haliosai-cli 2.0.4__tar.gz → 2.0.6__tar.gz

{haliosai_cli-2.0.4/haliosai_cli.egg-info → haliosai_cli-2.0.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: haliosai-cli
-Version: 2.0.4
+Version: 2.0.6
 Summary: Halios CLI for coding-agent evaluations, guardrails, and OpenTelemetry evidence
 Author-email: HaliosLabs <support@halios.ai>
 License-Expression: Apache-2.0
@@ -65,10 +65,13 @@ npx skills add HaliosAI/halios --skill halios
 Prompt your agent to set up evaluations for your project:
 
 ```text
-"Set up evals for this agent: inspect the repository, create realistic test scenarios and checks, and run a baseline."
+"Set up evals for this agent: inspect the repository, create realistic test scenarios and checks, run one smoke test, and summarize before asking whether to run the full suite."
 ```
 
-Your agent will inspect your application entrypoint, configure standard OpenTelemetry, draft scenarios in `.halios/`, and run baseline evaluations.
+Your agent will inspect your application entrypoint, configure standard OpenTelemetry, draft
+scenarios in `.halios/`, and run one smoke scenario. It stops with a verification summary before
+asking whether to run the full suite and how many trials per scenario to use. The summary includes
+direct Halios UI links for reviewing the materialized suite, smoke run, and trace evidence.
 
 ---
 
@@ -78,15 +81,19 @@ If you prefer to drive evaluations directly from the command line or CI:
 
 ```bash
 # Recommended: Install with uv tool
-uv tool install 'haliosai-cli>=2.0.0'
+uv tool install 'haliosai-cli>=2.0.6'
 
 # Or install with pipx
-pipx install 'haliosai-cli>=2.0.0'
+pipx install 'haliosai-cli>=2.0.6'
 
 # See available commands and usage
 halios --help
 ```
 
+The API and UI use the same origin by default. For a self-hosted deployment with separate origins,
+set `HALIOS_UI_URL` so structured review links point to the web application without changing the
+API endpoint used by `HALIOS_BASE_URL`.
+
 ---
 
 ## How It Works

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/README.md

@@ -28,10 +28,13 @@ npx skills add HaliosAI/halios --skill halios
 Prompt your agent to set up evaluations for your project:
 
 ```text
-"Set up evals for this agent: inspect the repository, create realistic test scenarios and checks, and run a baseline."
+"Set up evals for this agent: inspect the repository, create realistic test scenarios and checks, run one smoke test, and summarize before asking whether to run the full suite."
 ```
 
-Your agent will inspect your application entrypoint, configure standard OpenTelemetry, draft scenarios in `.halios/`, and run baseline evaluations.
+Your agent will inspect your application entrypoint, configure standard OpenTelemetry, draft
+scenarios in `.halios/`, and run one smoke scenario. It stops with a verification summary before
+asking whether to run the full suite and how many trials per scenario to use. The summary includes
+direct Halios UI links for reviewing the materialized suite, smoke run, and trace evidence.
 
 ---
 
@@ -41,15 +44,19 @@ If you prefer to drive evaluations directly from the command line or CI:
 
 ```bash
 # Recommended: Install with uv tool
-uv tool install 'haliosai-cli>=2.0.0'
+uv tool install 'haliosai-cli>=2.0.6'
 
 # Or install with pipx
-pipx install 'haliosai-cli>=2.0.0'
+pipx install 'haliosai-cli>=2.0.6'
 
 # See available commands and usage
 halios --help
 ```
 
+The API and UI use the same origin by default. For a self-hosted deployment with separate origins,
+set `HALIOS_UI_URL` so structured review links point to the web application without changing the
+API endpoint used by `HALIOS_BASE_URL`.
+
 ---
 
 ## How It Works

haliosai_cli-2.0.6/halios_cli/_version.py

@@ -0,0 +1 @@
+__version__ = "2.0.6"

