codeprobe 0.2.8__tar.gz → 0.3.0__tar.gz

{codeprobe-0.2.8 → codeprobe-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codeprobe
-Version: 0.2.8
+Version: 0.3.0
 Summary: Benchmark AI coding agents against your own codebase. Mine real tasks from repo history, run agents, interpret results.
 Author: codeprobe contributors
 License-Expression: Apache-2.0
@@ -24,6 +24,7 @@ Requires-Dist: anthropic>=0.39
 Requires-Dist: openai>=1.66
 Requires-Dist: tiktoken<1,>=0.7
 Requires-Dist: scipy<2,>=1.11
+Requires-Dist: rich<14,>=13.7
 Provides-Extra: dev
 Requires-Dist: pytest<9,>=8.0; extra == "dev"
 Requires-Dist: pytest-cov<6,>=5.0; extra == "dev"
@@ -84,18 +85,20 @@ codeprobe interpret .   # Get recommendations
 
 ## Commands
 
-| Command                  | Purpose                                          |
-| ------------------------ | ------------------------------------------------ |
-| `codeprobe assess`       | Score a codebase's benchmarking potential        |
-| `codeprobe init`         | Interactive wizard — choose what to compare      |
-| `codeprobe mine`         | Mine eval tasks from merged PRs/MRs              |
-| `codeprobe probe`        | Generate fast micro-benchmark probes (30s each)  |
-| `codeprobe experiment`   | Manage comparison experiments (init, add-config) |
-| `codeprobe run`          | Execute tasks against AI agents                  |
-| `codeprobe interpret`    | Analyze results, rank configurations             |
-| `codeprobe oracle-check` | Compare agent answer against oracle ground truth |
-| `codeprobe scaffold`     | Create/validate eval task directories            |
-| `codeprobe ratings`      | Record and analyze agent session quality ratings |
+| Command                    | Purpose                                          |
+| -------------------------- | ------------------------------------------------ |
+| `codeprobe assess`         | Score a codebase's benchmarking potential        |
+| `codeprobe init`           | Interactive wizard — choose what to compare      |
+| `codeprobe mine`           | Mine eval tasks from merged PRs/MRs              |
+| `codeprobe probe`          | Generate fast micro-benchmark probes (30s each)  |
+| `codeprobe experiment`     | Manage comparison experiments (init, add-config) |
+| `codeprobe run`            | Execute tasks against AI agents                  |
+| `codeprobe interpret`      | Analyze results, rank configurations             |
+| `codeprobe doctor`         | Check environment readiness (agents, keys, git)  |
+| `codeprobe preambles list` | List available preambles at all search levels    |
+| `codeprobe oracle-check`   | Compare agent answer against oracle ground truth |
+| `codeprobe scaffold`       | Create/validate eval task directories            |
+| `codeprobe ratings`        | Record and analyze agent session quality ratings |
 
 ## Two Ways to Generate Tasks
 
@@ -181,17 +184,32 @@ Template variables: `{{sg_repo}}`, `{{repo_name}}`, `{{repo_path}}`, `{{task_id}
 codeprobe run . --parallel 5          # Run 5 tasks concurrently (worktree-isolated)
 codeprobe run . --max-cost-usd 2.00   # Stop when cost budget is reached
 codeprobe run . --dry-run             # Estimate resource usage without running
+codeprobe run . --model opus-4        # Override experiment.json model
+codeprobe run . --timeout 600         # Override default 300s timeout
+codeprobe run . --repeats 3           # Run each task 3 times
+codeprobe run . --show-prompt         # Print resolved prompt without running agent
 
 # Mining
 codeprobe mine . --enrich             # Use LLM to improve weak task instructions
 codeprobe mine . --org-scale          # Mine comprehension tasks (not SDLC)
 codeprobe mine . --mcp-families       # Include MCP-optimized task families
 codeprobe mine . --sg-repo REPO       # Sourcegraph repo for ground truth enrichment
+codeprobe mine . --preset quick       # Quick scan: count=3
+codeprobe mine . --preset mcp         # MCP eval: org-scale + MCP families + enrich
+
+# Mine profiles (save/load custom flag combinations)
+codeprobe mine --save-profile my-setup --count 10 --org-scale .
+codeprobe mine --profile my-setup .   # Load saved flags
+codeprobe mine --list-profiles        # Show available profiles
 
