diff --git a/.gitignore b/.gitignore index e6fe104..9fd03b9 100644 --- a/.gitignore +++ b/.gitignore @@ -10,3 +10,7 @@ gitea-mcp*.json .vscode/ graphify-out/ branches/ +# Throwaway agent commit-encoding helpers (#261) — never commit. +/_encode_*.py +/_emit_*.py +/_inline_*.py diff --git a/agent_temp_artifacts.py b/agent_temp_artifacts.py new file mode 100644 index 0000000..5020521 --- /dev/null +++ b/agent_temp_artifacts.py @@ -0,0 +1,26 @@ +"""Detect throwaway agent helper scripts left in the repo root (#261).""" + +from __future__ import annotations + +import fnmatch + +AGENT_TEMP_BASENAME_PATTERNS = ( + "_encode_*.py", + "_emit_*.py", + "_inline_*.py", +) + + +def find_agent_temp_artifacts_from_porcelain(porcelain: str) -> list[str]: + """Return untracked repo-root helper paths matching agent temp patterns.""" + found: list[str] = [] + for line in (porcelain or "").splitlines(): + if not line.startswith("??"): + continue + path = line[3:].strip() + if not path or "/" in path or "\\" in path: + continue + basename = path.split("/")[-1] + if any(fnmatch.fnmatch(basename, pat) for pat in AGENT_TEMP_BASENAME_PATTERNS): + found.append(path) + return sorted(found) \ No newline at end of file diff --git a/docs/llm-workflow-runbooks.md b/docs/llm-workflow-runbooks.md index 1b04cb3..25a11a9 100644 --- a/docs/llm-workflow-runbooks.md +++ b/docs/llm-workflow-runbooks.md @@ -316,6 +316,30 @@ may edit another issue's branch folder unless explicitly assigned to that issue. No LLM may clean another issue's branch folder unless the PR is merged or closed and cleanup is explicitly part of the task. +## Agent temp artifact cleanup (#261) + +Failed or aborted MCP commit attempts sometimes leave throwaway helper scripts in +the **repository root**. These are not part of any issue scope and pollute +`git status`, which can break `gitea_lock_issue` and preflight checks. + +**Patterns (repo root only, untracked):** + +- `_encode_*.py` — base64 payload encoders +- `_emit_*.py` — commit payload emitters +- `_inline_*.py` — inline encoding helpers + +**Required cleanup (after MCP commit completes or aborts):** + +1. Delete any matching files at the repo root (`rm ./_encode_*.py` etc.). +2. Confirm `git status` is clean on the orchestration checkout before + `gitea_lock_issue`. +3. Prefer native `gitea_commit_files` / gated commit paths — do not leave shell + encoding fallbacks behind. + +Root-level matches are listed in `.gitignore` so they never get committed. +`gitea_get_runtime_context` and `gitea_lock_issue` surface **warnings** (not +hard blocks) when these artifacts are still present. + Implementation work and review work must use separate branch folders. For example, an implementation branch might live under `branches/fix-issue-123-example`, while a review branch for the resulting PR diff --git a/gitea_mcp_server.py b/gitea_mcp_server.py index 60e65fb..46353fb 100644 --- a/gitea_mcp_server.py +++ b/gitea_mcp_server.py @@ -260,9 +260,19 @@ def assess_preflight_status() -> dict: "Reviewer profile modified tracked workspace files after capability resolution " f"(offending files: {_format_preflight_files(_preflight_reviewer_violation_files)})" ) + warnings: list[str] = [] + agent_artifacts = agent_temp_artifacts.find_agent_temp_artifacts_from_porcelain( + _get_workspace_porcelain() + ) + if agent_artifacts: + warnings.append( + "Agent temp artifacts detected at repo root (delete before mutations): " + f"{_format_preflight_files(agent_artifacts)}" + ) return { "preflight_ready": not reasons, "preflight_block_reasons": reasons, + "preflight_warnings": warnings, "preflight_whoami_verified": _preflight_whoami_called, "preflight_capability_resolved": _preflight_capability_called, "preflight_whoami_violation_files": list(_preflight_whoami_violation_files), @@ -373,6 +383,7 @@ import role_session_router # noqa: E402 import role_namespace_gate # noqa: E402 import task_capability_map # noqa: E402 import review_proofs # noqa: E402 +import agent_temp_artifacts import issue_lock_worktree # noqa: E402 @@ -877,7 +888,10 @@ def gitea_lock_issue( except Exception as e: raise