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/halios_cli/cli_eval.py

@@ -24,8 +24,10 @@ from jsonschema import Draft202012Validator
 
 from .cli_support import (
     ApiClient,
+    emit_review_links,
     evaluation_suite_digest,
     git_provenance,
+    halios_ui_links,
     load_project_config,
     load_yaml,
     resolve_credentials,
@@ -354,8 +356,8 @@ def _eval_quality_gaps(eval_plan: dict[str, Any], scenarios: list[dict[str, Any]
         ):
             gaps.append(
                 f"Scenario '{scenario_id}': simulator_context is empty for a multi-turn scenario; "
-                "populate private user facts/preferences/state in simulator_context so the simulator "
-                "can reveal them dynamically upon agent request"
+                "populate private user facts/preferences/state in simulator_context so the "
+                "simulator can reveal them dynamically upon agent request"
             )
     return list(dict.fromkeys(gaps))
 
@@ -801,9 +803,28 @@ def _raise_for_failed_run(report: dict[str, Any], run_id: str) -> None:
         message = str(error.get("message") or trial.get("outcome") or trial.get("state"))
         details.append(f"{trial.get('scenario_id') or trial.get('id')}: {message}")
     suffix = f" ({'; '.join(details)})" if details else ""
+    run_url = str((report.get("links") or {}).get("evaluation_run") or "")
+    review = f" Review in Halios: {run_url}." if run_url else ""
     raise typer.BadParameter(
         f"Evaluation run {run_id} failed{suffix}. "
-        f"Inspect `halios eval report {run_id} --failures --json`."
+        f"Inspect `halios eval report {run_id} --failures --json`.{review}"
+    )
+
+
+def _representative_trace_id(report: dict[str, Any]) -> str | None:
+    for trial in report.get("trials") or []:
+        if isinstance(trial, dict) and trial.get("trace_id"):
+            return str(trial["trace_id"])
+    return None
+
+
+def _evaluation_links(*, base_url: str, agent_id: str, report: dict[str, Any]) -> dict[str, str]:
+    return halios_ui_links(
+        base_url,
+        agent_id,
+        include_evaluations=True,
+        run_tag=str(report.get("run_tag") or "") or None,
+        trace_id=_representative_trace_id(report),
     )
 
 
@@ -1028,6 +1049,9 @@ def run(
                 f"Evaluation run {run_id} did not finish within {timeout}s. "
                 f"Inspect `halios eval report {run_id} --failures --json`."
             )
+        report["links"] = _evaluation_links(
+            base_url=credentials.ui_base_url, agent_id=agent_id, report=report
+        )
         _raise_for_failed_run(report, run_id)
         if expected_roots:
             report["telemetry_verification"] = _verify_simulation_telemetry(
@@ -1047,6 +1071,7 @@ def run(
         if telemetry.get("verified"):
             typer.echo(f"Telemetry: verified ({telemetry['trace_count']} traces)")
         typer.echo(f"Run {run_id}: pass@{k}={pass_at_k:.1%} gate={gate}")
+        emit_review_links(report["links"])
     if not report.get("gate_passed"):
         raise typer.Exit(2)
 
@@ -1072,6 +1097,9 @@ def report(
             **result,
             "trials": [item for item in result.get("trials", []) if not item.get("passed")],
         }
+    result["links"] = _evaluation_links(
+        base_url=credentials.ui_base_url, agent_id=agent_id, report=result
+    )
     if json_output:
         typer.echo(json.dumps(result, indent=2, sort_keys=True))
     else:
@@ -1086,3 +1114,4 @@ def report(
             f"trial_failures={int(result.get('evaluation_failed_count') or 0)} "
             f"revision={result.get('report_revision')}{delta}"
         )
+        emit_review_links(result["links"])

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/halios_cli/cli_optimize.py

@@ -8,7 +8,14 @@ from typing import Any
 
 import typer
 
-from .cli_support import ApiClient, atomic_write_text, load_project_config, resolve_credentials
+from .cli_support import (
+    ApiClient,
+    atomic_write_text,
+    emit_review_links,
+    halios_ui_links,
+    load_project_config,
+    resolve_credentials,
+)
 
 app = typer.Typer(
     help="Guide, record, and verify coding-agent prompt optimization.",
@@ -27,7 +34,16 @@ def _emit(value: dict[str, Any], json_output: bool) -> None:
     if json_output:
         typer.echo(json.dumps(value, indent=2, sort_keys=True, default=str))
     else:
-        typer.echo(json.dumps(value, indent=2, default=str))
+        links = value.get("links")
+        typer.echo(
+            json.dumps(
+                {key: item for key, item in value.items() if key != "links"},
+                indent=2,
+                default=str,
+            )
+        )
+        if isinstance(links, dict):
+            emit_review_links(links)
 
 
 def _resolve_baseline(api: ApiClient, agent_id: str, explicit_run_id: str | None) -> dict[str, Any]:
@@ -149,6 +165,12 @@ def start(
             f"`halios optimize record {run_id} --evaluation-run <run-id> "
             f"--prompt-file {prompt_file}`."
         ),
+        "links": halios_ui_links(
+            credentials.ui_base_url,
+            agent_id,
+            run_tag=str(baseline.get("run_tag") or "") or None,
+            optimization_run_id=run_id,
+        ),
     }
     _emit(result, json_output)
 