 # Experiment configs
 codeprobe experiment add-config . --preamble sourcegraph  # Attach MCP preamble
 codeprobe experiment add-config . --mcp-config config.json  # Attach MCP server
 
+# Diagnostics
+codeprobe doctor                      # Check agents, API keys, git, Python
+codeprobe preambles list              # Show available preambles at all levels
+
 # Output
 codeprobe interpret . --format csv    # Export for pivot tables
 codeprobe interpret . --format html   # Self-contained HTML report
@@ -210,14 +228,9 @@ GitHub, GitLab, Bitbucket, Azure DevOps, Gitea/Forgejo, and local repos.
 
 ## Configuration
 
-Create a `.evalrc.yaml` in your repo root:
+Configuration lives in `experiment.json` (created by `codeprobe init` or `codeprobe experiment init`). CLI flags override experiment.json values — precedence: built-in defaults < experiment.json < CLI flags.
 
-```yaml
-name: my-experiment
-agents: [claude, copilot]
-models: [claude-sonnet-4-6, claude-opus-4-6]
-tasks_dir: .codeprobe/tasks
-```
+Run-time observability is on by default: Rich Live dashboard in TTY, JSON event lines with `--log-format json` for CI. Cost budget warnings at 80% and 100% thresholds are always visible on stderr.
 
 ## License
 

{codeprobe-0.2.8 → codeprobe-0.3.0}/README.md

@@ -49,18 +49,20 @@ codeprobe interpret .   # Get recommendations
 
 ## Commands
 
-| Command                  | Purpose                                          |
-| ------------------------ | ------------------------------------------------ |
-| `codeprobe assess`       | Score a codebase's benchmarking potential        |
-| `codeprobe init`         | Interactive wizard — choose what to compare      |
-| `codeprobe mine`         | Mine eval tasks from merged PRs/MRs              |
-| `codeprobe probe`        | Generate fast micro-benchmark probes (30s each)  |
-| `codeprobe experiment`   | Manage comparison experiments (init, add-config) |
-| `codeprobe run`          | Execute tasks against AI agents                  |
-| `codeprobe interpret`    | Analyze results, rank configurations             |
-| `codeprobe oracle-check` | Compare agent answer against oracle ground truth |
-| `codeprobe scaffold`     | Create/validate eval task directories            |
-| `codeprobe ratings`      | Record and analyze agent session quality ratings |
+| Command                    | Purpose                                          |
+| -------------------------- | ------------------------------------------------ |
+| `codeprobe assess`         | Score a codebase's benchmarking potential        |
+| `codeprobe init`           | Interactive wizard — choose what to compare      |
+| `codeprobe mine`           | Mine eval tasks from merged PRs/MRs              |
+| `codeprobe probe`          | Generate fast micro-benchmark probes (30s each)  |
+| `codeprobe experiment`     | Manage comparison experiments (init, add-config) |
+| `codeprobe run`            | Execute tasks against AI agents                  |
+| `codeprobe interpret`      | Analyze results, rank configurations             |
+| `codeprobe doctor`         | Check environment readiness (agents, keys, git)  |
+| `codeprobe preambles list` | List available preambles at all search levels    |
+| `codeprobe oracle-check`   | Compare agent answer against oracle ground truth |
+| `codeprobe scaffold`       | Create/validate eval task directories            |
+| `codeprobe ratings`        | Record and analyze agent session quality ratings |
 
