humanbound-cli 0.6.1__tar.gz → 0.6.2__tar.gz

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: humanbound-cli
-Version: 0.6.1
+Version: 0.6.2
 Summary: Humanbound CLI - command line interface for AI agent security testing.
 Author-email: Kostas Siabanis <hello@humanbound.ai>, Demetris Gerogiannis <hello@humanbound.ai>
 License: Apache-2.0

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/humanbound_cli/commands/connect.py

@@ -43,7 +43,7 @@ def _derive_agent_name(endpoint: str) -> str:
 
 
 @click.command("connect")
-@click.option("--endpoint", "-e", help="Bot config JSON or file path (agent path)")
+@click.option("--endpoint", "-e", help="Agent config JSON or file path (agent path)")
 @click.option(
     "--vendor", "-v",
     type=click.Choice(["microsoft"]),
@@ -53,7 +53,7 @@ def _derive_agent_name(endpoint: str) -> str:
 @click.option("--prompt", "-p", type=click.Path(exists=True), help="System prompt file (agent path)")
 @click.option("--repo", "-r", type=click.Path(exists=True), help="Repository path (agent path)")
 @click.option("--openapi", "-o", type=click.Path(exists=True), help="OpenAPI spec file (agent path)")
-@click.option("--serve", "-s", is_flag=True, help="Launch repo bot locally (agent path, requires --repo)")
+@click.option("--serve", "-s", is_flag=True, help="Launch repo agent locally (agent path, requires --repo)")
 @click.option("--tenant", help="Azure tenant ID (platform path, bypasses browser)")
 @click.option("--client-id", "client_id", help="Service principal client ID (platform path)")
 @click.option("--client-secret", "client_secret", help="Service principal secret (platform path)")
@@ -69,7 +69,7 @@ def connect_command(endpoint, vendor, name, prompt, repo, openapi, serve,
     \b
     Agent path (--endpoint):
       hb connect --endpoint ./bot-config.json
-      Probes your bot, extracts scope, creates project, runs first test.
+      Probes your agent, extracts scope, creates project, runs first test.
 
     \b
     Platform path (--vendor):
@@ -97,10 +97,10 @@ def connect_command(endpoint, vendor, name, prompt, repo, openapi, serve,
     else:
         console.print("[yellow]Specify a path:[/yellow]")
         console.print()
-        console.print("  [bold]Agent:[/bold]     hb connect --endpoint ./bot-config.json")
+        console.print("  [bold]Agent:[/bold]      hb connect --endpoint ./bot-config.json")
         console.print("  [bold]Platform:[/bold]  hb connect --vendor microsoft")
         console.print()
-        console.print("[dim]Use --endpoint to connect an AI agent, or --vendor to scan your cloud.[/dim]")
+        console.print("[dim]Use --endpoint to connect your AI agent, or --vendor to scan your cloud.[/dim]")
         raise SystemExit(1)
 
 
@@ -192,7 +192,7 @@ def _connect_agent(endpoint, name, prompt, repo, openapi, serve, context, yes, t
 
             if runtime_info and not serve and not endpoint and not yes:
                 console.print()
-                console.print(f"  [cyan]i[/cyan] Detected runnable bot: [bold]{runtime_info.framework.title()}[/bold] ({runtime_info.entry_point})")
+                console.print(f"  [cyan]i[/cyan] Detected runnable agent: [bold]{runtime_info.framework.title()}[/bold] ({runtime_info.entry_point})")
                 console.print(f"    Start command: [dim]{runtime_info.start_cmd.replace('{port}', str(runtime_info.port))}[/dim]")
                 from rich.prompt import Confirm
                 if Confirm.ask("    Launch it for live probing?", default=False):
@@ -209,7 +209,7 @@ def _connect_agent(endpoint, name, prompt, repo, openapi, serve, context, yes, t
             if spec_result:
                 operations = spec_result.get("operations", [])
                 console.print(f"  [green]\u2713[/green] OpenAPI spec: {len(operations)} operations")
-                summary_parts = [spec_result.get("description", "API-based bot")]
+                summary_parts = [spec_result.get("description", "API-based agent")]
                 for op in operations:
                     summary_parts.append(
                         f"- {op.get('method', 'GET')} {op.get('path', '')}: {op.get('summary', '')}"
@@ -228,7 +228,7 @@ def _connect_agent(endpoint, name, prompt, repo, openapi, serve, context, yes, t
         # -- Serve lifecycle: start server + tunnel ----------------------------
         if serve and repo:
             if not runtime_info:
-                console.print("[yellow]Could not detect a runnable bot in the repository.[/yellow]")
+                console.print("[yellow]Could not detect a runnable agent in the repository.[/yellow]")
                 console.print("[dim]Continuing with static analysis only.[/dim]")
             else:
                 _server, _tunnel, serve_source = _start_serve(
@@ -513,7 +513,7 @@ def _connect_platform(vendor, name, tenant, client_id, client_secret, yes, timeo
 def _auto_test(client, project_id, default_integration, context=None):
     """Run first test automatically and show results inline."""
     if not default_integration:
-        console.print("\n[dim]No bot integration configured -- skipping auto-test.[/dim]")
+        console.print("\n[dim]No agent integration configured -- skipping auto-test.[/dim]")
         console.print("[dim]Run 'hb test -e ./bot-config.json' to test manually.[/dim]")
         return
 