@@ -156,9 +178,10 @@ def start(
 @app.command("guidance")
 def guidance(run_id: str, json_output: bool = typer.Option(False, "--json")) -> None:
     """Return the next bounded edit contract and negative memory for a coding agent."""
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         result = api.request("POST", f"/api/v1/optimization-runs/{run_id}/next-action")
+    result["links"] = halios_ui_links(credentials.ui_base_url, agent_id, optimization_run_id=run_id)
     _emit(result, json_output)
 
 
@@ -170,7 +193,7 @@ def record(
     json_output: bool = typer.Option(False, "--json"),
 ) -> None:
     """Record one candidate using evidence from the unchanged canonical eval suite."""
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     prompt_after = prompt_file.read_text(encoding="utf-8")
     with ApiClient(credentials) as api:
         run = api.request("GET", f"/api/v1/optimization-runs/{run_id}")
@@ -229,6 +252,12 @@ def record(
             if accepted
             else "Revert the rejected prompt edit, inspect next_action, and try one different edit."
         ),
+        "links": halios_ui_links(
+            credentials.ui_base_url,
+            agent_id,
+            run_tag=str(report.get("run_tag") or "") or None,
+            optimization_run_id=run_id,
+        ),
     }
     _emit(result, json_output)
     if not accepted:
@@ -242,7 +271,7 @@ def apply_candidate(
     json_output: bool = typer.Option(False, "--json"),
 ) -> None:
     """Retrieve one backend-approved prompt candidate for repository application."""
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         handoff = api.request(
             "POST",
@@ -257,6 +286,11 @@ def apply_candidate(
         "Run `halios eval run --json` after applying the prompt, then use "
         f"`halios optimize verify {handoff['optimization_run_id']} --evaluation-run <run-id>`."
     )
+    handoff["links"] = halios_ui_links(
+        credentials.ui_base_url,
+        agent_id,
+        optimization_run_id=str(handoff["optimization_run_id"]),
+    )
     if json_output:
         _emit(handoff, True)
     elif output:
@@ -272,13 +306,14 @@ def verify_candidate(
     json_output: bool = typer.Option(False, "--json"),
 ) -> None:
     """Verify the applied candidate against the frozen baseline and unchanged suite."""
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         result = api.request(
             "POST",
             f"/api/v1/optimization-runs/{run_id}/verify",
             json={"evaluation_run_id": evaluation_run_id},
         )
+    result["links"] = halios_ui_links(credentials.ui_base_url, agent_id, optimization_run_id=run_id)
     _emit(result, json_output)
     if not result.get("passed"):
         raise typer.Exit(2)
@@ -294,15 +329,17 @@ def list_runs(json_output: bool = typer.Option(False, "--json")) -> None:
 
 @app.command("status")
 def status(run_id: str, json_output: bool = typer.Option(False, "--json")) -> None:
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         result = api.request("GET", f"/api/v1/optimization-runs/{run_id}")
+    result["links"] = halios_ui_links(credentials.ui_base_url, agent_id, optimization_run_id=run_id)
     _emit(result, json_output)
 
 
 @app.command("cancel")
 def cancel(run_id: str, json_output: bool = typer.Option(False, "--json")) -> None:
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         result = api.request("POST", f"/api/v1/optimization-runs/{run_id}/cancel")
+    result["links"] = halios_ui_links(credentials.ui_base_url, agent_id, optimization_run_id=run_id)
     _emit(result, json_output)