 ## Two Ways to Generate Tasks
 
@@ -146,17 +148,32 @@ Template variables: `{{sg_repo}}`, `{{repo_name}}`, `{{repo_path}}`, `{{task_id}
 codeprobe run . --parallel 5          # Run 5 tasks concurrently (worktree-isolated)
 codeprobe run . --max-cost-usd 2.00   # Stop when cost budget is reached
 codeprobe run . --dry-run             # Estimate resource usage without running
+codeprobe run . --model opus-4        # Override experiment.json model
+codeprobe run . --timeout 600         # Override default 300s timeout
+codeprobe run . --repeats 3           # Run each task 3 times
+codeprobe run . --show-prompt         # Print resolved prompt without running agent
 
 # Mining
 codeprobe mine . --enrich             # Use LLM to improve weak task instructions
 codeprobe mine . --org-scale          # Mine comprehension tasks (not SDLC)
 codeprobe mine . --mcp-families       # Include MCP-optimized task families
 codeprobe mine . --sg-repo REPO       # Sourcegraph repo for ground truth enrichment
+codeprobe mine . --preset quick       # Quick scan: count=3
+codeprobe mine . --preset mcp         # MCP eval: org-scale + MCP families + enrich
+
+# Mine profiles (save/load custom flag combinations)
+codeprobe mine --save-profile my-setup --count 10 --org-scale .
+codeprobe mine --profile my-setup .   # Load saved flags
+codeprobe mine --list-profiles        # Show available profiles
 
 # Experiment configs
 codeprobe experiment add-config . --preamble sourcegraph  # Attach MCP preamble
 codeprobe experiment add-config . --mcp-config config.json  # Attach MCP server
 
+# Diagnostics
+codeprobe doctor                      # Check agents, API keys, git, Python
+codeprobe preambles list              # Show available preambles at all levels
+
 # Output
 codeprobe interpret . --format csv    # Export for pivot tables
 codeprobe interpret . --format html   # Self-contained HTML report
@@ -175,14 +192,9 @@ GitHub, GitLab, Bitbucket, Azure DevOps, Gitea/Forgejo, and local repos.
 
 ## Configuration
 
-Create a `.evalrc.yaml` in your repo root:
+Configuration lives in `experiment.json` (created by `codeprobe init` or `codeprobe experiment init`). CLI flags override experiment.json values — precedence: built-in defaults < experiment.json < CLI flags.
 
-```yaml
-name: my-experiment
-agents: [claude, copilot]
-models: [claude-sonnet-4-6, claude-opus-4-6]
-tasks_dir: .codeprobe/tasks
-```
+Run-time observability is on by default: Rich Live dashboard in TTY, JSON event lines with `--log-format json` for CI. Cost budget warnings at 80% and 100% thresholds are always visible on stderr.
 
 ## License
 

{codeprobe-0.2.8 → codeprobe-0.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "codeprobe"
-version = "0.2.8"
+version = "0.3.0"
 description = "Benchmark AI coding agents against your own codebase. Mine real tasks from repo history, run agents, interpret results."
 readme = "README.md"
 license = "Apache-2.0"
@@ -25,6 +25,7 @@ dependencies = [
     "openai>=1.66",
     "tiktoken>=0.7,<1",
     "scipy>=1.11,<2",
+    "rich>=13.7,<14",
 ]
 
 [project.urls]
@@ -57,6 +58,12 @@ claude = "codeprobe.adapters.session:ClaudeSessionCollector"
 codex = "codeprobe.adapters.session:CodexSessionCollector"
 copilot = "codeprobe.adapters.session:CopilotSessionCollector"
 
+[project.entry-points."codeprobe.scorers"]
+binary = "codeprobe.core.scoring:BinaryScorer"
+continuous = "codeprobe.core.scoring:ContinuousScorer"
+checkpoint = "codeprobe.core.scoring:CheckpointScorer"
+test_ratio = "codeprobe.core.scoring:ContinuousScorer"
+
 [build-system]
 requires = ["setuptools>=68", "wheel"]
 build-backend = "setuptools.build_meta"

{codeprobe-0.2.8 → codeprobe-0.3.0}/src/codeprobe/__init__.py

@@ -1,3 +1,3 @@
 """codeprobe — Benchmark AI coding agents against your own codebase."""
 