@@ -569,5 +569,7 @@ def _auto_test(client, project_id, default_integration, context=None):
         console.print(f"  [dim]View logs:[/dim]       hb logs {exp_id}")
 
     except Exception as e:
+        import traceback
         console.print(f"\n[yellow]Auto-test failed:[/yellow] {e}")
+        console.print(f"[dim]{traceback.format_exc()}[/dim]")
         console.print("[dim]Run 'hb test' to try again.[/dim]")

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/humanbound_cli/commands/posture.py

@@ -18,7 +18,9 @@ console = Console()
 @click.option("--trends", is_flag=True, help="Show posture history over time")
 @click.option("--org", is_flag=True, help="Show org-level posture (3 dimensions)")
 @click.option("--coverage", is_flag=True, help="Include test coverage breakdown")
-def posture_command(project: str, as_json: bool, trends: bool, org: bool, coverage: bool):
+def posture_command(
+    project: str, as_json: bool, trends: bool, org: bool, coverage: bool
+):
     """View security posture score for a project.
 
     The posture score is a composite metric (0-100) reflecting:
@@ -52,10 +54,13 @@ def posture_command(project: str, as_json: bool, trends: bool, org: bool, covera
                 raise SystemExit(1)
 
             with console.status("Calculating organisation posture..."):
-                response = client.get(f"organisations/{org_id}/posture", include_project=False)
+                response = client.get(
+                    f"organisations/{org_id}/posture", include_project=False
+                )
 
             if as_json:
                 import json
+
                 print(json.dumps(response, indent=2, default=str))
                 return
 
@@ -82,6 +87,7 @@ def posture_command(project: str, as_json: bool, trends: bool, org: bool, covera
 
             if as_json:
                 import json
+
                 print(json.dumps(response, indent=2, default=str))
                 return
 
@@ -90,10 +96,13 @@ def posture_command(project: str, as_json: bool, trends: bool, org: bool, covera
 
         # Get posture from API
         with console.status("Calculating posture..."):
-            response = client.get(f"projects/{project_id}/posture", include_project=True)
+            response = client.get(
+                f"projects/{project_id}/posture", include_project=True
+            )
 
         if as_json:
             import json
+
             print(json.dumps(response, indent=2, default=str))
             return
 
@@ -125,7 +134,7 @@ def posture_command(project: str, as_json: bool, trends: bool, org: bool, covera
 
 def _display_posture(posture: dict):
     """Display posture score with visual breakdown."""
-    score = posture.get("score", 0)
+    score = posture.get("overall_score", 0)
     grade = posture.get("grade", _score_to_grade(score))
 
     # Color based on score
@@ -140,16 +149,21 @@ def _display_posture(posture: dict):
         emoji = "✗"
 
     # Main score panel
-    console.print(Panel(
-        f"[bold {score_color}]{emoji} {score}/100[/bold {score_color}]  [dim]Grade: {grade}[/dim]",
-        title="Security Posture",
-        border_style=score_color,
-        padding=(1, 4),
-    ))
+    console.print(
+        Panel(
+            f"[bold {score_color}]{emoji} {score}/100[/bold {score_color}]  [dim]Grade: {grade}[/dim]",
+            title="Security Posture",
+            border_style=score_color,
+            padding=(1, 4),
+        )
+    )
 
