mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-04-27 01:11:40 +00:00
feat: devex help, add Makefile, ruff, pre-commit, and modernize CI
This commit is contained in:
Brooklyn Nicholson
parent
172a38c344
commit
f4d7e6a29e
111 changed files with 11655 additions and 10200 deletions
|
|
@ -31,7 +31,7 @@ import time
|
|||
import uuid
|
||||
|
||||
_IS_WINDOWS = platform.system() == "Windows"
|
||||
from typing import Any, Dict, List, Optional
|
||||
from typing import Any
|
||||
|
||||
# Availability gate: UDS requires a POSIX OS
|
||||
logger = logging.getLogger(__name__)
|
||||
|
|
@ -40,21 +40,23 @@ SANDBOX_AVAILABLE = sys.platform != "win32"
|
|||
|
||||
# The 7 tools allowed inside the sandbox. The intersection of this list
|
||||
# and the session's enabled tools determines which stubs are generated.
|
||||
SANDBOX_ALLOWED_TOOLS = frozenset([
|
||||
"web_search",
|
||||
"web_extract",
|
||||
"read_file",
|
||||
"write_file",
|
||||
"search_files",
|
||||
"patch",
|
||||
"terminal",
|
||||
])
|
||||
SANDBOX_ALLOWED_TOOLS = frozenset(
|
||||
[
|
||||
"web_search",
|
||||
"web_extract",
|
||||
"read_file",
|
||||
"write_file",
|
||||
"search_files",
|
||||
"patch",
|
||||
"terminal",
|
||||
]
|
||||
)
|
||||
|
||||
# Resource limit defaults (overridable via config.yaml → code_execution.*)
|
||||
DEFAULT_TIMEOUT = 300 # 5 minutes
|
||||
DEFAULT_TIMEOUT = 300 # 5 minutes
|
||||
DEFAULT_MAX_TOOL_CALLS = 50
|
||||
MAX_STDOUT_BYTES = 50_000 # 50 KB
|
||||
MAX_STDERR_BYTES = 10_000 # 10 KB
|
||||
MAX_STDOUT_BYTES = 50_000 # 50 KB
|
||||
MAX_STDERR_BYTES = 10_000 # 10 KB
|
||||
|
||||
|
||||
def check_sandbox_requirements() -> bool:
|
||||
|
|
@ -114,7 +116,7 @@ _TOOL_STUBS = {
|
|||
}
|
||||
|
||||
|
||||
def generate_hermes_tools_module(enabled_tools: List[str]) -> str:
|
||||
def generate_hermes_tools_module(enabled_tools: list[str]) -> str:
|
||||
"""
|
||||
Build the source code for the hermes_tools.py stub module.
|
||||
|
||||
|
|
@ -128,11 +130,7 @@ def generate_hermes_tools_module(enabled_tools: List[str]) -> str:
|
|||
if tool_name not in _TOOL_STUBS:
|
||||
continue
|
||||
func_name, sig, doc, args_expr = _TOOL_STUBS[tool_name]
|
||||
stub_functions.append(
|
||||
f"def {func_name}({sig}):\n"
|
||||
f" {doc}\n"
|
||||
f" return _call({func_name!r}, {args_expr})\n"
|
||||
)
|
||||
stub_functions.append(f"def {func_name}({sig}):\n {doc}\n return _call({func_name!r}, {args_expr})\n")
|
||||
export_names.append(func_name)
|
||||
|
||||
header = '''\
|
||||
|
|
@ -223,7 +221,7 @@ def _rpc_server_loop(
|
|||
server_sock: socket.socket,
|
||||
task_id: str,
|
||||
tool_call_log: list,
|
||||
tool_call_counter: list, # mutable [int] so the thread can increment
|
||||
tool_call_counter: list, # mutable [int] so the thread can increment
|
||||
max_tool_calls: int,
|
||||
allowed_tools: frozenset,
|
||||
):
|
||||
|
|
@ -243,7 +241,7 @@ def _rpc_server_loop(
|
|||
while True:
|
||||
try:
|
||||
chunk = conn.recv(65536)
|
||||
except socket.timeout:
|
||||
except TimeoutError:
|
||||
break
|
||||
if not chunk:
|
||||
break
|
||||
|
|
@ -270,23 +268,22 @@ def _rpc_server_loop(
|
|||
# Enforce the allow-list
|
||||
if tool_name not in allowed_tools:
|
||||
available = ", ".join(sorted(allowed_tools))
|
||||
resp = json.dumps({
|
||||
"error": (
|
||||
f"Tool '{tool_name}' is not available in execute_code. "
|
||||
f"Available: {available}"
|
||||
)
|
||||
})
|
||||
resp = json.dumps(
|
||||
{"error": (f"Tool '{tool_name}' is not available in execute_code. Available: {available}")}
|
||||
)
|
||||
conn.sendall((resp + "\n").encode())
|
||||
continue
|
||||
|
||||
# Enforce tool call limit
|
||||
if tool_call_counter[0] >= max_tool_calls:
|
||||
resp = json.dumps({
|
||||
"error": (
|
||||
f"Tool call limit reached ({max_tool_calls}). "
|
||||
"No more tool calls allowed in this execution."