-__version__ = "0.2.8"
+__version__ = "0.3.0"

{codeprobe-0.2.8 → codeprobe-0.3.0}/src/codeprobe/cli/__init__.py

@@ -84,6 +84,10 @@ def main(verbose: int, quiet: bool, log_format: str) -> None:
     and interpret the results to find which setup works best for YOUR code.
     """
     _configure_logging(verbose=verbose, quiet=quiet, log_format=log_format)
+    ctx = click.get_current_context()
+    ctx.ensure_object(dict)
+    ctx.obj["log_format"] = log_format
+    ctx.obj["quiet"] = quiet
 
 
 @main.command()
@@ -101,6 +105,32 @@ def init(path: str) -> None:
 
 @main.command()
 @click.argument("path", default=".")
+@click.option(
+    "--preset",
+    type=click.Choice(["quick", "mcp"], case_sensitive=False),
+    default=None,
+    help="Apply a named preset: 'quick' (count=3) or 'mcp' (org-scale + MCP families).",
+)
+@click.option(
+    "--profile",
+    "profile_name",
+    default=None,
+    help="Load a user-defined profile from ~/.codeprobe/mine-profiles.json "
+    "or .codeprobe/mine-profiles.json. Explicit flags override profile values.",
+)
+@click.option(
+    "--save-profile",
+    "save_profile_name",
+    default=None,
+    help="Save current flag values as a named profile to ~/.codeprobe/mine-profiles.json.",
+)
+@click.option(
+    "--list-profiles",
+    "list_profiles_flag",
+    is_flag=True,
+    default=False,
+    help="Show available profiles from user and project levels.",
+)
 @click.option("--count", default=5, help="Number of tasks to mine (3-20).")
 @click.option(
     "--source",
@@ -206,8 +236,14 @@ def init(path: str) -> None:
     "(e.g. github.com/sg-evals/numpy). Defaults to github.com/sg-evals/{repo_name} "
     "when --mcp-families is used. Requires SOURCEGRAPH_TOKEN env var.",
 )
+@click.pass_context
 def mine(
+    ctx: click.Context,
     path: str,
+    preset: str | None,
+    profile_name: str | None,
+    save_profile_name: str | None,
+    list_profiles_flag: bool,
     count: int,
     source: str,
     min_files: int,
@@ -232,6 +268,21 @@ def mine(
     Extracts real code-change tasks from merged PRs/MRs with ground truth,
     test scripts, and scoring rubrics.
 
+    \b
+    Presets (--preset):
+      quick  — Fast scan: count=3, default SDLC mode
+      mcp    — MCP eval: count=8, org-scale + MCP families + enrich
+
+    \b
+    Profiles (--profile / --save-profile / --list-profiles):
+      Save:  codeprobe mine --save-profile my-setup --count 10 --org-scale .
+      Load:  codeprobe mine --profile my-setup /path/to/repo
+      List:  codeprobe mine --list-profiles
+
+    \b
+    Precedence: built-in defaults < profile < --preset < explicit CLI flags.
+
+    \b
     Use --org-scale to mine comprehension/IR tasks with oracle verification
     instead of SDLC code-change tasks.
 