     # Breakdown table
-    breakdown = posture.get("breakdown", {})
-    if breakdown:
+    finding_metrics = posture.get("finding_metrics", {})
+    coverage_metrics = posture.get("coverage_metrics", {})
+    resilience_metrics = posture.get("resilience_metrics", {})
+
+    if finding_metrics or coverage_metrics or resilience_metrics:
         console.print("\n[bold]Score Breakdown:[/bold]\n")
 
         table = Table(show_header=True, header_style="bold")
@@ -159,15 +173,19 @@ def _display_posture(posture: dict):
         table.add_column("Bar", width=30)
 
         components = [
-            ("Findings", breakdown.get("finding_score", 0), "40%"),
-            ("Confidence", breakdown.get("confidence_score", 0), "25%"),
-            ("Coverage", breakdown.get("coverage_score", 0), "20%"),
-            ("Drift", breakdown.get("drift_score", 0), "15%"),
+            ("Findings", finding_metrics.get("score", 0), "40%"),
+            ("Confidence", finding_metrics.get("avg_confidence", 0), "25%"),
+            ("Coverage", coverage_metrics.get("score", 0), "20%"),
+            ("Resilience", resilience_metrics.get("score", 0), "15%"),
         ]
 
         for name, comp_score, weight in components:
             bar = _score_bar(comp_score)
-            color = "green" if comp_score >= 80 else ("yellow" if comp_score >= 60 else "red")
+            color = (
+                "green"
+                if comp_score >= 80
+                else ("yellow" if comp_score >= 60 else "red")
+            )
             table.add_row(
                 name,
                 f"[{color}]{comp_score:.0f}[/{color}]",
@@ -309,14 +327,11 @@ def _calculate_fallback_posture(client: HumanboundClient, project_id: str):
         score = min(100, pass_rate)
 
         posture = {
-            "score": score,
+            "overall_score": score,
             "grade": _score_to_grade(score),
-            "breakdown": {
-                "finding_score": pass_rate,
-                "confidence_score": 80,  # Placeholder
-                "coverage_score": 70,    # Placeholder
-                "drift_score": 85,       # Placeholder
-            },
+            "finding_metrics": {"score": pass_rate},
+            "coverage_metrics": {"score": 70},
+            "resilience_metrics": {"score": 85},
             "recommendations": [],
             "last_tested": latest.get("created_at", "")[:10],
         }
@@ -353,12 +368,14 @@ def _display_org_posture(response: dict):
         emoji = "✗"
 
     # Main score panel
-    console.print(Panel(
-        f"[bold {score_color}]{emoji} {score}/100[/bold {score_color}]  [dim]Grade: {grade}[/dim]",
-        title="Organisation Posture",
-        border_style=score_color,
-        padding=(1, 4),
-    ))
+    console.print(
+        Panel(
+            f"[bold {score_color}]{emoji} {score}/100[/bold {score_color}]  [dim]Grade: {grade}[/dim]",
+            title="Organisation Posture",
+            border_style=score_color,
+            padding=(1, 4),
+        )
+    )
 
     # Dimension breakdown
     dimensions = response.get("dimensions", {})
@@ -378,9 +395,13 @@ def _display_org_posture(response: dict):
 
         for key, label in dimension_labels.items():
             dim_data = dimensions.get(key, {})
-            dim_score = dim_data.get("score", 0) if isinstance(dim_data, dict) else dim_data
+            dim_score = (
+                dim_data.get("score", 0) if isinstance(dim_data, dict) else dim_data
+            )
             bar = _score_bar(dim_score)
-            color = "green" if dim_score >= 80 else ("yellow" if dim_score >= 60 else "red")
+            color = (
+                "green" if dim_score >= 80 else ("yellow" if dim_score >= 60 else "red")
+            )
             table.add_row(
                 label,
                 f"[{color}]{dim_score:.0f}[/{color}]",
@@ -410,7 +431,9 @@ def _display_coverage_section(response: dict):
     empty = bar_width - filled
     bar = f"[{cov_color}]{'█' * filled}[/{cov_color}][dim]{'░' * empty}[/dim]"
 
-    console.print(f"\n[bold]Test Coverage:[/bold]  {bar}  [{cov_color}]{overall:.0f}%[/{cov_color}]")
+    console.print(
+        f"\n[bold]Test Coverage:[/bold]  {bar}  [{cov_color}]{overall:.0f}%[/{cov_color}]"
+    )
 
     # Category breakdown
     categories = response.get("categories", response.get("by_category", []))
@@ -427,7 +450,9 @@ def _display_coverage_section(response: dict):
 
             if total > 0:
                 rate = (passed / total) * 100
-                rate_color = "green" if rate >= 80 else ("yellow" if rate >= 50 else "red")
+                rate_color = (
+                    "green" if rate >= 80 else ("yellow" if rate >= 50 else "red")
+                )
                 rate_str = f"[{rate_color}]{rate:.0f}%[/{rate_color}]"
             else:
                 rate_str = "[dim]-[/dim]"
@@ -441,7 +466,11 @@ def _display_coverage_section(response: dict):
     if gap_list:
         console.print(f"\n[yellow]Gaps ({len(gap_list)} untested):[/yellow]")
         for gap in gap_list[:5]:
-            name = gap.get("category", gap.get("name", str(gap))) if isinstance(gap, dict) else str(gap)
+            name = (
+                gap.get("category", gap.get("name", str(gap)))
+                if isinstance(gap, dict)
+                else str(gap)
+            )
             console.print(f"  - {name}")
         if len(gap_list) > 5:
             console.print(f"  [dim]... and {len(gap_list) - 5} more[/dim]")
@@ -457,7 +486,9 @@ def _print_next(org: bool = False, has_coverage: bool = False):
         suggestions.append("hb report --org         Generate org posture report")
     else:
         if not has_coverage:
-            suggestions.append("hb posture --coverage   Include test coverage breakdown")
+            suggestions.append(
+                "hb posture --coverage   Include test coverage breakdown"
+            )
         suggestions.append("hb posture --trends     View posture over time")
         suggestions.append("hb posture --org        View org-level posture")
         suggestions.append("hb test                 Run tests to improve posture")

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/humanbound_cli/commands/test.py

@@ -104,8 +104,8 @@ def _print_next(suggestions: list):
 )
 @click.option(
     "--endpoint", "-e",
-    help="Bot integration config — JSON string or path to JSON file. "
-         "Same shape as 'hb init --endpoint'. Overrides the project's default integration."
+    help="Agent integration config — JSON string or path to JSON file. "
+         "Same shape as 'hb connect --endpoint'. Overrides the project's default integration."
 )
 @click.option(
     "--category",

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/humanbound_cli/mcp_server.py

@@ -48,6 +48,8 @@ mcp = FastMCP(
         "the user wants a quick one-off scan or has no connector set up.\n\n"
 
         "CORE WORKFLOWS:\n"
+        "  One-shot agent onboarding (fastest):\n"
+        "    hb_connect — scan + create project + auto-test in a single call\n"
         "  Discovery → Testing:\n"
         "    hb_trigger_discovery → hb_list_inventory → hb_onboard_inventory_asset → hb_run_test\n"
         "  Security Testing:\n"
@@ -62,7 +64,7 @@ mcp = FastMCP(
 
         "CLI-ONLY COMMANDS (suggest when relevant):\n"
         "  • 'hb discover' — browser-based shadow AI discovery (no connector needed)\n"
-        "  • 'hb init' — interactive project setup with scope extraction from a URL\n"
+        "  • 'hb connect --vendor microsoft' — browser-based platform discovery\n"
         "  • 'hb sentinel setup' — configure continuous monitoring sentinel"
     ),
 )
@@ -1303,6 +1305,166 @@ def hb_break_campaign(campaign_id: str, project_id: Optional[str] = None) -> str
         return _err(e)
 
