit_file and zero-risk 1-command rollback (scout undo).
05. Codebase Due-Diligence
Generate multilingual onboarding briefs (scout brief), scan fragile hotspots (scout risky), and run security/dependency audits (scout audit).
06. Live Web & Documentation Search
Fetch up-to-date web context, API specs, and technical documentation live inside your terminal execution loop (scout search / scout web).
CLI Command Reference
Complete Command Suite
Commands available via scout or ft in your terminal workspace.
Scout AI Agent Commands
$ scout agent "Build a responsive snake game in HTML and JS"
# Launch autonomous multi-step execution loop for complex tasks
$ scout task "Fix typescript build errors in src/utils" -v "npm run build"
# Execute task with explicit automated verification command
$ scout goal "Refactor database query logic" -y
# Autonomous goal runner with auto-approval mode
$ scout agent
# Start persistent interactive Scout Agent session with slash commands
$ scout agent --create src/utils/math.ts "Create math helper functions"
# Generate and create new source code files from prompt
$ scout agent --run npm test
# Run terminal command with AI error diagnostics & self-healing
$ scout agent --scan / scout agent --plan "Add rate limiting"
# Perform bug scan or generate feasibility roadmap
$ scout undo
# Revert Scout's last action & restore backed-up file contents
ReAct Tool Engine
16 Autonomous ReAct Tools
Built-in capabilities available to Scout during multi-step goal execution.
read_file
Read full content & specific line ranges of workspace files.
write_file
Create new source files or overwrite existing workspace files.
edit_file
Search and replace exact target code snippets with precision.
multi_edit_file
Apply batch snippet replacements in a single atomic tool call.
run_command
Execute terminal commands (npm test, tsc, pytest).
list_dir
List contents and structure of workspace directories.
tree_view
Generate visual ASCII directory tree structures for paths.
file_info
Get file metadata, exact line counts, size & mod timestamps.
grep_search
String and regex pattern search across entire codebase.
glob_search
Filter workspace files matching specific file glob patterns.
fetch_url
Fetch web API documentation, specs, or external URLs.
git_diff
Query uncommitted workspace git diff inside execution loop.
ask_user
Request missing secret credentials or API keys interactively.
open_app
Launch or focus external applications (browser, terminal, editor).
app_action
Perform interactive actions & navigation inside opened apps.
task_completed
Conclude goal execution with structured teardown report.
Post-Execution Review Menu
Total agency with zero loss of control.
When Scout completes a goal, you are presented with an interactive post-execution menu: Keep Changes, Run Verification Tests, Create Pull Request, Export Session Report, or Revert Changes.