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/halios_cli/cli_project.py

@@ -17,8 +17,10 @@ import typer
 from .cli_support import (
     ApiClient,
     ApiError,
+    emit_review_links,
     evaluation_suite_digest,
     git_provenance,
+    halios_ui_links,
     load_project_config,
     load_yaml,
     preserve_suite_recovery,
@@ -92,6 +94,7 @@ def init(
     command: str = typer.Option(
         "", "--command", help="Project adapter command; may be added later by the coding agent."
     ),
+    json_output: bool = typer.Option(False, "--json"),
 ) -> None:
     """Create a fresh agent, or explicitly link one by UUID, and initialize the checkout."""
     root = pathlib.Path.cwd().resolve()
@@ -167,16 +170,37 @@ def init(
         created.append(".halios/scenarios.yml")
 
     verb = "Created" if created_agent else "Linked"
-    typer.echo(f"{verb} Halios agent: {resolved_agent['name']}")
-    typer.echo(f"Agent ID: {resolved_agent['id']}")
-    typer.echo(f"Agent URL: {credentials.base_url}/agents/{resolved_agent['id']}")
-    if not created_agent:
+    links = halios_ui_links(credentials.ui_base_url, str(resolved_agent["id"]))
+    result = {
+        "created": created_agent,
+        "agent": {
+            "id": str(resolved_agent["id"]),
+            "name": str(resolved_agent["name"]),
+            "slug": str(resolved_agent["slug"]),
+        },
+        "suite": {
+            "revision": int(suite.get("revision") or 0),
+            "check_count": len((suite.get("eval") or {}).get("checks") or []),
+            "scenario_count": len((suite.get("scenarios") or {}).get("scenarios") or []),
+        },
+        "created_files": created,
+        "links": links,
+    }
+    if json_output:
+        typer.echo(json.dumps(result, indent=2, sort_keys=True))
+    else:
+        typer.echo(f"{verb} Halios agent: {resolved_agent['name']}")
+        typer.echo(f"Agent ID: {resolved_agent['id']}")
+        if not created_agent:
+            typer.echo(
+                f"Existing evaluation suite: revision {suite['revision']}, "
+                f"{result['suite']['check_count']} checks, "
+                f"{result['suite']['scenario_count']} scenarios"
+            )
         typer.echo(
-            f"Existing evaluation suite: revision {suite['revision']}, "
-            f"{len((suite.get('eval') or {}).get('checks') or [])} checks, "
-            f"{len((suite.get('scenarios') or {}).get('scenarios') or [])} scenarios"
+            "Created: " + ", ".join(created) if created else "Project was already initialized."
         )
-    typer.echo("Created: " + ", ".join(created) if created else "Project was already initialized.")
+        emit_review_links(links)
 
 
 def _apply_suite_response(root: pathlib.Path, response: dict[str, Any]) -> None:
@@ -244,6 +268,7 @@ def configure(json_output: bool = typer.Option(False, "--json")) -> None:
         "revision": response["revision"],
         "digest": response.get("digest"),
         "verification": verification,
+        "links": halios_ui_links(credentials.ui_base_url, agent_id, include_suite=True),
     }
     if json_output:
         typer.echo(json.dumps(result, indent=2, sort_keys=True))
@@ -253,6 +278,7 @@ def configure(json_output: bool = typer.Option(False, "--json")) -> None:
             f"{verification['check_count']} checks, {verification['rule_count']} rules, "
             f"{verification['rubric_count']} rubrics, {verification['scenario_count']} scenarios"
         )
+        emit_review_links(result["links"])
 
 
 @app.command("refresh")
@@ -261,10 +287,12 @@ def refresh() -> None:
     root, config = load_project_config()
     agent_id = str((config.get("agent") or {}).get("id") or "")
     profile = str(config.get("profile") or "default")
-    with ApiClient(resolve_credentials(profile, agent_id)) as api:
+    credentials = resolve_credentials(profile, agent_id)
+    with ApiClient(credentials) as api:
         response = api.request("GET", f"/api/v1/agents/{agent_id}/evaluation-suite")
     _apply_suite_response(root, response)
     typer.echo(f"Refreshed evaluation suite revision {response['revision']} from Halios")
+    emit_review_links(halios_ui_links(credentials.ui_base_url, agent_id, include_suite=True))
 