 
+# =========================================================================
+# CONNECT — one-shot agent onboarding (scan → project → test)
+# =========================================================================
+
+@mcp.tool()
+def hb_connect(
+    endpoint_config: str,
+    name: Optional[str] = None,
+    prompt_text: Optional[str] = None,
+    context: Optional[str] = None,
+    timeout: int = 180,
+) -> str:
+    """Connect an AI agent: probe it, create a project, and run the first security test — all in one call.
+
+    This is the fastest way to onboard an agent. It replicates the CLI
+    command ``hb connect --endpoint <config> --yes``.
+
+    Flow:
+      1. Build extraction sources from the endpoint config (and optional prompt).
+      2. POST /scan — the backend probes the agent and extracts its scope.
+      3. Create a project with the extracted scope.
+      4. Auto-start the first security test (owasp_agentic, unit level).
+
+    After this tool returns, poll hb_get_experiment_status with the
+    returned experiment_id until status is "Finished", then call
+    hb_get_experiment_logs and hb_get_posture for results.
+
+    Args:
+        endpoint_config: JSON string with the agent's chat-completion config.
+            Example: {"streaming": false, "chat_completion": {"endpoint": "https://...",
+            "headers": {"Authorization": "Bearer ..."}, "payload": {"content": "$PROMPT"}}}
+        name: Project name (auto-derived from endpoint hostname if omitted).
+        prompt_text: Optional system prompt text to include as an additional
+            extraction source (improves scope quality).
+        context: Extra context for the judge, e.g. "Authenticated as Alice,
+            her PII is expected" (max 1500 chars).
+        timeout: Request timeout in seconds for the scan call (default 180).
+    """
+    try:
+        client = _get_client()
+
+        if not client.is_authenticated():
+            return _err(ValueError("Not authenticated. The user must run 'hb login' first."))
+        if not client.organisation_id:
+            return _err(ValueError("No organisation selected. Use hb_set_organisation first."))
+
+        # -- Parse endpoint config ----------------------------------------
+        try:
+            bot_config = json.loads(endpoint_config)
+        except json.JSONDecodeError as e:
+            return _err(ValueError(f"Invalid JSON in endpoint_config: {e}"))
+
+        # -- Build sources array ------------------------------------------
+        sources = [{"source": "endpoint", "data": bot_config}]
+
+        if prompt_text:
+            sources.append({"source": "text", "data": {"text": prompt_text}})
+
+        # -- Derive project name ------------------------------------------
+        if not name:
+            try:
+                ep_url = bot_config.get("chat_completion", {}).get("endpoint", "")
+                if ep_url:
+                    from urllib.parse import urlparse
+                    hostname = urlparse(ep_url).hostname
+                    if hostname:
+                        name = hostname
+            except Exception:
+                pass
+            if not name:
+                name = "My Agent"
+
+        # -- POST /scan ---------------------------------------------------
+        scan_response = client.post(
+            "scan",
+            data={"sources": sources},
+            include_project=False,
+            timeout=timeout,
+        )
+
+        scope = scan_response.get("scope", {})
+        risk_profile = scan_response.get("risk_profile", {})
+        default_integration = scan_response.get("default_integration")
+
+        # -- Create project -----------------------------------------------
+        project_data = {
+            "name": name,
+            "description": f"Project created via MCP hb_connect",
+            "scope": scope,
+        }
+        if default_integration:
+            project_data["default_integration"] = default_integration
+
+        project_result = client.post("projects", data=project_data)
+        project_id = project_result.get("id")
+
+        # Auto-select the project
+        client.set_project(project_id)
+
+        # -- Auto-test ----------------------------------------------------
+        experiment_id = None
+        auto_test_error = None
+
+        if default_integration:
+            providers = client.list_providers()
+            if providers:
+                provider = next((p for p in providers if p.get("is_default")), providers[0])
+
+                configuration = {}
+                if context:
+                    if len(context) > 1500:
+                        return _err(ValueError(f"Context too long ({len(context)} chars). Maximum is 1,500."))
+                    configuration["context"] = context
+
+                import time as _time
+                experiment_data = {
+                    "name": f"connect-{_time.strftime('%Y%m%d-%H%M%S')}",
+                    "description": "Initial assessment from hb_connect (MCP)",
+                    "test_category": "humanbound/adversarial/owasp_agentic",
+                    "testing_level": "unit",
+                    "provider_id": provider.get("id"),
+                    "auto_start": True,
+                    "configuration": configuration,
+                }
+
+                try:
+                    exp_result = client.post("experiments", data=experiment_data, include_project=True)
+                    experiment_id = exp_result.get("id")
+                except Exception as e:
+                    auto_test_error = str(e)
+            else:
+                auto_test_error = "No providers configured. Add one with hb_add_provider, then run hb_run_test."
+        else:
+            auto_test_error = "No agent integration detected from scan — skipped auto-test. Run hb_run_test manually."
+
+        # -- Build response -----------------------------------------------
+        result = {
+            "project_id": project_id,
+            "project_name": name,
+            "scope": scope,
+            "risk_profile": risk_profile,
+            "has_integration": bool(default_integration),
+        }
+
+        if experiment_id:
+            result["experiment_id"] = experiment_id
+            result["next_step"] = (
+                f"Poll hb_get_experiment_status(experiment_id='{experiment_id}') "
+                "until status is 'Finished', then call hb_get_experiment_logs "
+                "and hb_get_posture for results."
+            )
+        elif auto_test_error:
+            result["auto_test_skipped"] = auto_test_error
+
+        return _ok(result)
+
+    except HumanboundError as e:
+        return _err(e)
+
+
 # =========================================================================
 # UPLOAD
 # =========================================================================

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/humanbound_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: humanbound-cli
-Version: 0.6.1
+Version: 0.6.2
 Summary: Humanbound CLI - command line interface for AI agent security testing.
 Author-email: Kostas Siabanis <hello@humanbound.ai>, Demetris Gerogiannis <hello@humanbound.ai>
 License: Apache-2.0

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/humanbound_cli.egg-info/top_level.txt

@@ -1,3 +1,4 @@
+build
 dist
 docs
 humanbound_cli

{humanbound_cli-0.6.1 → humanbound_cli-0.6.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "humanbound-cli"
-version = "0.6.1"
+version = "0.6.2"
 authors = [
   { name="Kostas Siabanis", email="hello@humanbound.ai" },
   { name="Demetris Gerogiannis", email="hello@humanbound.ai" },