cleveragents-core/tools/mcp_git_server.py

#!/usr/bin/env python3
"""MCP server wrapping git plumbing for the auto-agents pipeline.

Replaces the fleet of single-op subagents in ``.opencode/agents/`` —
``git-isolator-util``, ``git-clone-util``, ``git-fetch-util``,
``git-checkout-util``, ``git-stage-util``, ``git-commit-util``,
``git-create-commit-util``, ``git-commit-and-push-util``,
``git-push-util``, ``git-rebase-util``, ``git-rebase-and-push-util``,
``git-force-push-with-lease-util``, ``git-cleanup-util``. That layer
existed only as a permission-scoping pattern (each agent's narrow
bash allowlist covered exactly one git operation). With this MCP,
agents call typed tools — no subagent overhead, no per-op prompt
boilerplate, and the worktree-path allowlist lives in one file
instead of being scattered across 13.

Path safety
-----------

Every tool that takes a ``worktree`` argument validates the path
against an allowlist of acceptable bases — currently
``/tmp/cleveragents-implementer-worktrees/`` and
``/tmp/cleveragents-review-worktrees/``. A request to operate on
``/etc``, ``/`` or the host repo is rejected with a clear error
before any git invocation runs. The allowlist can be extended via
``MCP_GIT_WORKTREE_BASES`` (colon-separated paths) for edge cases
like local development or future dispatchers.

Identity
--------

Pushes authenticate as HAL9000 (``FORGEJO_PAT``) — the worker
identity for code changes. The reviewer never commits or pushes
under normal operation; if a future workflow needs it, add an
explicit ``as_reviewer=True`` parameter to ``push`` rather than
flipping a runtime toggle. Author/committer identity for
:func:`commit` comes from ``GIT_USER_NAME`` / ``GIT_USER_EMAIL``
which ``tools/launch_fork.sh`` sets to ``CleverThis`` /
``hal9000@cleverthis.com``.

Tools
-----

``isolate(pr, head_sha, head_ref=None, kind="implementer")``
    Pre-clone a PR's head SHA into a fresh worktree under the
    kind-specific base. Wraps ``_pr_clone.prepare_pr_worktree``.
    Returns ``{worktree, branch, kind, run_tag}`` or
    ``{error: str}``.

``status(worktree)`` / ``stage(worktree, paths)`` /
``commit(worktree, message, author_name?, author_email?)`` /
``push(worktree, remote="origin", force_with_lease=False)`` /
``fetch(worktree, remote="origin")`` /
``rebase(worktree, onto)`` / ``rebase_continue(worktree)`` /
``cleanup(worktree)``
    Standard git plumbing — see each tool's docstring for shape.
    ``rebase_continue`` resumes a conflict-stopped rebase after the
    agent has resolved + staged the current step (mid-rebase mode);
    ``status`` also reports ``rebase_in_progress`` /
    ``merge_in_progress`` / a ``conflicts`` list so the conflict-
    resolver can drive either a mid-rebase or a mid-merge worktree.

Each write tool returns a small structured dict with the next-step
fact the agent needs (e.g. ``commit`` returns ``{sha}``,
``push`` returns ``{remote_sha}``, ``rebase`` returns
``{success, conflicts}``). On failure the tool returns
``{error: str, stderr: str}`` so the agent can surface the real
git error without re-reading transcript output.
"""

from __future__ import annotations

import os
import shutil
import subprocess
import sys
import tempfile
import uuid
from pathlib import Path
from typing import Any

from mcp.server.fastmcp import FastMCP

sys.path.insert(0, str(Path(__file__).parent))
from _mcp_common import (  # noqa: E402
    ForgejoCfg,
    bootstrap_loader,
    error_envelope as _err,
    make_main,
    require_token,
)

load_sibling = bootstrap_loader()
_pr_clone = load_sibling("_pr_clone", "_pr_clone.py")

server = FastMCP("git")


# ─── Worktree allowlist ────────────────────────────────────────────
def _allowed_bases() -> tuple[Path, ...]:
    """Resolve the set of base directories any ``worktree`` argument
    is allowed to live under. The defaults match the dispatcher's
    pre-clone targets; the env var lets operators extend (e.g. for
    a one-off debug worktree elsewhere)."""
    defaults = (
        "/tmp/cleveragents-implementer-worktrees",
        "/tmp/cleveragents-review-worktrees",
        # The state-machine controller pre-clones every role's worktree
        # under this root (workspace.py DEFAULT_WORKSPACE_ROOT). Without
        # it the conflict-resolver's git_* MCP calls all error and the
        # agent silently falls back to bash.
        "/tmp/cleveragents-controller",
    )
    extra = os.environ.get("MCP_GIT_WORKTREE_BASES", "").strip()
    extras = tuple(p for p in extra.split(":") if p) if extra else ()
    return tuple(Path(p).resolve() for p in (defaults + extras))


def _validate_worktree(worktree: str) -> tuple[Path | None, str | None]:
    """Return ``(resolved_path, None)`` if ``worktree`` is allowed,
    or ``(None, error_message)`` if not. Resolves symlinks so an
    agent cannot escape the allowlist by linking through /tmp."""
    if not worktree or not worktree.strip():
        return None, "worktree path must be non-empty"
    try:
        resolved = Path(worktree).expanduser().resolve()
    except (OSError, RuntimeError) as exc:
        return None, f"could not resolve {worktree!r}: {exc}"
    if not resolved.exists():
        return None, f"worktree does not exist at {resolved}"
    if not resolved.is_dir():
        return None, f"worktree is not a directory: {resolved}"
    for base in _allowed_bases():
        try:
            resolved.relative_to(base)
            return resolved, None
        except ValueError:
            continue
    bases = ", ".join(str(b) for b in _allowed_bases())
    return None, (
        f"worktree {resolved} is outside the allowed bases ({bases}). "
        "Set MCP_GIT_WORKTREE_BASES to extend if this is intentional."
    )


# ─── Subprocess wrapping ───────────────────────────────────────────
_DEFAULT_GIT_TIMEOUT_S = int(os.environ.get("MCP_GIT_TIMEOUT_S", "300"))


def _ensure_askpass_for_hal9000() -> Path:
    """Create a 0700 askpass shim that returns the HAL9000 PAT.