 
 @app.command("check")
@@ -345,6 +373,7 @@ def check(
         "git": {"branch": branch, "commit": commit, "dirty": dirty == "true"},
         "evaluation_ai": evaluation_status,
         "suite": {"verified": True, "revision": suite["revision"]},
+        "links": halios_ui_links(credentials.ui_base_url, agent_id, include_suite=True),
     }
     if json_output:
         typer.echo(json.dumps(result, indent=2, sort_keys=True))
@@ -355,6 +384,7 @@ def check(
         typer.echo(f"Git: {branch}@{commit} dirty={dirty}")
         typer.echo(f"Evaluation AI: {evaluation_status}")
         typer.echo(f"Evaluation suite: verified (revision {suite['revision']})")
+        emit_review_links(result["links"])
 
 
 @app.command("instrumentation")

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/halios_cli/cli_support.py

@@ -10,6 +10,7 @@ import platform
 import re
 import subprocess
 import tempfile
+import urllib.parse
 from dataclasses import dataclass
 from datetime import datetime, timezone
 from typing import Any
@@ -31,6 +32,77 @@ def normalize_url(value: str) -> str:
     return (value if "://" in value else f"http://{value}").rstrip("/")
 
 
+def _ui_origin(base_url: str) -> str:
+    """Return a credential-free origin suitable for links shown to users."""
+    parsed = urllib.parse.urlsplit(normalize_url(base_url))
+    hostname = parsed.hostname
+    if parsed.scheme not in {"http", "https"} or not hostname:
+        raise typer.BadParameter(f"Invalid Halios base URL: {base_url}")
+    rendered_host = f"[{hostname}]" if ":" in hostname else hostname
+    if parsed.port is not None:
+        rendered_host = f"{rendered_host}:{parsed.port}"
+    return urllib.parse.urlunsplit((parsed.scheme, rendered_host, "", "", ""))
+
+
+def halios_ui_links(
+    base_url: str,
+    agent_id: str,
+    *,
+    include_suite: bool = False,
+    include_evaluations: bool = False,
+    run_tag: str | None = None,
+    trace_id: str | None = None,
+    optimization_run_id: str | None = None,
+) -> dict[str, str]:
+    """Build stable, secret-free Halios UI review links for durable resources."""
+    if not agent_id:
+        return {}
+    origin = _ui_origin(base_url)
+    encoded_agent = urllib.parse.quote(agent_id, safe="")
+    agent_path = f"{origin}/agents/{encoded_agent}"
+    links = {"agent": agent_path}
+    if include_suite:
+        links["scenarios"] = (
+            f"{agent_path}/evaluations?{urllib.parse.urlencode({'tab': 'scenarios'})}"
+        )
+        links["rules"] = f"{agent_path}/evaluations?{urllib.parse.urlencode({'tab': 'rules'})}"
+    if include_evaluations or run_tag:
+        links["evaluations"] = f"{agent_path}/evaluations"
+    if run_tag:
+        encoded_run_tag = urllib.parse.quote(run_tag, safe="")
+        links["evaluation_run"] = f"{agent_path}/evaluations/{encoded_run_tag}"
+        links["evaluation_traces"] = (
+            f"{agent_path}/traces?{urllib.parse.urlencode({'eval_run': run_tag})}"
+        )
+    if trace_id:
+        links["trace"] = f"{agent_path}/traces/{urllib.parse.quote(trace_id, safe='')}"
+    if optimization_run_id:
+        encoded_run_id = urllib.parse.quote(optimization_run_id, safe="")
+        links["optimization_run"] = f"{agent_path}/optimize/{encoded_run_id}"
+    return links
+
+
+_LINK_LABELS = {
+    "agent": "Agent overview",
+    "scenarios": "Scenarios",
+    "rules": "Rules and rubrics",
+    "evaluations": "Evaluation runs",
+    "evaluation_run": "Evaluation run",
+    "evaluation_traces": "Run traces",
+    "trace": "Trace evidence",
+    "optimization_run": "Optimization run",
+}
+
+
+def emit_review_links(links: dict[str, str]) -> None:
+    """Print a compact human-review handoff without making UI usage mandatory."""
+    if not links:
+        return
+    typer.echo("Review in Halios:")
+    for key, url in links.items():
+        typer.echo(f"- {_LINK_LABELS.get(key, key.replace('_', ' ').title())}: {url}")
+
+
 def credentials_path() -> pathlib.Path:
     if os.getenv("HALIOS_CONFIG_HOME"):
         root = pathlib.Path(os.environ["HALIOS_CONFIG_HOME"])
@@ -77,6 +149,7 @@ def save_profile(
     *,
     base_url: str,
     api_key: str,
+    ui_base_url: str | None = None,
     organization_id: str | None = None,
     api_key_id: int | None = None,
     expires_at: str | None = None,
@@ -87,6 +160,7 @@ def save_profile(
     profiles[profile] = {
         **previous,
         "base_url": normalize_url(base_url),
+        "ui_base_url": normalize_url(ui_base_url or base_url),
         "api_key": api_key,
         "organization_id": organization_id,
         "api_key_id": api_key_id,
@@ -117,6 +191,7 @@ def save_agent_ingest_token(profile: str, agent_id: str, token: str) -> None:
 class Credentials:
     profile: str
     base_url: str
+    ui_base_url: str
     api_key: str
     organization_id: str | None
     otlp_token: str | None = None
@@ -133,6 +208,9 @@ def stored_profile_credentials(profile: str = "default") -> Credentials | None:
     return Credentials(
         profile=profile,
         base_url=normalize_url(str(entry.get("base_url") or DEFAULT_BASE_URL)),
+        ui_base_url=normalize_url(
+            str(entry.get("ui_base_url") or entry.get("base_url") or DEFAULT_BASE_URL)
+        ),
         api_key=str(entry["api_key"]),
         organization_id=entry.get("organization_id"),
         api_key_id=raw_key_id if isinstance(raw_key_id, int) else None,
@@ -150,6 +228,10 @@ def resolve_credentials(profile: str = "default", agent_id: str | None = None) -
     base_url = normalize_url(
         os.getenv("HALIOS_BASE_URL") or str(entry.get("base_url") or DEFAULT_BASE_URL)
     )
+    ui_base_url = normalize_url(
+        os.getenv("HALIOS_UI_URL")
+        or str(entry.get("ui_base_url") or entry.get("base_url") or base_url)
+    )
     # INTENT: CI runners need both control-plane and agent-scoped ingest credentials without
     # writing a persistent profile to the ephemeral filesystem.
     token = os.getenv("HALIOS_OTLP_TOKEN")
@@ -160,6 +242,7 @@ def resolve_credentials(profile: str = "default", agent_id: str | None = None) -
     return Credentials(
         profile=profile,
         base_url=base_url,
+        ui_base_url=ui_base_url,
         api_key=env_key or str(entry.get("api_key") or ""),
         organization_id=entry.get("organization_id"),
         otlp_token=token,