@@ -242,10 +293,98 @@ def mine(
     choosing an eval goal, task count, and git host before mining.
     Use --no-interactive to skip the prompts and use defaults/flags directly.
     """
-    from codeprobe.cli.mine_cmd import run_mine
+    from pathlib import Path as _Path
+
+    from codeprobe.cli.mine_cmd import (
+        list_profiles,
+        load_profile,
+        run_mine,
+        save_profile,
+    )
+
+    # --list-profiles: show and exit
+    if list_profiles_flag:
+        repo_path = _Path(path).resolve() if path != "." else _Path.cwd()
+        entries = list_profiles(repo_path)
+        if not entries:
+            click.echo("No profiles found.")
+        else:
+            click.echo(f"{'Name':<20s} {'Source':<10s} {'Settings'}")
+            click.echo("-" * 60)
+            for name, source_label, prof in entries:
+                summary = ", ".join(f"{k}={v}" for k, v in sorted(prof.items()))
+                click.echo(f"{name:<20s} {source_label:<10s} {summary}")
+        return
+
+    # --save-profile: save current flags and exit
+    if save_profile_name is not None:
+        # Collect all current param values, keeping only those that differ
+        # from Click defaults.
+        param_defaults = {p.name: p.default for p in ctx.command.params}
+        # Exclude meta-params that aren't mining flags
+        _EXCLUDE_FROM_PROFILE = frozenset(
+            {
+                "path",
+                "profile_name",
+                "save_profile_name",
+                "list_profiles_flag",
+            }
+        )
+        values = {
+            k: (list(v) if isinstance(v, tuple) else v)
+            for k, v in ctx.params.items()
+            if k not in _EXCLUDE_FROM_PROFILE and v != param_defaults.get(k)
+        }
+        saved_path = save_profile(save_profile_name, values)
+        click.echo(f"Profile '{save_profile_name}' saved to {saved_path}")
+        return
+
+    # --profile: load profile values as defaults, then apply preset and CLI overrides
+    if profile_name is not None:
+        repo_path = _Path(path).resolve() if path != "." else _Path.cwd()
+        prof = load_profile(profile_name, repo_path)
+
+        # Determine which params were explicitly set on the CLI
+        explicitly_set = {
+            p.name
+            for p in ctx.command.params
+            if ctx.get_parameter_source(p.name) is not None
+            and ctx.get_parameter_source(p.name).name == "COMMANDLINE"
+        }
+
+        # Apply profile values for params NOT explicitly set on CLI.
+        # Tuple-typed params (click multiple=True) need list→tuple coercion.
+        _TUPLE_PARAMS = frozenset({"subsystem", "family", "repos", "backends"})
+
+        def _prof_val(key: str, current: object) -> object:
+            if key in explicitly_set or key not in prof:
+                return current
+            v = prof[key]
+            return tuple(v) if key in _TUPLE_PARAMS else v
+
+        count = _prof_val("count", count)  # type: ignore[assignment]
+        source = _prof_val("source", source)  # type: ignore[assignment]
+        min_files = _prof_val("min_files", min_files)  # type: ignore[assignment]
+        enrich = _prof_val("enrich", enrich)  # type: ignore[assignment]
+        org_scale = _prof_val("org_scale", org_scale)  # type: ignore[assignment]
+        mcp_families = _prof_val("mcp_families", mcp_families)  # type: ignore[assignment]
+        no_llm = _prof_val("no_llm", no_llm)  # type: ignore[assignment]
+        discover_subsystems = _prof_val("discover_subsystems", discover_subsystems)  # type: ignore[assignment]
+        scan_timeout = _prof_val("scan_timeout", scan_timeout)  # type: ignore[assignment]
+        validate_flag = _prof_val("validate_flag", validate_flag)  # type: ignore[assignment]
+        curate = _prof_val("curate", curate)  # type: ignore[assignment]
+        verify_curation_flag = _prof_val("verify_curation_flag", verify_curation_flag)  # type: ignore[assignment]
+        sg_repo = _prof_val("sg_repo", sg_repo)  # type: ignore[assignment]
+        subsystem = _prof_val("subsystem", subsystem)  # type: ignore[assignment]
+        family = _prof_val("family", family)  # type: ignore[assignment]
+        repos = _prof_val("repos", repos)  # type: ignore[assignment]
+        backends = _prof_val("backends", backends)  # type: ignore[assignment]
+        interactive = _prof_val("interactive", interactive)  # type: ignore[assignment]
+        preset = _prof_val("preset", preset)  # type: ignore[assignment]
 