    Mirrors the pattern in ``_pr_clone_creds._ensure_askpass_script``
    but pins to ``FORGEJO_PAT`` (HAL9000) instead of
    ``FORGEJO_REVIEWER_PAT`` (HAL9001). The MCP server is per-process
    long-lived, so the script is created lazily on first push/fetch
    and reused thereafter.
    """
    global _ASKPASS_PATH
    if _ASKPASS_PATH is not None and _ASKPASS_PATH.exists():
        return _ASKPASS_PATH
    fd, path = tempfile.mkstemp(prefix="mcp-git-askpass-", suffix=".sh")
    with os.fdopen(fd, "w") as f:
        f.write(
            "#!/bin/sh\n"
            "# Generated by tools/mcp_git_server.py — HAL9000 (FORGEJO_PAT).\n"
            'prompt="${1:-}"\n'
            'lower=$(printf "%s" "$prompt" | tr "[:upper:]" "[:lower:]")\n'
            'case "$lower" in\n'
            "  *username*)\n"
            '    printf "%s\\n" "${FORGEJO_USERNAME:-x-token-auth}"\n'
            "    ;;\n"
            "  *)\n"
            '    printf "%s\\n" "${FORGEJO_PAT:-${GITEA_TOKEN:-}}"\n'
            "    ;;\n"
            "esac\n"
        )
    os.chmod(path, 0o700)
    _ASKPASS_PATH = Path(path)
    return _ASKPASS_PATH


_ASKPASS_PATH: Path | None = None


def _git_env_for_push() -> dict[str, str]:
    """Build a clean env for a git subprocess that pushes as HAL9000.

    Matches the ``_pr_clone_creds._GIT_ENV_PASSTHROUGH`` allowlist —
    only proxy / SSL / locale / TMPDIR vars carry through; everything
    else is dropped to keep unrelated secrets out of any git hook the
    push triggers.
    """
    askpass = _ensure_askpass_for_hal9000()
    # The worktree credential helper (worker/workspace.py) sources the
    # token from $FORGEJO_TOKEN. Resolve it with the same precedence
    # the askpass shim uses (FORGEJO_PAT → GITEA_TOKEN) so a push
    # authenticates whether it goes through the shim OR the worktree
    # helper. Without FORGEJO_TOKEN the helper sends an empty password
    # and Forgejo rejects it as "Credentials are incorrect or expired".
    forgejo_token = os.environ.get("FORGEJO_PAT") or os.environ.get("GITEA_TOKEN") or os.environ.get("HAL_9000_API_KEY") or ""
    env: dict[str, str] = {
        "PATH": os.environ.get("PATH", "/usr/bin:/bin"),
        "HOME": os.environ.get("HOME", "/tmp"),
        "GIT_ASKPASS": str(askpass),
        "GIT_TERMINAL_PROMPT": "0",
        "FORGEJO_PAT": os.environ.get("FORGEJO_PAT", ""),
        "FORGEJO_TOKEN": forgejo_token,
        "FORGEJO_USERNAME": os.environ.get("FORGEJO_USERNAME", "x-token-auth"),
        "GITEA_TOKEN": os.environ.get("GITEA_TOKEN", ""),
    }
    for name in (
        "SSL_CERT_FILE",
        "SSL_CERT_DIR",
        "GIT_SSL_CAINFO",
        "GIT_SSL_CAPATH",
        "GIT_SSL_NO_VERIFY",
        "http_proxy",
        "https_proxy",
        "no_proxy",
        "HTTP_PROXY",
        "HTTPS_PROXY",
        "NO_PROXY",
        "LANG",
        "LC_ALL",
        "LC_CTYPE",
        "TMPDIR",
        "SSH_AUTH_SOCK",
        "SSH_AGENT_PID",
    ):
        v = os.environ.get(name)
        if v is not None:
            env[name] = v
    return env


def _run_git(
    args: list[str],
    *,
    cwd: Path | None = None,
    env: dict[str, str] | None = None,
    timeout: int = _DEFAULT_GIT_TIMEOUT_S,
) -> tuple[int, str, str]:
    """Run a git subprocess and return ``(returncode, stdout, stderr)``.

    Catches timeouts and FileNotFoundError; returns ``(-1, "",
    error_message)`` in those cases so callers can branch on
    ``returncode`` uniformly.
    """
    cmd = ["git", *args]
    try:
        result = subprocess.run(
            cmd,
            cwd=str(cwd) if cwd is not None else None,
            env=env,
            capture_output=True,
            text=True,
            timeout=timeout,
            check=False,
        )
    except subprocess.TimeoutExpired:
        return -1, "", f"git timed out after {timeout}s: {' '.join(cmd)}"
    except FileNotFoundError:
        return -1, "", "git binary not found on PATH"
    return result.returncode, result.stdout or "", result.stderr or ""


# ``_err`` is the shared :func:`_mcp_common.error_envelope` imported above.


def _rel_path_in_worktree(wt: Path, path: str) -> tuple[str, str | None]:
    """Validate ``path`` is a worktree-relative path that stays inside
    ``wt``.

