mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-05-24 05:41:40 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 6

feat(discord): allow_any_attachment config to accept arbitrary file types

Browse source 
The Discord adapter silently dropped any attachment whose extension wasn't
in the SUPPORTED_DOCUMENT_TYPES allowlist (PDF, text family, zip, office).
Users uploading .wav / .bin / other unrecognized formats saw nothing in
their conversation — the file got logged as 'Unsupported document type'
and discarded before the agent ever saw it.

Add discord.allow_any_attachment (default false) to bypass the allowlist.
When on:
  - Any file is downloaded, cached under ~/.hermes/cache/documents/, and
    surfaced as a DOCUMENT-typed event with application/octet-stream MIME
  - gateway/run.py already emits a context note with the cached path,
    auto-translated via to_agent_visible_cache_path() for Docker/Modal
    sandboxed terminals
  - File body is NOT inlined — only the path — so binary uploads don't
    blow up the context window
  - Allowlisted text formats (.txt/.md/.log) keep their 100 KiB inline
    behavior unchanged

Also adds discord.max_attachment_bytes (default 32 MiB matches the
historical hardcoded cap; 0 = unlimited) since users opting into arbitrary
types may want to raise the cap. The whole attachment is held in memory
while being cached, so unlimited carries a real memory cost.

Env overrides: DISCORD_ALLOW_ANY_ATTACHMENT, DISCORD_MAX_ATTACHMENT_BYTES.

Discord-only by deliberate scope. Telegram has hard 20 MB API limits and
Slack has its own caps — extending the same flag there is a separate
follow-up if/when requested.
This commit is contained in:
teknium1 2026-05-16 20:25:47 -07:00 • committed by Teknium
parent 3b39096904
commit 407a11b419
5 changed files with 258 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