     run_mine(
         path,
+        preset=preset,
         count=count,
         source=source,
         min_files=min_files,
@@ -294,7 +433,39 @@ def mine(
     default=False,
     help="Print estimated resource requirements without executing any agents.",
 )
+@click.option(
+    "--force-plain",
+    is_flag=True,
+    default=False,
+    help="Force plain-text output even in a TTY (disable Rich dashboard).",
+)
+@click.option(
+    "--force-rich",
+    is_flag=True,
+    default=False,
+    help="Force Rich Live dashboard even in non-TTY environments.",
+)
+@click.option(
+    "--timeout",
+    default=None,
+    type=int,
+    help="Timeout in seconds per task (overrides experiment.json extra.timeout_seconds).",
+)
+@click.option(
+    "--repeats",
+    default=None,
+    type=int,
+    help="Number of repeats per task (overrides default of 1).",
+)
+@click.option(
+    "--show-prompt",
+    is_flag=True,
+    default=False,
+    help="Print the fully-resolved prompt for the first task and exit (no agent spawned).",
+)
+@click.pass_context
 def run(
+    ctx: click.Context,
     path: str,
     agent: str,
     model: str | None,
@@ -302,6 +473,11 @@ def run(
     max_cost_usd: float | None,
     parallel: int,
     dry_run: bool,
+    force_plain: bool,
+    force_rich: bool,
+    timeout: int | None,
+    repeats: int | None,
+    show_prompt: bool,
 ) -> None:
     """Run eval tasks against an AI coding agent.
 
@@ -310,6 +486,16 @@ def run(
     """
     from codeprobe.cli.run_cmd import run_eval
 
+    ctx.ensure_object(dict)
+    log_format = ctx.obj.get("log_format", "text")
+    quiet = ctx.obj.get("quiet", False)
+
+    if show_prompt:
+        from codeprobe.cli.run_cmd import show_prompt_and_exit
+
+        show_prompt_and_exit(path, config=config, agent=agent, model=model)
+        return
+
     run_eval(
         path,
         agent=agent,
@@ -318,6 +504,12 @@ def run(
         max_cost_usd=max_cost_usd,
         parallel=parallel,
         dry_run=dry_run,
+        log_format=log_format,
+        quiet=quiet,
+        force_plain=force_plain,
+        force_rich=force_rich,
+        timeout=timeout,
+        repeats=repeats if repeats is not None else 1,
     )
 
 
@@ -488,3 +680,13 @@ main.add_command(scaffold)
 from codeprobe.cli.probe_cmd import probe  # noqa: E402
 
 main.add_command(probe)
+
+# Register the preambles subcommand group
+from codeprobe.cli.preamble_cmd import preambles  # noqa: E402
+
+main.add_command(preambles)
+
+# Register the doctor command
+from codeprobe.cli.doctor_cmd import doctor  # noqa: E402
+
+main.add_command(doctor)