    Returns ``(normalised_relative_path, None)`` on success or
    ``("", error_message)`` on rejection. Rejects empty / non-string
    paths, absolute paths, and any path that resolves outside the
    worktree — ``..`` traversal OR a symlink escape (``resolve()``
    follows links, then ``relative_to`` confirms containment).
    """
    if not path or not isinstance(path, str):
        return "", "path must be a non-empty string"
    if path.startswith("/"):
        return "", f"path must be worktree-relative, got absolute: {path!r}"
    if ".." in path.split("/"):
        return "", f"path must not contain '..': {path!r}"
    try:
        resolved = (wt / path).resolve()
    except (OSError, RuntimeError) as exc:
        return "", f"could not resolve path {path!r}: {exc}"
    try:
        rel = resolved.relative_to(wt)
    except ValueError:
        return "", f"path {path!r} resolves outside the worktree"
    return str(rel), None


# ─── Tools ─────────────────────────────────────────────────────────


@server.tool()
def isolate(
    pr: int,
    head_sha: str,
    head_ref: str | None = None,
    kind: str = "implementer",
) -> dict[str, Any]:
    """Pre-clone a PR's HEAD SHA into a fresh worktree.

    Wraps ``_pr_clone.prepare_pr_worktree`` which uses a bare mirror
    at ``/tmp/.cleveragents-mirror.git`` plus ``git worktree add``
    for speed. Requires ``IMPLEMENTER_DISPATCHER_PRECLONE=1`` (or
    the review-side equivalent) — same gating as the dispatcher's
    own pre-clone path.

    Parameters:
      pr:       PR number.
      head_sha: the SHA to materialise (caller is responsible for
                fetching this from Forgejo first — typically via the
                forgejo MCP's ``fetch_pr`` and reading ``head.sha``).
      head_ref: optional branch ref (saves a ``git for-each-ref``
                call inside the helper). Passes through verbatim.
      kind:     ``"implementer"`` or ``"review"``. Selects the
                worktree base and gating flag.

    Returns ``{worktree, branch, kind, run_tag}`` on success or
    ``{error: str}`` if the pre-clone is gated off / mirror fetch
    fails / SHA invalid. On gate-off the caller should fall back to
    a plain `clone` operation (not yet implemented as a tool).
    """
    cfg = ForgejoCfg()
    err = require_token(cfg, "default")
    if err:
        return _err(err, pr=pr)
    try:
        handle = _pr_clone.prepare_pr_worktree(
            cfg,
            int(pr),
            head_sha,
            head_ref=head_ref or "",
            kind=kind,
        )
    except Exception as exc:
        return _err(f"prepare_pr_worktree raised: {exc!r}", pr=pr)
    if handle is None:
        return _err(
            "pre-clone returned None (gated off, mirror fetch failed, or "
            "SHA invalid — check dispatcher logs)",
            pr=pr,
        )
    return {
        "worktree": str(handle.path),
        "branch": handle.head_ref or "",
        "kind": handle.kind,
        "run_tag": handle.path.name.rsplit("-", 1)[-1],
    }


# Porcelain XY codes that mark an unmerged (conflicted) path. Both
# ``status`` and the conflict-resolver use this set to enumerate the
# files git considers still in conflict.
_UNMERGED_CODES = frozenset(("UU", "AA", "DD", "AU", "UA", "DU", "UD"))


def _unmerged_paths(cwd: Path) -> list[str]:
    """Worktree paths git reports as unmerged (conflicted), keyed on the
    porcelain XY codes in ``_UNMERGED_CODES``.

    Shared by ``rebase`` / ``rebase_continue`` so their conflict
    detection is identical to ``status`` — the prior ad-hoc
    ``startswith(("U", "AA", "DD"))`` parse missed the ``AU`` / ``UA`` /
    ``DU`` / ``UD`` unmerged codes.
    """
    rc, out, _ = _run_git(["status", "--porcelain=v1", "-z"], cwd=cwd)
    if rc != 0:
        return []
    paths: list[str] = []
    for entry in out.split("\x00"):
        if len(entry) < 4:
            continue
        if entry[:2] in _UNMERGED_CODES:
            paths.append(entry[3:])
    return paths


@server.tool()
def status(worktree: str) -> dict[str, Any]:
    """Compact working-tree status: branch, HEAD SHA, file lists.

    Returns:
      ``{worktree, branch, head_sha, staged, unstaged, untracked,
      rebase_in_progress, merge_in_progress, conflicts}``
      where ``staged`` / ``unstaged`` / ``untracked`` / ``conflicts``
      are lists of file paths relative to the worktree root. Easier for
      the agent than parsing ``git status --porcelain`` itself.