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/halios_cli/cli_trace.py

@@ -7,7 +7,13 @@ import re
 
 import typer
 
-from .cli_support import ApiClient, load_project_config, resolve_credentials
+from .cli_support import (
+    ApiClient,
+    emit_review_links,
+    halios_ui_links,
+    load_project_config,
+    resolve_credentials,
+)
 
 app = typer.Typer(help="Inspect trace evidence and production failures.", no_args_is_help=True)
 
@@ -23,7 +29,17 @@ def _emit(value: object, json_output: bool) -> None:
     if json_output:
         typer.echo(json.dumps(value, indent=2, sort_keys=True, default=str))
     else:
-        items = value.get("data", value) if isinstance(value, dict) else value
+        links = value.get("links") if isinstance(value, dict) else None
+        display_value = (
+            {key: item for key, item in value.items() if key != "links"}
+            if isinstance(value, dict)
+            else value
+        )
+        items = (
+            display_value.get("data", display_value)
+            if isinstance(display_value, dict)
+            else display_value
+        )
         if isinstance(items, list):
             for item in items:
                 if isinstance(item, dict):
@@ -33,7 +49,9 @@ def _emit(value: object, json_output: bool) -> None:
                         )
                     )
         else:
-            typer.echo(json.dumps(value, indent=2, default=str))
+            typer.echo(json.dumps(display_value, indent=2, default=str))
+        if isinstance(links, dict):
+            emit_review_links(links)
 
 
 @app.command("list")
@@ -57,7 +75,7 @@ def show(
     include: str = typer.Option("spans,checks", "--include"),
     json_output: bool = typer.Option(False, "--json"),
 ) -> None:
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         result = api.request("GET", f"/api/v1/traces/{trace_id}")
     allowed = {item.strip() for item in include.split(",") if item.strip()}
@@ -66,6 +84,7 @@ def show(
             result.pop("spans", None)
         if "checks" not in allowed:
             result.pop("check_executions", None)
+        result["links"] = halios_ui_links(credentials.ui_base_url, agent_id, trace_id=trace_id)
     _emit(result, json_output)
 