88
gateway/platforms/discord.py
View file

@ -3564,6 +3564,43 @@ class DiscordAdapter(BasePlatformAdapter):
return bool(configured)
return os.getenv("DISCORD_REQUIRE_MENTION", "true").lower() not in {"false", "0", "no", "off"}
def _discord_allow_any_attachment(self) -> bool:
"""Return whether Discord attachments bypass the SUPPORTED_DOCUMENT_TYPES allowlist.
When True, any uploaded file is cached to disk and surfaced to the
agent as a local path so it can be inspected via terminal / read_file
/ ffprobe / etc. Default False preserves the historical behaviour of
dropping unsupported types with a warning log.
"""
configured = self.config.extra.get("allow_any_attachment")
if configured is not None:
if isinstance(configured, str):
return configured.lower() not in {"false", "0", "no", "off", ""}
return bool(configured)
return os.getenv("DISCORD_ALLOW_ANY_ATTACHMENT", "false").lower() in {"true", "1", "yes", "on"}
def _discord_max_attachment_bytes(self) -> int:
"""Return the per-attachment byte cap. 0 means unlimited.
The whole attachment is held in memory while being written to the
cache, so unlimited carries a real memory cost. Default 32 MiB
matches the historical hardcoded value.
"""
configured = self.config.extra.get("max_attachment_bytes")
if configured is None:
configured = os.getenv("DISCORD_MAX_ATTACHMENT_BYTES")
if configured is None or configured == "":
return 32 * 1024 * 1024
try:
value = int(configured)
except (TypeError, ValueError):
logger.warning(
"[Discord] Invalid max_attachment_bytes value %r, falling back to 32 MiB",
configured,
)
return 32 * 1024 * 1024
return max(0, value)
def _discord_free_response_channels(self) -> set:
"""Return Discord channel IDs where no bot mention is required.
@ -4495,6 +4532,7 @@ class DiscordAdapter(BasePlatformAdapter):
if normalized_content.startswith("/"):
msg_type = MessageType.COMMAND
elif all_attachments:
_allow_any = self._discord_allow_any_attachment()
# Check attachment types
for att in all_attachments:
if att.content_type:
@ -4509,9 +4547,15 @@ class DiscordAdapter(BasePlatformAdapter):
if att.filename:
_, doc_ext = os.path.splitext(att.filename)
doc_ext = doc_ext.lower()
if doc_ext in SUPPORTED_DOCUMENT_TYPES:
if doc_ext in SUPPORTED_DOCUMENT_TYPES or _allow_any:
msg_type = MessageType.DOCUMENT
break
elif _allow_any:
# No content_type at all (rare — discord usually fills it
# in). Treat as a document so downstream pipelines surface
# the path to the agent.
msg_type = MessageType.DOCUMENT
break
# When auto-threading kicked in, route responses to the new thread
effective_channel = auto_threaded_channel or message.channel
@ -4594,31 +4638,48 @@ class DiscordAdapter(BasePlatformAdapter):
if not ext and content_type:
mime_to_ext = {v: k for k, v in SUPPORTED_DOCUMENT_TYPES.items()}
ext = mime_to_ext.get(content_type, "")
if ext not in SUPPORTED_DOCUMENT_TYPES:
allow_any_attachment = self._discord_allow_any_attachment()
in_allowlist = ext in SUPPORTED_DOCUMENT_TYPES
if not in_allowlist and not allow_any_attachment:
logger.warning(
"[Discord] Unsupported document type '%s' (%s), skipping",
ext or "unknown", content_type,
)
else:
MAX_DOC_BYTES = 32 * 1024 * 1024
if att.size and att.size > MAX_DOC_BYTES:
max_doc_bytes = self._discord_max_attachment_bytes()
if max_doc_bytes and att.size and att.size > max_doc_bytes:
logger.warning(
"[Discord] Document too large (%s bytes), skipping: %s",
att.size, att.filename,
"[Discord] Document too large (%s bytes > cap %s), skipping: %s",
att.size, max_doc_bytes, att.filename,
)
else:
try:
raw_bytes = await self._cache_discord_document(att, ext)
cached_path = cache_document_from_bytes(
raw_bytes, att.filename or f"document{ext}"
raw_bytes, att.filename or f"document{ext or '.bin'}"
)
doc_mime = SUPPORTED_DOCUMENT_TYPES[ext]
if in_allowlist:
doc_mime = SUPPORTED_DOCUMENT_TYPES[ext]
else:
# allow_any_attachment path: untyped file. Use the
# source content_type if discord gave us one,
# otherwise fall back to octet-stream so the agent
# knows it's binary and reaches for terminal tools.
doc_mime = (
content_type
if content_type and content_type != "unknown"
else "application/octet-stream"
)
media_urls.append(cached_path)
media_types.append(doc_mime)
logger.info("[Discord] Cached user document: %s", cached_path)
logger.info(
"[Discord] Cached user %s: %s",
"document" if in_allowlist else "attachment",
cached_path,
)
# Inject text content for plain-text documents (capped at 100 KB)
MAX_TEXT_INJECT_BYTES = 100 * 1024
if ext in {".md", ".txt", ".log"} and len(raw_bytes) <= MAX_TEXT_INJECT_BYTES:
if in_allowlist and ext in {".md", ".txt", ".log"} and len(raw_bytes) <= MAX_TEXT_INJECT_BYTES:
try:
text_content = raw_bytes.decode("utf-8")
display_name = att.filename or f"document{ext}"
@ -4630,6 +4691,13 @@ class DiscordAdapter(BasePlatformAdapter):
pending_text_injection = injection
except UnicodeDecodeError:
pass
# NOTE: for the allow_any_attachment path we deliberately
# do NOT inject a path string here. ``gateway/run.py``
# already detects DOCUMENT-typed events with
# ``application/octet-stream`` MIME and emits a context
# note with the sandbox-translated cache path via
# ``to_agent_visible_cache_path()`` (important for
# Docker/Modal terminal backends).
except Exception as e:
logger.warning(
"[Discord] Failed to cache document %s: %s",