    ``rebase_in_progress`` is True when ``.git/rebase-merge`` or
    ``.git/rebase-apply`` exists (a rebase has stopped, typically on a
    conflict). ``merge_in_progress`` is True when ``.git/MERGE_HEAD``
    exists (a ``git merge`` stopped on a conflict). ``conflicts`` lists
    every path with an unmerged porcelain code (``UU``/``AA``/``DD``/
    ``AU``/``UA``/``DU``/``UD``) — the conflict-resolver uses it to
    enumerate what it still has to resolve in either mode.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    branch_rc, branch_out, _ = _run_git(["rev-parse", "--abbrev-ref", "HEAD"], cwd=wt)
    sha_rc, sha_out, _ = _run_git(["rev-parse", "HEAD"], cwd=wt)
    porcelain_rc, porcelain_out, porcelain_err = _run_git(
        ["status", "--porcelain=v1", "-z"], cwd=wt
    )
    if porcelain_rc != 0:
        return _err(f"git status failed: {porcelain_err.strip()}", worktree=str(wt))
    staged: list[str] = []
    unstaged: list[str] = []
    untracked: list[str] = []
    conflicts: list[str] = []
    # -z: NUL-separated entries; "XY path" where X/Y are status codes
    for entry in porcelain_out.split("\x00"):
        if not entry or len(entry) < 3:
            continue
        x, y, path = entry[0], entry[1], entry[3:]
        if x == "?" and y == "?":
            untracked.append(path)
            continue
        if (x + y) in _UNMERGED_CODES:
            conflicts.append(path)
        if x != " ":
            staged.append(path)
        if y != " ":
            unstaged.append(path)
    git_dir = wt / ".git"
    rebase_in_progress = (git_dir / "rebase-merge").exists() or (
        git_dir / "rebase-apply"
    ).exists()
    merge_in_progress = (git_dir / "MERGE_HEAD").exists()
    return {
        "worktree": str(wt),
        "branch": branch_out.strip() if branch_rc == 0 else "",
        "head_sha": sha_out.strip() if sha_rc == 0 else "",
        "staged": staged,
        "unstaged": unstaged,
        "untracked": untracked,
        "rebase_in_progress": rebase_in_progress,
        "merge_in_progress": merge_in_progress,
        "conflicts": conflicts,
    }


@server.tool()
def stage(worktree: str, paths: list[str]) -> dict[str, Any]:
    """``git add`` the given paths (relative to worktree).

    Returns ``{staged: [...], skipped: [{path, reason}]}``. Paths
    are passed individually so a single bad path doesn't fail the
    whole batch — each one's outcome is reported.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not paths:
        return _err("paths must be a non-empty list", worktree=str(wt))
    staged: list[str] = []
    skipped: list[dict[str, str]] = []
    for p in paths:
        if not p or not isinstance(p, str):
            skipped.append({"path": str(p), "reason": "empty or non-string"})
            continue
        # Reject absolute paths or path traversal that could escape
        # the worktree. ``git add`` itself would reject most of these
        # but a clear error here is friendlier than a cryptic git one.
        if p.startswith("/") or ".." in p.split("/"):
            skipped.append({"path": p, "reason": "absolute or contains '..'"})
            continue
        rc, _, err_out = _run_git(["add", "--", p], cwd=wt)
        if rc != 0:
            skipped.append({"path": p, "reason": err_out.strip()[:200]})
        else:
            staged.append(p)
    return {"staged": staged, "skipped": skipped}


@server.tool()
def commit(
    worktree: str,
    message: str,
    author_name: str | None = None,
    author_email: str | None = None,
) -> dict[str, Any]:
    """``git commit -m`` the staged changes.

    Returns ``{sha}`` on success or ``{error, stderr}`` on failure.
    ``author_name`` / ``author_email`` default to
    ``GIT_USER_NAME`` / ``GIT_USER_EMAIL`` from env which
    ``tools/launch_fork.sh`` sets to ``CleverThis`` /
    ``hal9000@cleverthis.com``. A commit with nothing staged returns
    ``{error: "nothing to commit"}`` rather than the git-default
    error message.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not message or not message.strip():
        return _err("commit message must be non-empty", worktree=str(wt))
    name = author_name or os.environ.get("GIT_USER_NAME") or "CleverThis"
    email = author_email or os.environ.get("GIT_USER_EMAIL") or "hal9000@cleverthis.com"
    env = dict(os.environ)
    env["GIT_AUTHOR_NAME"] = name
    env["GIT_AUTHOR_EMAIL"] = email
    env["GIT_COMMITTER_NAME"] = name
    env["GIT_COMMITTER_EMAIL"] = email
    rc, _, err_out = _run_git(["commit", "-m", message], cwd=wt, env=env)
    if rc != 0:
        lowered = err_out.lower()
        if "nothing to commit" in lowered or "no changes added" in lowered:
            return _err("nothing to commit", worktree=str(wt))
        return _err(f"git commit failed", stderr=err_out.strip(), worktree=str(wt))
    sha_rc, sha_out, _ = _run_git(["rev-parse", "HEAD"], cwd=wt)
    return {"sha": sha_out.strip() if sha_rc == 0 else ""}


@server.tool()
def checkout_conflict(worktree: str, path: str, side: str) -> dict[str, Any]:
    """Resolve ONE conflicted file by taking an entire side, then stage it.

    Runs ``git checkout --ours|--theirs -- <path>`` followed by
    ``git add -- <path>`` so the file ends up fully resolved (working
    tree + index). This is the conflict-resolver's deterministic
    side-selection primitive.

    Use this — an MCP tool, permission-matched by tool NAME (``git*``)
    — rather than the built-in ``edit``/``write`` tools, which are
    matched by a file-path glob that does not reach deep worktree
    paths (the PR-40 incident, 2026-05-21).