 
@@ -122,7 +141,7 @@ def verify(
     json_output: bool = typer.Option(False, "--json"),
 ) -> None:
     """Fail closed unless a stored runtime trace has usable standard OTel evidence."""
-    _agent_id, credentials = _context()
+    agent_id, credentials = _context()
     with ApiClient(credentials) as api:
         detail = api.request("GET", f"/api/v1/traces/{trace_id}")
     spans = detail.get("spans") or []
@@ -169,6 +188,7 @@ def verify(
         "span_count": len(spans),
         "root_count": len(roots),
         "issues": issues,
+        "links": halios_ui_links(credentials.ui_base_url, agent_id, trace_id=trace_id),
     }
     _emit(result, json_output)
     if issues:

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6/haliosai_cli.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: haliosai-cli
-Version: 2.0.4
+Version: 2.0.6
 Summary: Halios CLI for coding-agent evaluations, guardrails, and OpenTelemetry evidence
 Author-email: HaliosLabs <support@halios.ai>
 License-Expression: Apache-2.0
@@ -65,10 +65,13 @@ npx skills add HaliosAI/halios --skill halios
 Prompt your agent to set up evaluations for your project:
 
 ```text
-"Set up evals for this agent: inspect the repository, create realistic test scenarios and checks, and run a baseline."
+"Set up evals for this agent: inspect the repository, create realistic test scenarios and checks, run one smoke test, and summarize before asking whether to run the full suite."
 ```
 
-Your agent will inspect your application entrypoint, configure standard OpenTelemetry, draft scenarios in `.halios/`, and run baseline evaluations.
+Your agent will inspect your application entrypoint, configure standard OpenTelemetry, draft
+scenarios in `.halios/`, and run one smoke scenario. It stops with a verification summary before
+asking whether to run the full suite and how many trials per scenario to use. The summary includes
+direct Halios UI links for reviewing the materialized suite, smoke run, and trace evidence.
 
 ---
 
@@ -78,15 +81,19 @@ If you prefer to drive evaluations directly from the command line or CI:
 
 ```bash
 # Recommended: Install with uv tool
-uv tool install 'haliosai-cli>=2.0.0'
+uv tool install 'haliosai-cli>=2.0.6'
 
 # Or install with pipx
-pipx install 'haliosai-cli>=2.0.0'
+pipx install 'haliosai-cli>=2.0.6'
 
 # See available commands and usage
 halios --help
 ```
 
+The API and UI use the same origin by default. For a self-hosted deployment with separate origins,
+set `HALIOS_UI_URL` so structured review links point to the web application without changing the
+API endpoint used by `HALIOS_BASE_URL`.
+
 ---
 
 ## How It Works

{haliosai_cli-2.0.4 → haliosai_cli-2.0.6}/tests/test_cli_surface.py

@@ -1,11 +1,15 @@
 from __future__ import annotations
 
+import json
 import sys
+from types import SimpleNamespace
 from typing import Any
 
+import pytest
+import typer
 from typer.testing import CliRunner
 
-from halios_cli import cli_support, cli_trace
+from halios_cli import cli_project, cli_support, cli_trace
 from halios_cli.cli import app
 from halios_cli.cli_eval import _invoke_adapter, _otlp_root_payload, _scenario_schema_errors
 
@@ -30,7 +34,8 @@ class FakeApiClient:
 
 def _patch_trace_context(monkeypatch) -> None:
     FakeApiClient.calls = []
-    monkeypatch.setattr(cli_trace, "_context", lambda: ("agent-id", object()))
+    credentials = SimpleNamespace(ui_base_url="https://app.halios.ai")
+    monkeypatch.setattr(cli_trace, "_context", lambda: ("agent-id", credentials))
     monkeypatch.setattr(cli_trace, "ApiClient", FakeApiClient)
 