|
||||
)
|
||||
})
|
||||
resp = json.dumps(
|
||||
{
|
||||
"error": (
|
||||
f"Tool call limit reached ({max_tool_calls}). "
|
||||
"No more tool calls allowed in this execution."
|
||||
)
|
||||
}
|
||||
)
|
||||
conn.sendall((resp + "\n").encode())
|
||||
continue
|
||||
|
||||
|
|
@ -303,9 +300,7 @@ def _rpc_server_loop(
|
|||
sys.stdout = open(os.devnull, "w")
|
||||
sys.stderr = open(os.devnull, "w")
|
||||
try:
|
||||
result = handle_function_call(
|
||||
tool_name, tool_args, task_id=task_id
|
||||
)
|
||||
result = handle_function_call(tool_name, tool_args, task_id=task_id)
|
||||
finally:
|
||||
sys.stdout.close()
|
||||
sys.stderr.close()
|
||||
|
|
@ -318,15 +313,17 @@ def _rpc_server_loop(
|
|||
|
||||
# Log for observability
|
||||
args_preview = str(tool_args)[:80]
|
||||
tool_call_log.append({
|
||||
"tool": tool_name,
|
||||
"args_preview": args_preview,
|
||||
"duration": round(call_duration, 2),
|
||||
})
|
||||
tool_call_log.append(
|
||||
{
|
||||
"tool": tool_name,
|
||||
"args_preview": args_preview,
|
||||
"duration": round(call_duration, 2),
|
||||
}
|
||||
)
|
||||
|
||||
conn.sendall((result + "\n").encode())
|
||||
|
||||
except socket.timeout:
|
||||
except TimeoutError:
|
||||
pass
|
||||
except OSError:
|
||||
pass
|
||||
|
|
@ -342,10 +339,11 @@ def _rpc_server_loop(
|
|||
# Main entry point
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
|
||||
def execute_code(
|
||||
code: str,
|
||||
task_id: Optional[str] = None,
|
||||
enabled_tools: Optional[List[str]] = None,
|
||||
task_id: str | None = None,
|
||||
enabled_tools: list[str] | None = None,
|
||||
) -> str:
|
||||
"""
|
||||
Run a Python script in a sandboxed child process with RPC access
|
||||
|
|
@ -361,9 +359,7 @@ def execute_code(
|
|||
JSON string with execution results.
|
||||
"""
|
||||
if not SANDBOX_AVAILABLE:
|
||||
return json.dumps({
|
||||
"error": "execute_code is not available on Windows. Use normal tool calls instead."
|
||||
})
|
||||
return json.dumps({"error": "execute_code is not available on Windows. Use normal tool calls instead."})
|
||||
|
||||
if not code or not code.strip():
|
||||
return json.dumps({"error": "No code provided."})
|
||||
|
|
@ -397,9 +393,7 @@ def execute_code(
|
|||
|
||||
try:
|
||||
# Write the auto-generated hermes_tools module
|
||||
tools_src = generate_hermes_tools_module(
|
||||
list(sandbox_tools) if enabled_tools else list(SANDBOX_ALLOWED_TOOLS)
|
||||
)
|
||||
tools_src = generate_hermes_tools_module(list(sandbox_tools) if enabled_tools else list(SANDBOX_ALLOWED_TOOLS))
|
||||
with open(os.path.join(tmpdir, "hermes_tools.py"), "w") as f:
|
||||
f.write(tools_src)
|
||||
|
||||
|
|
@ -415,8 +409,12 @@ def execute_code(
|
|||
rpc_thread = threading.Thread(
|
||||
target=_rpc_server_loop,
|
||||
args=(
|
||||
server_sock, task_id, tool_call_log,
|
||||
tool_call_counter, max_tool_calls, sandbox_tools,
|
||||
server_sock,
|
||||
task_id,
|
||||
tool_call_log,
|
||||
tool_call_counter,
|
||||
max_tool_calls,
|
||||
sandbox_tools,
|
||||
),
|
||||
daemon=True,
|
||||
)