    ``side`` is ``"ours"`` or ``"theirs"``. NOTE the rebase sense:
    during ``git rebase`` ``ours`` is the branch being rebased ONTO
    (the base) and ``theirs`` is the PR commit being replayed. The
    caller owns that semantic choice; this tool only passes it on.

    Returns ``{path, side, resolved: true}`` or ``{error, stderr}``.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if side not in ("ours", "theirs"):
        return _err(
            f"side must be 'ours' or 'theirs', got {side!r}", worktree=str(wt)
        )
    rel, perr = _rel_path_in_worktree(wt, path)
    if perr:
        return _err(perr, worktree=str(wt))
    rc, _, err_out = _run_git(["checkout", f"--{side}", "--", rel], cwd=wt)
    if rc != 0:
        return _err(
            "git checkout failed", stderr=err_out.strip(), worktree=str(wt), path=rel
        )
    add_rc, _, add_err = _run_git(["add", "--", rel], cwd=wt)
    if add_rc != 0:
        return _err(
            "git add failed after checkout",
            stderr=add_err.strip(),
            worktree=str(wt),
            path=rel,
        )
    return {"path": rel, "side": side, "resolved": True}


@server.tool()
def write_file(worktree: str, path: str, content: str) -> dict[str, Any]:
    """Write ``content`` to ``path`` (relative to ``worktree``),
    overwriting any existing file and creating parent dirs as needed.

    The conflict-resolver's tool for three-way / manual merges — when
    no single side wins and it must compose the resolved file by hand.

    Use this — an MCP tool, permission-matched by tool NAME (``git*``)
    — rather than the built-in ``write``/``edit`` tools, which are
    matched by a file-path glob that does not reach deep worktree
    paths (the PR-40 incident, 2026-05-21).

    Does NOT stage — call ``git_stage`` afterwards (mirrors the
    write-then-stage protocol). ``path`` must stay inside the worktree;
    absolute paths and ``..`` traversal are rejected.

    Returns ``{path, bytes_written}`` or ``{error}``.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not isinstance(content, str):
        return _err("content must be a string", worktree=str(wt))
    rel, perr = _rel_path_in_worktree(wt, path)
    if perr:
        return _err(perr, worktree=str(wt))
    target = wt / rel
    try:
        target.parent.mkdir(parents=True, exist_ok=True)
        data = content.encode("utf-8")
        target.write_bytes(data)
    except OSError as exc:
        return _err(f"write failed: {exc}", worktree=str(wt), path=rel)
    return {"path": rel, "bytes_written": len(data)}


@server.tool()
def push(
    worktree: str,
    remote: str = "origin",
    force_with_lease: bool = False,
) -> dict[str, Any]:
    """``git push`` the current branch.

    Authenticates as HAL9000 via the MCP's own askpass shim (does
    NOT use the reviewer PAT). Returns ``{remote_sha, branch}`` on
    success or ``{error, stderr}`` on failure.

    ``force_with_lease=True`` uses ``--force-with-lease`` which is
    safe against concurrent pushes (refuses to overwrite if the
    remote moved since the worker last fetched). Plain ``--force``
    is intentionally NOT supported — it's the kind of thing whose
    blast radius justifies a human in the loop.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not os.environ.get("FORGEJO_PAT") and not os.environ.get("GITEA_TOKEN"):
        return _err(
            "FORGEJO_PAT / GITEA_TOKEN not set in MCP environment — push "
            "would prompt for credentials (which terminal_prompt=0 refuses)."
        )
    branch_rc, branch_out, _ = _run_git(["rev-parse", "--abbrev-ref", "HEAD"], cwd=wt)
    if branch_rc != 0 or not branch_out.strip():
        return _err("could not resolve current branch", worktree=str(wt))
    branch = branch_out.strip()
    if branch == "HEAD":
        return _err(
            "worktree is in detached HEAD state — checkout a branch first",
            worktree=str(wt),
        )
    env = _git_env_for_push()
    # Refresh the remote-tracking ref BEFORE pushing. Without this,
    # ``--force-with-lease`` (no value form) uses whatever
    # ``refs/remotes/{remote}/{branch}`` was at the start of the
    # cycle, which is usually stale by the time the worker reaches
    # the push step — leading to spurious
    # ``! [rejected] (stale info)`` failures even though there is no
    # actual concurrent push to race against. The fetch + explicit
    # ``--force-with-lease=<ref>:<sha>`` form pin the lease to the
    # tip we just observed, eliminating the false-rejection class.
    # Live-confirmed on 2026-05-16 (PR #30): five bash-path pushes
    # all failed with "stale remote state info" while the worker
    # made no other progress; this MCP path is engineered to avoid
    # that mode.
    # Explicit refspec form (``+<branch>:refs/remotes/origin/<branch>``)
    # is critical here: the bare ``git fetch origin <branch>`` shape
    # would update the local ``<branch>`` if it exists, and git
    # REFUSES that when the local ``<branch>`` is currently
    # checked out ("fatal: refusing to fetch into branch ... checked
    # out at ..."). The explicit-refspec form only touches the
    # remote-tracking ref, never the local branch, so it works
    # regardless of checkout state. The leading ``+`` is the
    # force-update marker — without it, a force-push on the remote
    # since our last sync would fail the fetch with non-fast-forward.
    # Live-confirmed root cause on 2026-05-16 (run-16 PR #29): the
    # bare-shape fetch failed silently, MCP fell through to bare
    # ``--force-with-lease`` with a stale tracking ref, push
    # rejected stale-info. Worker burned ~10 min troubleshooting.
    fetch_rc, _, fetch_err = _run_git(
        ["fetch", remote, f"+{branch}:refs/remotes/{remote}/{branch}"],
        cwd=wt,
        env=env,
        timeout=60,
    )
    if fetch_rc != 0:
        # Non-fatal: pre-fetch failed (network / auth blip). Push
        # with the bare ``--force-with-lease`` form as a fallback;
        # the lease is stale but the agent at least sees a real
        # rejection if there's a genuine race.
        lease_sha = ""
    else:
        rt_rc, rt_out, _ = _run_git(
            ["rev-parse", f"{remote}/{branch}"], cwd=wt, env=env, timeout=15
        )
        lease_sha = rt_out.strip() if rt_rc == 0 else ""
    cmd = ["push"]
    if force_with_lease:
        if lease_sha:
            cmd.append(f"--force-with-lease=refs/heads/{branch}:{lease_sha}")
        else:
            cmd.append("--force-with-lease")
    cmd += [remote, branch]
    rc, _, err_out = _run_git(cmd, cwd=wt, env=env)
    if rc != 0:
        # One-shot retry for the specific "stale info" class —
        # another fetch + a fresh lease often clears it (the remote
        # may have advanced between our pre-fetch and the push).
        lower = (err_out or "").lower()
        if force_with_lease and "stale info" in lower:
            _run_git(["fetch", remote, branch], cwd=wt, env=env, timeout=60)
            rt_rc, rt_out, _ = _run_git(
                ["rev-parse", f"{remote}/{branch}"],
                cwd=wt,
                env=env,
                timeout=15,
            )
            retry_lease = rt_out.strip() if rt_rc == 0 else ""
            retry_cmd = ["push"]
            if retry_lease:
                retry_cmd.append(
                    f"--force-with-lease=refs/heads/{branch}:{retry_lease}"
                )
            else:
                retry_cmd.append("--force-with-lease")
            retry_cmd += [remote, branch]
            rc, _, err_out = _run_git(retry_cmd, cwd=wt, env=env)
        if rc != 0:
            return _err(
                "git push failed",
                stderr=(err_out or "").strip()[:1000],
                worktree=str(wt),
                branch=branch,
            )
    # Read the remote-tracking ref after the push to confirm what landed.
    rt_rc, rt_out, _ = _run_git(["rev-parse", f"{remote}/{branch}"], cwd=wt, env=env)
    return {
        "remote_sha": rt_out.strip() if rt_rc == 0 else "",
        "branch": branch,
        "remote": remote,
    }