 
@@ -46,12 +51,96 @@ def test_ci_credentials_accept_agent_scoped_otlp_token_from_environment(monkeypa
     monkeypatch.setenv("HALIOS_API_KEY", "halios_control_plane_key")
     monkeypatch.setenv("HALIOS_OTLP_TOKEN", "halios_agent_ingest_token")
     monkeypatch.setenv("HALIOS_BASE_URL", "https://api.halios.ai")
+    monkeypatch.setenv("HALIOS_UI_URL", "https://app.halios.ai")
 
     credentials = cli_support.resolve_credentials(agent_id="agent-id")
 
     assert credentials.api_key == "halios_control_plane_key"
     assert credentials.otlp_token == "halios_agent_ingest_token"
     assert credentials.base_url == "https://api.halios.ai"
+    assert credentials.ui_base_url == "https://app.halios.ai"
+
+
+def test_ui_links_are_encoded_and_strip_credentials() -> None:
+    links = cli_support.halios_ui_links(
+        "https://user:secret@app.halios.ai/api?token=do-not-copy",
+        "agent/id",
+        include_suite=True,
+        run_tag="run:hello world",
+        trace_id="1" * 32,
+        optimization_run_id="opt/id",
+    )
+
+    assert links["agent"] == "https://app.halios.ai/agents/agent%2Fid"
+    assert links["scenarios"].endswith("/evaluations?tab=scenarios")
+    assert links["rules"].endswith("/evaluations?tab=rules")
+    assert links["evaluation_run"].endswith("/evaluations/run%3Ahello%20world")
+    assert links["evaluation_traces"].endswith("/traces?eval_run=run%3Ahello+world")
+    assert links["optimization_run"].endswith("/optimize/opt%2Fid")
+    assert "secret" not in "\n".join(links.values())
+    assert "token" not in "\n".join(links.values())
+
+
+def test_ui_links_require_agent_identity() -> None:
+    assert cli_support.halios_ui_links("https://app.halios.ai", "") == {}
+
+    with pytest.raises(typer.BadParameter, match="Invalid Halios base URL"):
+        cli_support.halios_ui_links("ftp://app.halios.ai", "agent-id")
+
+
+def test_project_init_json_returns_agent_review_link(monkeypatch, tmp_path) -> None:
+    agent_id = "11111111-1111-4111-8111-111111111111"
+    credentials = SimpleNamespace(
+        base_url="https://api.halios.ai",
+        ui_base_url="https://app.halios.ai",
+        otlp_token="stored-token",
+    )
+
+    class InitApi:
+        def __init__(self, _credentials: object) -> None:
+            pass
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, *_args):
+            return None
+
+        def request(self, method: str, path: str, **_kwargs):
+            if method == "GET" and path == "/api/v1/ai/capability":
+                return {"evaluation_available": True}
+            if method == "POST" and path == "/api/v1/agents":
+                return {"id": agent_id, "name": "Support agent", "slug": "support-agent"}
+            if path == f"/api/v1/agents/{agent_id}/evaluation-suite":
+                return {
+                    "revision": 0,
+                    "digest": "",
+                    "eval": {},
+                    "scenarios": {"version": 1, "scenarios": []},
+                }
+            raise AssertionError(f"Unexpected request: {method} {path}")
+
+    monkeypatch.chdir(tmp_path)
+    monkeypatch.setattr(cli_project, "resolve_credentials", lambda *_args: credentials)
+    monkeypatch.setattr(cli_project, "ApiClient", InitApi)
+
+    result = CliRunner().invoke(
+        app,
+        [
+            "project",
+            "init",
+            "--agent",
+            "Support agent",
+            "--command",
+            "python adapter.py",
+            "--json",
+        ],
+    )
+
+    assert result.exit_code == 0
+    payload = json.loads(result.output)
+    assert payload["agent"]["id"] == agent_id
+    assert payload["links"]["agent"] == f"https://app.halios.ai/agents/{agent_id}"
 
 
 def test_trace_list_uses_backend_traffic_scope(monkeypatch) -> None:
@@ -100,6 +189,7 @@ def test_trace_verify_accepts_complete_root(monkeypatch) -> None:
     result = CliRunner().invoke(app, ["trace", "verify", trace_id, "--json"])
     assert result.exit_code == 0
     assert '"verified": true' in result.output
+    assert '"trace": "https://app.halios.ai/agents/agent-id/traces/' in result.output
 
 
 def test_cli_root_span_uses_standard_genai_message_parts() -> None:

haliosai_cli-2.0.4/halios_cli/_version.py

@@ -1 +0,0 @@
-__version__ = "2.0.4"