|
||||
|
|
@ -426,11 +424,24 @@ def execute_code(
|
|||
# Build a minimal environment for the child. We intentionally exclude
|
||||
# API keys and tokens to prevent credential exfiltration from LLM-
|
||||
# generated scripts. The child accesses tools via RPC, not direct API.
|
||||
_SAFE_ENV_PREFIXES = ("PATH", "HOME", "USER", "LANG", "LC_", "TERM",
|
||||
"TMPDIR", "TMP", "TEMP", "SHELL", "LOGNAME",
|
||||
"XDG_", "PYTHONPATH", "VIRTUAL_ENV", "CONDA")
|
||||
_SECRET_SUBSTRINGS = ("KEY", "TOKEN", "SECRET", "PASSWORD", "CREDENTIAL",
|
||||
"PASSWD", "AUTH")
|
||||
_SAFE_ENV_PREFIXES = (
|
||||
"PATH",
|
||||
"HOME",
|
||||
"USER",
|
||||
"LANG",
|
||||
"LC_",
|
||||
"TERM",
|
||||
"TMPDIR",
|
||||
"TMP",
|
||||
"TEMP",
|
||||
"SHELL",
|
||||
"LOGNAME",
|
||||
"XDG_",
|
||||
"PYTHONPATH",
|
||||
"VIRTUAL_ENV",
|
||||
"CONDA",
|
||||
)
|
||||
_SECRET_SUBSTRINGS = ("KEY", "TOKEN", "SECRET", "PASSWORD", "CREDENTIAL", "PASSWD", "AUTH")
|
||||
child_env = {}
|
||||
for k, v in os.environ.items():
|
||||
if any(s in k.upper() for s in _SECRET_SUBSTRINGS):
|
||||
|
|
@ -515,7 +526,7 @@ def execute_code(
|
|||
rpc_thread.join(timeout=3)
|
||||
|
||||
# Build response
|
||||
result: Dict[str, Any] = {
|
||||
result: dict[str, Any] = {
|
||||
"status": status,
|
||||
"output": stdout_text,
|
||||
"tool_calls_made": tool_call_counter[0],
|
||||
|
|
@ -538,17 +549,21 @@ def execute_code(
|
|||
except Exception as exc:
|
||||
duration = round(time.monotonic() - exec_start, 2)
|
||||
logging.exception("execute_code failed")
|
||||
return json.dumps({
|
||||
"status": "error",
|
||||
"error": str(exc),
|
||||
"tool_calls_made": tool_call_counter[0],
|
||||
"duration_seconds": duration,
|
||||
}, ensure_ascii=False)
|
||||
return json.dumps(
|
||||
{
|
||||
"status": "error",
|
||||
"error": str(exc),
|
||||
"tool_calls_made": tool_call_counter[0],
|
||||
"duration_seconds": duration,
|
||||
},
|
||||
ensure_ascii=False,
|
||||
)
|
||||
|
||||
finally:
|
||||
# Cleanup temp dir and socket
|
||||
try:
|
||||
import shutil
|
||||
|
||||
shutil.rmtree(tmpdir, ignore_errors=True)
|
||||
except Exception as e:
|
||||
logger.debug("Could not clean temp dir: %s", e)
|
||||
|
|
@ -592,6 +607,7 @@ def _load_config() -> dict:
|
|||
"""Load code_execution config from CLI_CONFIG if available."""