@server.tool()
def fetch(
    worktree: str,
    remote: str = "origin",
    branch: str | None = None,
) -> dict[str, Any]:
    """``git fetch`` from ``remote``. Authenticates as HAL9000.

    Parameters:
      remote: defaults to ``origin``.
      branch: optional. When provided, fetches ONLY this branch using
        the explicit refspec ``+<branch>:refs/remotes/<remote>/<branch>``
        so the operation works regardless of whether the local branch
        by that name is currently checked out. (The bare
        ``git fetch origin <branch>`` shape would refuse with "fatal:
        refusing to fetch into branch ... checked out at ..." in the
        checked-out case — same bug that bit the ``push`` MCP's
        internal pre-fetch on 2026-05-16, run-16 PR #29.) When
        ``branch`` is omitted the fetch updates every remote-tracking
        ref the default refspec covers.

    Returns ``{ok: bool}`` on success or ``{error, stderr}`` on
    failure. Output (which is usually noise like ``From https://...``)
    is not returned; if you need it, run ``status`` afterward.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    env = _git_env_for_push()
    if branch:
        cmd = [
            "fetch",
            remote,
            f"+{branch}:refs/remotes/{remote}/{branch}",
        ]
    else:
        cmd = ["fetch", remote]
    rc, _, err_out = _run_git(cmd, cwd=wt, env=env)
    if rc != 0:
        return _err(
            "git fetch failed",
            stderr=err_out.strip(),
            worktree=str(wt),
        )
    return {"ok": True}


@server.tool()
def checkout(
    worktree: str,
    branch: str,
    create: bool = False,
    force: bool = False,
) -> dict[str, Any]:
    """``git checkout`` — switch to (or create) a branch.

    Parameters:
      branch: target branch name.
      create: when True, uses ``-B`` to create-or-reset the branch
        at the current HEAD. This is the common pattern for converting
        a detached HEAD (left by the dispatcher's pre-clone
        ``worktree add --detach`` flow) back into a named branch
        before pushing. Without this, ``git push`` rejects with
        "src refspec HEAD does not match any" / "you are not
        currently on a branch."
      force: when True, uses ``-f`` to discard local changes that
        would otherwise block the checkout. Use sparingly — silent
        data loss risk if the worker had uncommitted work.

    Returns ``{branch, head_sha}`` on success or ``{error, stderr}``
    on failure.

    The 2026-05-16 (run-16 PR #29) trace showed task-implementor
    repeatedly dropping to bash for ``git checkout -B <branch>``
    because no MCP tool existed for it. This tool closes that gap
    so the MCP path covers the full git workflow.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not branch or not branch.strip():
        return _err("branch must be non-empty", worktree=str(wt))
    cmd = ["checkout"]
    if create:
        cmd.append("-B")
    elif force:
        cmd.append("-f")
    cmd.append(str(branch))
    rc, _, err_out = _run_git(cmd, cwd=wt)
    if rc != 0:
        return _err(
            "git checkout failed",
            stderr=err_out.strip(),
            worktree=str(wt),
            branch=branch,
        )
    # Report the post-checkout state so the caller can verify they
    # ended up on the expected branch + SHA.
    sha_rc, sha_out, _ = _run_git(["rev-parse", "HEAD"], cwd=wt)
    return {
        "branch": branch,
        "head_sha": sha_out.strip() if sha_rc == 0 else "",
    }


@server.tool()
def rebase(worktree: str, onto: str) -> dict[str, Any]:
    """``git rebase`` the current branch onto ``onto`` (e.g.
    ``origin/master``).