RuntimeError(f"Could not write issue lock file: {e}") - return { + agent_artifacts = agent_temp_artifacts.find_agent_temp_artifacts_from_porcelain( + git_state.get("porcelain_status") or "" + ) + result = { "success": True, "message": ( f"Successfully locked issue #{issue_number} to branch '{branch_name}' " @@ -887,6 +901,12 @@ def gitea_lock_issue( "branch_name": branch_name, "worktree_path": resolved_worktree, } + if agent_artifacts: + result["warnings"] = [ + "Agent temp artifacts at repo root (delete before implementation): " + + ", ".join(agent_artifacts) + ] + return result @mcp.tool() @@ -2275,6 +2295,106 @@ def gitea_get_file( } +def _prepare_commit_payload_files(files: list[dict]) -> tuple[list[dict], list[dict]]: + import base64 + import json + import os + + processed_files = [] + source_proofs = [] + + lock_data = {} + if os.path.exists(ISSUE_LOCK_FILE): + try: + with open(ISSUE_LOCK_FILE, "r", encoding="utf-8") as f: + lock_data = json.load(f) + except Exception: + pass + + locked_worktree = lock_data.get("worktree_path") + if locked_worktree: + locked_worktree = os.path.realpath(locked_worktree) + + for f in files: + f_copy = dict(f) + path = f_copy.get("path", "") + + content_keys = [k for k in ["content", "content_plain", "workspace_path", "local_path"] if k in f_copy] + if len(content_keys) > 1: + raise ValueError( + f"Multiple content sources specified for file '{path}'; provide exactly one of " + f"'content', 'content_plain', 'workspace_path', or 'local_path' (fail closed)" + ) + + source = "none" + if "content_plain" in f_copy: + content_plain = f_copy.pop("content_plain") + if content_plain is not None: + content_bytes = content_plain.encode("utf-8") + f_copy["content"] = base64.b64encode(content_bytes).decode("utf-8") + source = "inline_plain" + elif "workspace_path" in f_copy: + workspace_path = f_copy.pop("workspace_path") + if not locked_worktree: + raise RuntimeError( + f"Issue lock is missing or does not define a worktree path. " + f"Cannot resolve workspace_path '{workspace_path}' (fail closed)." + ) + target_path = os.path.realpath(os.path.abspath(os.path.join(locked_worktree, workspace_path))) + + is_inside = target_path == locked_worktree or target_path.startswith(os.path.join(locked_worktree, "")) + is_tmp = target_path.startswith(os.path.realpath("/tmp") + os.sep) or target_path.startswith("/tmp" + os.sep) + + if not (is_inside or is_tmp): + raise ValueError( + f"workspace_path '{workspace_path}' resolves to '{target_path}' which falls outside " + f"of locked worktree '{locked_worktree}' (fail closed)" + ) + + if not os.path.exists(target_path): + raise FileNotFoundError(f"File not found: {target_path} (fail closed)") + + with open(target_path, "rb") as fh: + file_bytes = fh.read() + f_copy["content"] = base64.b64encode(file_bytes).decode("utf-8") + source = "workspace_path" + elif "local_path" in f_copy: + local_path = f_copy.pop("local_path") + if not locked_worktree: + raise RuntimeError( + f"Issue lock is missing or does not define a worktree path. " + f"Cannot resolve local_path '{local_path}' (fail closed)." + ) + target_path = os.path.realpath(os.path.abspath(local_path)) + + is_inside = target_path == locked_worktree or target_path.startswith(os.path.join(locked_worktree, "")) + is_tmp = target_path.startswith(os.path.realpath("/tmp") + os.sep) or target_path.startswith("/tmp" + os.sep) + + if not (is_inside or is_tmp): + raise ValueError( + f"local_path '{local_path}' resolves to '{target_path}' which falls outside " + f"of locked worktree '{locked_worktree}' (fail closed)" + ) + + if not os.path.exists(target_path): + raise FileNotFoundError(f"File not found: {target_path} (fail closed)") + + with open(target_path, "rb") as fh: + file_bytes = fh.read() + f_copy["content"] = base64.b64encode(file_bytes).decode("utf-8") + source = "local_path" + elif "content" in f_copy: + source = "inline_base64" + + processed_files.append(f_copy) + source_proofs.append({ + "path": path, + "source": source + }) + + return processed_files, source_proofs + + @mcp.tool() def gitea_commit_files( files: list[dict], @@ -2289,7 +2409,7 @@ def gitea_commit_files( """Commit changes to multiple files in a Gitea repository in a single atomic commit. Args: - files: List of file operations. Each file dict must contain 'operation' ('create', 'update', 'delete', 'rename'), 'path', and 'content' (base64 encoded for create/update), and optionally 'sha' (required for update/delete) or 'from_path' (for rename). + files: List of file operations. Each file dict must contain 'operation' ('create', 'update', 'delete', 'rename'), 'path', and one of content payload sources: 'content' (base64), 'content_plain', 'workspace_path', or 'local_path'. message: The commit message. branch: Optional existing branch to start/commit from. new_branch: Optional new branch name to create for this commit. @@ -2325,12 +2445,14 @@ def gitea_commit_files( return blocked verify_preflight_purity(remote) + processed_files, source_proofs = _prepare_commit_payload_files(files) + h, o, r = _resolve(remote, host, org, repo) auth = _auth(h) url = f"{repo_api_url(h, o, r)}/contents" payload = { - "files": files, + "files": processed_files, "message": message, } if branch is not None: @@ -2341,13 +2463,14 @@ def gitea_commit_files( with _audited("commit_files", host=h, remote=remote, org=o, repo=r, target_branch=(new_branch or branch), request_metadata={"message": message, - "paths": [f.get("path") for f in files], - "operations": [f.get("operation") for f in files]}): + "paths": [f.get("path") for f in processed_files], + "operations": [f.get("operation") for f in processed_files]}): data = api_request("POST", url, auth, payload) return { "success": True, "commit": data.get("commit", {}).get("sha", ""), "branch": data.get("branch", {}).get("name", ""), + "content_source_proof": source_proofs, } diff --git a/tests/test_agent_temp_artifacts.py b/tests/test_agent_temp_artifacts.py new file mode 100644 index 0000000..2e14385 --- /dev/null +++ b/tests/test_agent_temp_artifacts.py @@ -0,0 +1,87 @@ +"""Tests for agent temp artifact detection and preflight warnings (#261).""" +import json +import os +import sys +import tempfile +import unittest +from unittest.mock import patch + +sys.path.insert(0, str(__import__("pathlib").Path(__file__).resolve().parent.parent)) + +import agent_temp_artifacts +import mcp_server + + +class TestAgentTempArtifactPatterns(unittest.TestCase): + def test_detects_root_level_encode_emit_inline(self): + porcelain = ( + "?? _encode_commit_payload.py\n" + "?? _emit_payload.py\n" + "?? _inline_b64.py\n" + ) + self.assertEqual( + agent_temp_artifacts.find_agent_temp_artifacts_from_porcelain(porcelain), + ["_emit_payload.py", "_encode_commit_payload.py", "_inline_b64.py"], + ) + + def test_ignores_nested_and_unrelated_untracked(self): + porcelain = ( + "?? scripts/_encode_x.py\n" + "?? README-draft.md\n" + " M task_capability_map.py\n" + ) + self.assertEqual( + agent_temp_artifacts.find_agent_temp_artifacts_from_porcelain(porcelain), + [], + ) + + +class TestPreflightWarnings(unittest.TestCase): + def setUp(self): + self._snapshot = ( + mcp_server._preflight_whoami_called, + mcp_server._preflight_capability_called, + ) + mcp_server._preflight_whoami_called = True + mcp_server._preflight_capability_called = True + + def tearDown(self): + mcp_server._preflight_whoami_called, mcp_server._preflight_capability_called = ( + self._snapshot + ) + + @patch( + "mcp_server._get_workspace_porcelain", + return_value="?? _encode_commit_payload.py\n", + ) + def test_assess_preflight_surfaces_warning_not_block(self, _porcelain): + status = mcp_server.assess_preflight_status() + self.assertTrue(status["preflight_ready"]) + self.assertEqual(status["preflight_block_reasons"], []) + self.assertEqual(len(status["preflight_warnings"]), 1) + self.assertIn("_encode_commit_payload.py", status["preflight_warnings"][0]) + + +class TestIssueLockArtifactWarning(unittest.TestCase): + @patch("mcp_server.api_get_all", return_value=[]) + @patch("mcp_server._auth", return_value="token x") + @patch("mcp_server._resolve", return_value=("h", "o", "r")) + @patch("mcp_server.ISSUE_LOCK_FILE", new_callable=lambda: tempfile.mktemp()) + @patch("issue_lock_worktree.read_worktree_git_state") + def test_lock_success_includes_artifact_warning(self, mock_state, _lock_file, *_mocks): + mock_state.return_value = { + "current_branch": "master", + "porcelain_status": "?? _emit_payload.py\n", + } + result = mcp_server.gitea_lock_issue( + issue_number=261, + branch_name="docs/issue-261-agent-artifact-cleanup", + remote="prgs", + ) + self.assertTrue(result["success"]) + self.assertIn("warnings", result) + self.assertIn("_emit_payload.py", result["warnings"][0]) + + +if __name__ == "__main__": + unittest.main() \ No newline at end of file diff --git a/tests/test_commit_payloads.py b/tests/test_commit_payloads.py new file mode 100644 index 0000000..1b3a80d --- /dev/null +++ b/tests/test_commit_payloads.py @@ -0,0 +1,251 @@ +import json +import os +import tempfile +import unittest +from unittest.mock import MagicMock, patch + +import gitea_config +import mcp_server +import task_capability_map + +CONFIG = { + "version": 2, + "contexts": { + "ctx": { + "enabled": True, + "gitea": {"enabled": True, "base_url": "https://gitea.example.com"}, + } + }, + "profiles": { + "full-author": { + "enabled": True, + "context": "ctx", + "role": "author", + "username": "author-user", + "auth": {"type": "env", "name": "GITEA_TOKEN_AUTHOR"}, + "allowed_operations": [ + "gitea.read", + "gitea.repo.commit", + "gitea.pr.create", + "gitea.branch.push", + ], + "forbidden_operations": [ + "gitea.pr.approve", + "gitea.pr.merge", + "gitea.pr.review", + ], + "execution_profile": "full-author", + }, + }, +} + + +class TestCommitPayloads(unittest.TestCase): + def setUp(self): + self._remotes = patch.dict(mcp_server.REMOTES, { + "prgs": {"host": "gitea.example.com", "org": "Example-Org", + "repo": "Example-Repo"}, + }) + self._remotes.start() + mcp_server._IDENTITY_CACHE.clear() + + # Setup temp directories and lock files + self._dir = tempfile.TemporaryDirectory() + self.config_path = os.path.join(self._dir.name, "profiles.json") + with open(self.config_path, "w", encoding="utf-8") as fh: + fh.write(json.dumps(CONFIG)) + + self.locked_worktree_dir = tempfile.TemporaryDirectory() + self.locked_worktree_path = os.path.realpath(self.locked_worktree_dir.name) + + self.lock_file_path = "/tmp/gitea_issue_lock.json" + self.lock_data = { + "issue_number": 263, + "branch_name": "feat/issue-263-native-commit-payloads", + "remote": "prgs", + "org": "Example-Org", + "repo": "Example-Repo", + "worktree_path": self.locked_worktree_path, + } + with open(self.lock_file_path, "w", encoding="utf-8") as fh: + fh.write(json.dumps(self.lock_data)) + + # Reset preflight status to bypass/pass verification in tests + self.orig_whoami_called = mcp_server._preflight_whoami_called + self.orig_capability_called = mcp_server._preflight_capability_called + mcp_server._preflight_whoami_called = True + mcp_server._preflight_capability_called = True + + def tearDown(self): + self._remotes.stop() + mcp_server._IDENTITY_CACHE.clear() + + mcp_server._preflight_whoami_called = self.orig_whoami_called + mcp_server._preflight_capability_called = self.orig_capability_called + + self._dir.cleanup() + self.locked_worktree_dir.cleanup() + if os.path.exists(self.lock_file_path): + os.remove(self.lock_file_path) + + def _env(self, profile: str) -> dict: + return { + "GITEA_MCP_CONFIG": self.config_path, + "GITEA_MCP_PROFILE": profile, + "GITEA_TOKEN_AUTHOR": "author-pass", + "GITEA_TEST_PORCELAIN": "", + } + + @patch("mcp_server.api_request") + @patch("mcp_server.get_auth_header", return_value="token author-pass") + def test_commit_files_content_plain(self, _auth, mock_api): + mock_api.return_value = { + "commit": {"sha": "sha-123"}, + "branch": {"name": "feat/issue-263-native-commit-payloads"} + } + with patch.dict(os.environ, self._env("full-author"), clear=True): + res = mcp_server.gitea_commit_files( + files=[ + { + "operation": "create", + "path": "hello.txt", + "content_plain": "Hello World!", + } + ], + message="Add hello.txt", + remote="prgs", + ) + print("DEBUG RES