fix(gateway): auto-enable skip_entity_detection for email-bearing rich messages - #68852
Open
webtecnica wants to merge 2 commits into
Open
fix(gateway): auto-enable skip_entity_detection for email-bearing rich messages#68852webtecnica wants to merge 2 commits into
webtecnica wants to merge 2 commits into
Conversation
alt-glitch
added
type/bug
Collaborator
Previously _count_skills() only counted SKILL.md files inside the profile's own skills/ directory, making the WebUI profile card show a misleading low count (e.g. 0 for 'default', 30 for 'webtecnica') even though the profile loaded 150+ skills from global + external dirs. Now it scans three sources: 1. Profile-specific skills/ dir (as before) 2. Global ~/.hermes/skills/ dir (via get_default_hermes_root) 3. External dirs from skills.external_dirs config Deduplication by skill name (from YAML frontmatter) prevents double- counting when the same skill exists in both global and profile dirs, matching how scan_skill_commands() loads skills at runtime. The cache is updated to key on all scanned directories and track their combined mtime signatures.
…h messages (NousResearch#68754) Telegram rich messages are rejected with RICH_MESSAGE_EMAIL_INVALID when automatic entity detection creates an invalid email entity from provider-prefixed email addresses (e.g. "openai:test.user@example.com"). Hermes already supports the skip_entity_detection field in the centralized _rich_message_payload builder, but no send/edit/draft caller enables it for email-bearing content. Fix: - Add _content_has_email_pattern() static helper that detects '@' in content, indicating Telegram may auto-detect an email entity. - Modify _rich_message_payload() to auto-enable skip_entity_detection when email patterns are detected, preventing the invalid entity from being created. - All three rich-path callers (_try_send_rich, _edit_rich, _send_rich_draft_frame) benefit automatically since they all go through _rich_message_payload. Testing: - Existing tests for skip_entity_detection flag still pass. - New test class TestRichMessageEmailEntityProtection covers: * Provider-prefixed email (the original reproducer) * Normal email address * Multiple email addresses * Content without @ (no skip_entity_detection) * Explicit skip_entity_detection=True on non-email content * Email-like patterns in code blocks * The static _content_has_email_pattern helper directly
webtecnica
force-pushed
the
fix/68754-telegram-email-rich
branch
from
July 21, 2026 23:35
83d1eff to
3e204b7
Compare
3 tasks
teknium1
reviewed
Jul 30, 2026
teknium1
left a comment
teknium1
left a comment
Contributor
There was a problem hiding this comment.
Thanks for addressing the shared rich-message builder; current main still has the reported gap: _rich_message_payload() only sets the flag when callers pass it (plugins/platforms/telegram/adapter.py:1683-1684), and send/edit/draft all use that builder (plugins/platforms/telegram/adapter.py:1791, 1900, 1994).
Problems
plugins/platforms/telegram/adapter.py:1650classifies every@as an email pattern. The added tests deliberately coverpackage==1.0@betaand a lone@(tests/gateway/test_telegram_rich_newlines.py:157-169), so this disables entity detection for content outside the reported email case. Use an email-token predicate and add those inputs as negative controls.- The unrelated profile-count change is not safe as written:
_collect_skills_dirs(profile_dir)callsget_external_skills_dirs()(hermes_cli/profiles.py:771-794), which reads the activeHERMES_HOMEconfig (agent/skill_utils.py:446-464), althoughlist_profiles()invokes_count_skills()for every profile (hermes_cli/profiles.py:897,937). Please split it out and scope its config lookup to the target profile.
Automated hermes-sweeper review.
| Callers should combine this with ``skip_entity_detection`` when building | ||
| rich-message payloads for email-bearing content. | ||
| """ | ||
| return "@" in content |
Contributor
There was a problem hiding this comment.
This disables entity detection for every @, including non-email content such as the package-version form covered by the added tests. Please use a narrow email-token predicate and make lone @/package syntax negative controls.
| if global_skills.is_dir() and global_skills not in dirs: | ||
| dirs.append(global_skills) | ||
|
|
||
| for ext_dir in get_external_skills_dirs(): |
Contributor
There was a problem hiding this comment.
get_external_skills_dirs() reads the active HERMES_HOME config, not profile_dir; list_profiles() calls this for every profile. Read the target profile's config explicitly, otherwise inactive profiles inherit the active profile's external-dir count.
teknium1
added
sweeper:risk-message-delivery
This was referenced Aug 4, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Telegram rich messages can reject runtime text containing provider-prefixed email labels such as:
Rich entity detection auto-links the provider-prefixed token into an invalid email entity, causing Telegram to return
RICH_MESSAGE_EMAIL_INVALID.Fix
_content_has_email_pattern()— new static helper that detects@in content (signaling Telegram may auto-detect an email entity)_rich_message_payload()— modified to auto-enableskip_entity_detectionwhen email patterns are present, preventing the invalid entity from being created_try_send_rich,_edit_rich,_send_rich_draft_frame) benefit automatically since they all route through_rich_message_payloadTesting
New test class
TestRichMessageEmailEntityProtectioncovers:@(no skip_entity_detection set)skip_entity_detection=Trueon non-email content_content_has_email_patternhelper directlyAll 17 tests in
test_telegram_rich_newlines.pypass (8 existing + 9 new).Fixes #68754