    Returns ``{success: True}`` on a clean rebase or ``{success:
    False, conflicts: [path, …]}`` when the rebase stops on a
    conflict — the worktree is left in the rebase-in-progress state
    so the agent can resolve conflicts and then call the
    ``rebase_continue`` tool to resume.

    On non-conflict errors returns ``{error, stderr}``.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not onto or not onto.strip():
        return _err("`onto` ref must be non-empty", worktree=str(wt))
    rc, _, err_out = _run_git(["rebase", onto], cwd=wt)
    if rc == 0:
        return {"success": True}
    # Conflict detection: a stopped rebase exits non-zero and leaves
    # unmerged paths in the index (``_unmerged_paths`` / ``_UNMERGED_CODES``
    # — the same set ``status`` reports).
    conflicts = _unmerged_paths(wt)
    if conflicts:
        return {
            "success": False,
            "conflicts": conflicts,
            "stderr": err_out.strip()[:500],
        }
    return _err(f"git rebase failed", stderr=err_out.strip(), worktree=str(wt))


@server.tool()
def rebase_continue(worktree: str) -> dict[str, Any]:
    """``git rebase --continue`` after the conflicts for the current
    rebase step have been resolved + staged.

    Runs with ``GIT_EDITOR=true`` so git never opens an interactive
    editor (a stopped rebase that resumes would otherwise block on a
    commit-message editor, hanging the MCP call).

    Returns ``{success: True}`` when the rebase ran to completion, or
    ``{success: False, conflicts: [path, …]}`` when it advanced and
    stopped on the NEXT step's conflicts — same shape + conflict
    detection as the ``rebase`` tool, so the agent loops
    resolve → ``rebase_continue`` until ``success`` is True.

    On a non-conflict failure (e.g. nothing staged so git refuses to
    continue, or no rebase in progress) returns ``{error, stderr}``.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    env = dict(os.environ)
    env["GIT_EDITOR"] = "true"
    rc, _, err_out = _run_git(["rebase", "--continue"], cwd=wt, env=env)
    if rc == 0:
        return {"success": True}
    # The rebase advanced and stopped on the NEXT step's conflicts:
    # same detection as the ``rebase`` tool — git exits non-zero and
    # leaves unmerged paths in the index.
    conflicts = _unmerged_paths(wt)
    if conflicts:
        return {
            "success": False,
            "conflicts": conflicts,
            "stderr": err_out.strip()[:500],
        }
    return _err(
        "git rebase --continue failed", stderr=err_out.strip(), worktree=str(wt)
    )


@server.tool()
def cleanup(worktree: str) -> dict[str, Any]:
    """Remove a worktree.

    Runs ``git worktree remove --force`` against the mirror so git's
    metadata stays consistent, then deletes the directory if any
    files remain. Returns ``{removed: True}`` on success or
    ``{error}`` on failure. Safe to call on an already-removed
    worktree.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        # If the directory doesn't exist that's actually success for
        # cleanup — the post-condition (path is gone) already holds.
        if "does not exist" in (err or ""):
            return {"removed": True, "note": "already absent"}
        return _err(err)
    # Best-effort: ask the mirror to deregister the worktree. The
    # mirror path follows the dispatcher's convention.
    mirror = Path("/tmp/.cleveragents-mirror.git")
    if mirror.is_dir():
        _run_git(
            ["--git-dir", str(mirror), "worktree", "remove", "--force", str(wt)],
        )
    # Whatever the worktree-remove result, scrub any remaining files
    # so the agent sees a clean post-condition.
    try:
        shutil.rmtree(wt, ignore_errors=True)
    except OSError as exc:
        return _err(f"rmtree failed: {exc}", worktree=str(wt))
    return {"removed": True}


# ─── Read-only inspection tools (2026-05-16, Step 4) ───────────────
# Added so the git MCP covers the full surface that the soon-to-be-
# retired ``git-*-util`` subagent fleet supported. Without these,
# callers migrating off ``git-commit-util`` / ``git-rebase-util`` /
# etc. would still need bash perms for ``git log`` / ``git diff`` /
# ``git show`` / ``git rev-parse`` / ``git merge-base`` to verify
# the state of their work — keeping the bash dependency the MCP
# migration was supposed to eliminate. All five are read-only with
# no side effects beyond stdout/stderr.


@server.tool()
def log(
    worktree: str,
    range: str | None = None,
    max_count: int = 20,
    oneline: bool = True,
    paths: list[str] | None = None,
) -> dict[str, Any]:
    """``git log`` — list commits. Defaults to oneline format capped
    at 20 commits.

    Parameters:
      range: e.g. ``"master..HEAD"`` (commits in HEAD not in master),
        ``"HEAD~5..HEAD"``, or omit for full history.
      max_count: ``-n`` flag; capped at 200 to bound output.
      oneline: ``--oneline`` (default). Set False for the full
        commit-message format.
      paths: limit log to commits touching these paths (e.g.
        ``["src/foo.py", "tests/test_foo.py"]``).