IS:", res) + self.assertTrue(res["success"]) + self.assertEqual(res["commit"], "sha-123") + self.assertEqual(res["content_source_proof"][0]["source"], "inline_plain") + + # Verify posted base64 content + post_args = mock_api.call_args[0] + payload = post_args[3] + self.assertEqual(payload["files"][0]["content"], "SGVsbG8gV29ybGQh") + + @patch("mcp_server.api_request") + @patch("mcp_server.get_auth_header", return_value="token author-pass") + def test_commit_files_workspace_path(self, _auth, mock_api): + mock_api.return_value = { + "commit": {"sha": "sha-456"}, + "branch": {"name": "feat/issue-263-native-commit-payloads"} + } + + # Create a workspace file + file_relative = "src/code.py" + file_abs = os.path.join(self.locked_worktree_path, file_relative) + os.makedirs(os.path.dirname(file_abs), exist_ok=True) + with open(file_abs, "wb") as fh: + fh.write(b"print('hello')") + + with patch.dict(os.environ, self._env("full-author"), clear=True): + res = mcp_server.gitea_commit_files( + files=[ + { + "operation": "create", + "path": "src/code.py", + "workspace_path": file_relative, + } + ], + message="Add code.py", + remote="prgs", + ) + self.assertTrue(res["success"]) + self.assertEqual(res["content_source_proof"][0]["source"], "workspace_path") + + post_args = mock_api.call_args[0] + payload = post_args[3] + # "print('hello')" in base64 is cHJpbnQoJ2hlbGxvJyk= + self.assertEqual(payload["files"][0]["content"], "cHJpbnQoJ2hlbGxvJyk=") + + @patch("mcp_server.api_request") + @patch("mcp_server.get_auth_header", return_value="token author-pass") + def test_commit_files_local_path(self, _auth, mock_api): + mock_api.return_value = { + "commit": {"sha": "sha-789"}, + "branch": {"name": "feat/issue-263-native-commit-payloads"} + } + + file_abs = os.path.join(self.locked_worktree_path, "local.bin") + with open(file_abs, "wb") as fh: + fh.write(b"\x00\x01\x02\x03\xff") + + with patch.dict(os.environ, self._env("full-author"), clear=True): + res = mcp_server.gitea_commit_files( + files=[ + { + "operation": "create", + "path": "local.bin", + "local_path": file_abs, + } + ], + message="Add local.bin", + remote="prgs", + ) + self.assertTrue(res["success"]) + self.assertEqual(res["content_source_proof"][0]["source"], "local_path") + + post_args = mock_api.call_args[0] + payload = post_args[3] + # b"\x00\x01\x02\x03\xff" in base64 is AAECA/8= + self.assertEqual(payload["files"][0]["content"], "AAECA/8=") + + @patch("mcp_server.api_request") + @patch("mcp_server.get_auth_header", return_value="token author-pass") + def test_commit_files_traversal_blocked(self, _auth, mock_api): + # Remove active lock file to ensure it fails on traversal/invalid locks + os.remove(self.lock_file_path) + + with patch.dict(os.environ, self._env("full-author"), clear=True): + with self.assertRaises(RuntimeError) as ctx: + mcp_server.gitea_commit_files( + files=[ + { + "operation": "create", + "path": "hack.txt", + "workspace_path": "any.txt", + } + ], + message="Add hack", + remote="prgs", + ) + self.assertIn("Issue lock is missing", str(ctx.exception)) + + @patch("mcp_server.api_request") + @patch("mcp_server.get_auth_header", return_value="token author-pass") + def test_commit_files_outside_scope_blocked(self, _auth, mock_api): + with patch.dict(os.environ, self._env("full-author"), clear=True): + with self.assertRaises(ValueError) as ctx: + mcp_server.gitea_commit_files( + files=[ + { + "operation": "create", + "path": "hack.txt", + "workspace_path": "../outside.txt", + } + ], + message="Add hack", + remote="prgs", + ) + self.assertIn("falls outside of locked worktree", str(ctx.exception)) + + @patch("mcp_server.api_request") + @patch("mcp_server.get_auth_header", return_value="token author-pass") + def test_commit_files_multiple_sources_blocked(self, _auth, mock_api): + with patch.dict(os.environ, self._env("full-author"), clear=True): + with self.assertRaises(ValueError) as ctx: + mcp_server.gitea_commit_files( + files=[ + { + "operation": "create", + "path": "hello.txt", + "content_plain": "Hello!", + "content": "SGVsbG8=", + } + ], + message="Add hello", + remote="prgs", + ) + self.assertIn("Multiple content sources specified", str(ctx.exception))