codeprobe-0.3.0/src/codeprobe/cli/doctor_cmd.py

@@ -0,0 +1,115 @@
+"""Doctor command — checks environment readiness for codeprobe."""
+
+from __future__ import annotations
+
+import os
+import shutil
+import subprocess
+import sys
+from dataclasses import dataclass
+
+import click
+
+
+@dataclass(frozen=True)
+class CheckResult:
+    name: str
+    passed: bool
+    detail: str
+    fix: str
+
+
+def _check_tool(name: str, fix: str) -> CheckResult:
+    found = shutil.which(name) is not None
+    return CheckResult(
+        name=f"{name} CLI",
+        passed=found,
+        detail="found" if found else "not found",
+        fix=fix,
+    )
+
+
+def _check_env_key(key: str, fix: str) -> CheckResult:
+    present = key in os.environ and len(os.environ[key]) > 0
+    return CheckResult(
+        name=key,
+        passed=present,
+        detail="set" if present else "not set",
+        fix=fix,
+    )
+
+
+def _check_git_repo() -> CheckResult:
+    try:
+        result = subprocess.run(
+            ["git", "rev-parse", "--is-inside-work-tree"],
+            capture_output=True,
+            text=True,
+            timeout=5,
+        )
+        is_repo = result.returncode == 0
+    except (FileNotFoundError, subprocess.TimeoutExpired):
+        is_repo = False
+    return CheckResult(
+        name="git repo",
+        passed=is_repo,
+        detail="inside git repo" if is_repo else "not a git repository",
+        fix="Run 'git init' or cd into an existing git repository.",
+    )
+
+
+def _check_python_version() -> CheckResult:
+    major, minor = sys.version_info[:2]
+    ok = (major, minor) >= (3, 11)
+    return CheckResult(
+        name="Python version",
+        passed=ok,
+        detail=f"{major}.{minor}",
+        fix="Install Python 3.11 or later. See https://www.python.org/downloads/",
+    )
+
+
+def run_checks() -> list[CheckResult]:
+    """Run all environment checks and return results."""
+    return [
+        _check_tool(
+            "claude",
+            "Install Claude Code: https://docs.anthropic.com/en/docs/claude-code",
+        ),
+        _check_tool(
+            "copilot",
+            "Install GitHub Copilot CLI: https://github.com/github/gh-copilot",
+        ),
+        _check_tool(
+            "codex", "Install OpenAI Codex CLI: https://github.com/openai/codex"
+        ),
+        _check_tool("aider", "Install aider: https://aider.chat/docs/install.html"),
+        _check_env_key(
+            "ANTHROPIC_API_KEY", "Set ANTHROPIC_API_KEY in your environment."
+        ),
+        _check_env_key("OPENAI_API_KEY", "Set OPENAI_API_KEY in your environment."),
+        _check_env_key(
+            "GITHUB_TOKEN",
+            "Set GITHUB_TOKEN in your environment. See https://github.com/settings/tokens",
+        ),
+        _check_git_repo(),
+        _check_python_version(),
+    ]
+
+
+@click.command("doctor")
+def doctor() -> None:
+    """Check environment readiness for running codeprobe."""
+    results = run_checks()
+    any_failed = False
+
+    for r in results:
+        if r.passed:
+            click.echo(f"  PASS  {r.name} ({r.detail})")
+        else:
+            any_failed = True
+            click.echo(f"  FAIL  {r.name} ({r.detail})")
+            click.echo(f"        -> {r.fix}")
+
+    if any_failed:
+        raise SystemExit(1)

{codeprobe-0.2.8 → codeprobe-0.3.0}/src/codeprobe/cli/experiment_cmd.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 
 import json
 import statistics
+import sys
 from datetime import datetime, timezone
 from pathlib import Path
 
@@ -56,6 +57,36 @@ def experiment_init(
     )
 
 
+def _interactive_mcp_selection() -> str | None:
+    """Offer interactive MCP config selection when available.
+
+    Returns a file path string if the user selects a config, or None to skip.
+    """
+    from codeprobe.core.mcp_discovery import discover_mcp_configs
+
+    discovered = discover_mcp_configs()
+    if not discovered:
+        return None
+
+    click.echo()
+    click.echo("Discovered MCP configurations:")
+    for i, (p, servers) in enumerate(discovered, 1):
+        click.echo(f"  {i}. {p}  ({len(servers)} servers)")
+        for s in servers:
+            click.echo(f"     - {s}")
+    click.echo(f"  {len(discovered) + 1}. Skip (no MCP config)")
+    click.echo()
+
+    choice = click.prompt(
+        "Select MCP config",
+        type=click.IntRange(1, len(discovered) + 1),
+        default=len(discovered) + 1,
+    )
+    if choice <= len(discovered):
+        return str(discovered[choice - 1][0])
+    return None
+
+
 def experiment_add_config(
     path: str,
     label: str,
@@ -84,7 +115,7 @@ def experiment_add_config(
         )
         raise SystemExit(1)
 
-    # Parse MCP config
+    # Parse MCP config — offer interactive discovery when omitted in a TTY
     mcp_config: dict | None = None
     if mcp_config_str:
         try:
@@ -99,6 +130,12 @@ def experiment_add_config(
                     err=True,
                 )
                 raise SystemExit(1)
+    elif sys.stderr.isatty():
+        mcp_config_str = _interactive_mcp_selection()
+        if mcp_config_str:
+            mcp_path = Path(mcp_config_str).expanduser().resolve()
+            if mcp_path.is_file():
+                mcp_config = json.loads(mcp_path.read_text(encoding="utf-8"))
 
     new_config = ExperimentConfig(
         label=label,