    Returns ``{output: str}`` on success or ``{error, stderr}`` on
    failure.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    cmd = ["log"]
    if oneline:
        cmd.append("--oneline")
    n = max(1, min(int(max_count), 200))
    cmd += ["-n", str(n)]
    if range:
        cmd.append(str(range))
    if paths:
        cmd.append("--")
        cmd += [str(p) for p in paths if p]
    rc, out, err_out = _run_git(cmd, cwd=wt)
    if rc != 0:
        return _err("git log failed", stderr=err_out.strip(), worktree=str(wt))
    return {"output": out}


@server.tool()
def diff(
    worktree: str,
    ref1: str | None = None,
    ref2: str | None = None,
    paths: list[str] | None = None,
    name_only: bool = False,
    stat: bool = False,
) -> dict[str, Any]:
    """``git diff`` between refs or against the working tree.

    Parameters:
      ref1 / ref2: omit both → diff working tree vs HEAD. Specify
        ref1 only → diff working tree vs ref1. Specify both →
        ``git diff ref1 ref2`` (use ``"master...HEAD"`` syntax in
        ref1 for triple-dot semantics).
      paths: limit diff to these paths.
      name_only: ``--name-only`` (list of changed files, no hunks).
      stat: ``--stat`` (insertions/deletions summary, no hunks).

    Returns ``{output: str}`` on success or ``{error, stderr}`` on
    failure. ``name_only`` and ``stat`` are mutually exclusive; if
    both are True ``stat`` wins.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    cmd = ["diff"]
    if stat:
        cmd.append("--stat")
    elif name_only:
        cmd.append("--name-only")
    if ref1:
        cmd.append(str(ref1))
    if ref2:
        cmd.append(str(ref2))
    if paths:
        cmd.append("--")
        cmd += [str(p) for p in paths if p]
    rc, out, err_out = _run_git(cmd, cwd=wt)
    if rc != 0:
        return _err("git diff failed", stderr=err_out.strip(), worktree=str(wt))
    return {"output": out}


@server.tool()
def show(
    worktree: str,
    ref: str,
    path: str | None = None,
) -> dict[str, Any]:
    """``git show`` a commit or a file's content at a specific ref.

    Without ``path``: shows the commit (message + diff) at ``ref``.
    With ``path``: shows the file content at ``ref:path`` (e.g.
    ``ref="HEAD"`` ``path="README.md"`` returns README at HEAD).

    Returns ``{output: str}`` on success or ``{error, stderr}`` on
    failure. Useful for reading the file as it existed in a previous
    commit without checking out.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not ref or not ref.strip():
        return _err("ref must be non-empty", worktree=str(wt))
    target = f"{ref}:{path}" if path else ref
    rc, out, err_out = _run_git(["show", target], cwd=wt)
    if rc != 0:
        return _err("git show failed", stderr=err_out.strip(), worktree=str(wt))
    return {"output": out}


@server.tool()
def rev_parse(
    worktree: str,
    ref: str,
    abbrev_ref: bool = False,
) -> dict[str, Any]:
    """``git rev-parse`` — resolve a ref to a SHA (or to its
    abbreviated-ref / branch name).

    Parameters:
      ref: anything git can parse — ``"HEAD"``, ``"master"``,
        ``"origin/main"``, ``"HEAD~3"``, a tag, a partial SHA.
      abbrev_ref: ``--abbrev-ref`` — return the branch name a ref
        points to (e.g. ``rev_parse("HEAD", abbrev_ref=True)`` →
        ``"feature/x"``) instead of the SHA. ``git_status`` already
        returns the branch name; this flag exists for the rarer
        case of resolving an arbitrary ref's branch.

    Returns ``{sha: str}`` (or ``{branch: str}`` when abbrev_ref) on
    success, or ``{error, stderr}`` on failure.
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not ref or not ref.strip():
        return _err("ref must be non-empty", worktree=str(wt))
    cmd = ["rev-parse"]
    if abbrev_ref:
        cmd.append("--abbrev-ref")
    cmd.append(str(ref))
    rc, out, err_out = _run_git(cmd, cwd=wt)
    if rc != 0:
        return _err("git rev-parse failed", stderr=err_out.strip(), worktree=str(wt))
    value = out.strip()
    return {"branch": value} if abbrev_ref else {"sha": value}


@server.tool()
def merge_base(
    worktree: str,
    ref1: str,
    ref2: str,
) -> dict[str, Any]:
    """``git merge-base`` — find the common ancestor SHA of two refs.

    Used to compute the "PR fork point" (``merge_base("HEAD",
    "origin/master")``) for diff-base resolution and for verifying a
    rebase landed on the expected base.

    Returns ``{sha: str}`` on success or ``{error, stderr}`` on
    failure (the most common failure is unrelated histories — e.g.
    you compared two refs that share no common commit).
    """
    wt, err = _validate_worktree(worktree)
    if err:
        return _err(err)
    if not ref1 or not ref2:
        return _err("both ref1 and ref2 must be non-empty", worktree=str(wt))
    rc, out, err_out = _run_git(["merge-base", str(ref1), str(ref2)], cwd=wt)
    if rc != 0:
        return _err(
            "git merge-base failed",
            stderr=err_out.strip(),
            worktree=str(wt),
        )
    return {"sha": out.strip()}


main = make_main(server, "mcp_git_server")


if __name__ == "__main__":
    sys.exit(main())