|
||||
try:
|
||||
from cli import CLI_CONFIG
|
||||
|
||||
return CLI_CONFIG.get("code_execution", {})
|
||||
except Exception:
|
||||
return {}
|
||||
|
|
@ -604,27 +620,37 @@ def _load_config() -> dict:
|
|||
# Per-tool documentation lines for the execute_code description.
|
||||
# Ordered to match the canonical display order.
|
||||
_TOOL_DOC_LINES = [
|
||||
("web_search",
|
||||
" web_search(query: str, limit: int = 5) -> dict\n"
|
||||
" Returns {\"data\": {\"web\": [{\"url\", \"title\", \"description\"}, ...]}}"),
|
||||
("web_extract",
|
||||
" web_extract(urls: list[str]) -> dict\n"
|
||||
" Returns {\"results\": [{\"url\", \"title\", \"content\", \"error\"}, ...]} where content is markdown"),
|
||||
("read_file",
|
||||
" read_file(path: str, offset: int = 1, limit: int = 500) -> dict\n"
|
||||
" Lines are 1-indexed. Returns {\"content\": \"...\", \"total_lines\": N}"),
|
||||
("write_file",
|
||||
" write_file(path: str, content: str) -> dict\n"
|
||||
" Always overwrites the entire file."),
|
||||
("search_files",
|
||||
" search_files(pattern: str, target=\"content\", path=\".\", file_glob=None, limit=50) -> dict\n"
|
||||
" target: \"content\" (search inside files) or \"files\" (find files by name). Returns {\"matches\": [...]}"),
|
||||
("patch",
|
||||
" patch(path: str, old_string: str, new_string: str, replace_all: bool = False) -> dict\n"
|
||||
" Replaces old_string with new_string in the file."),
|
||||
("terminal",
|
||||
" terminal(command: str, timeout=None, workdir=None) -> dict\n"
|
||||
" Foreground only (no background/pty). Returns {\"output\": \"...\", \"exit_code\": N}"),
|
||||
(
|
||||
"web_search",
|
||||
" web_search(query: str, limit: int = 5) -> dict\n"
|
||||
' Returns {"data": {"web": [{"url", "title", "description"}, ...]}}',
|
||||
),
|
||||
(
|
||||
"web_extract",
|
||||
" web_extract(urls: list[str]) -> dict\n"
|
||||
' Returns {"results": [{"url", "title", "content", "error"}, ...]} where content is markdown',
|
||||
),
|
||||
(
|
||||
"read_file",
|
||||
" read_file(path: str, offset: int = 1, limit: int = 500) -> dict\n"
|
||||
' Lines are 1-indexed. Returns {"content": "...", "total_lines": N}',
|
||||
),
|
||||
("write_file", " write_file(path: str, content: str) -> dict\n Always overwrites the entire file."),
|
||||
(
|
||||
"search_files",
|
||||
' search_files(pattern: str, target="content", path=".", file_glob=None, limit=50) -> dict\n'
|
||||
' target: "content" (search inside files) or "files" (find files by name). Returns {"matches": [...]}',
|
||||
),
|
||||
(
|
||||
"patch",
|
||||
" patch(path: str, old_string: str, new_string: str, replace_all: bool = False) -> dict\n"
|
||||
" Replaces old_string with new_string in the file.",
|
||||
),
|
||||
(
|
||||
"terminal",
|
||||
" terminal(command: str, timeout=None, workdir=None) -> dict\n"
|
||||
' Foreground only (no background/pty). Returns {"output": "...", "exit_code": N}',
|
||||
),
|
||||
]
|
||||
|
||||
|
||||
|
|
@ -639,9 +665,7 @@ def build_execute_code_schema(enabled_sandbox_tools: set = None) -> dict:
|
|||
enabled_sandbox_tools = SANDBOX_ALLOWED_TOOLS
|
||||
|
||||
# Build tool documentation lines for only the enabled tools
|
||||
tool_lines = "\n".join(
|
||||
doc for name, doc in _TOOL_DOC_LINES if name in enabled_sandbox_tools
|
||||
)
|
||||
tool_lines = "\n".join(doc for name, doc in _TOOL_DOC_LINES if name in enabled_sandbox_tools)
|
||||
|
||||
# Build example import list from enabled tools
|
||||
import_examples = [n for n in ("web_search", "terminal") if n in enabled_sandbox_tools]
|
||||
|
|
@ -702,8 +726,7 @@ registry.register(
|
|||
toolset="code_execution",
|
||||
schema=EXECUTE_CODE_SCHEMA,
|
||||
handler=lambda args, **kw: execute_code(
|
||||
code=args.get("code", ""),
|
||||
task_id=kw.get("task_id"),
|
||||
enabled_tools=kw.get("enabled_tools")),
|
||||
code=args.get("code", ""), task_id=kw.get("task_id"), enabled_tools=kw.get("enabled_tools")
|
||||
),
|
||||
check_fn=check_sandbox_requirements,
|
||||
)
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue