Bookmarks Skill
Triggered by /save ... messages or bookmark-related requests.
Commands
All bookmark operations use a single Telegram command with sub-parameters:
| /save help | handler.py help | Show command usage |
| /save show # | handler.py show <id> | Display specific bookmark |
| /save rm # | handler.py rm <id> | Remove a bookmark |
Commands
All bookmark operations use a single Telegram command with sub-parameters:
| Input | Maps to | Action |
|---|---|---|
/save (as reply) |
bookmark.py save |
Save replied-to message |
/save <text> |
bookmark.py save --content "text" |
Save inline text |
/save list [count] |
bookmark.py list --count N |
List recent bookmarks |
/save search <query> |
bookmark.py search "query" |
Search bookmarks |
/save show <#> |
bookmark.py show <id> |
Show specific bookmark |
/save rm <#> |
bookmark.py rm <id> |
Remove bookmark |
/save count |
bookmark.py json | jq length |
Count total bookmarks |
/save tags |
Parse tags from all entries | List unique tags |
/save help |
— | Show usage help |
Scripts
scripts/bookmark.py— CLI for save/list/search/json (workspace root)scripts/handler.py— Normalizes/save ...into bookmark.py calls
Workflow
- Detect
/savemessage (standalone or reply). - If replying to assistant message: fetch full text from session history to avoid truncation, pass via
--full-content. - Parse sub-command and arguments.
- Run appropriate
bookmark.pysubcommand via exec. - Return formatted results to user.
Save behavior
- Reply to assistant message: retrieve full response from session history, save complete text via
--full-contentoverride - Reply to user/other message: capture the replied-to message text as content (may be truncated if long — recommend inline
/save <text>for long messages) - Inline mode: use the text after
/saveas content - Auto-tags: extract from content using keyword heuristics
- Manual tags: user can append
#tag1 #tag2to note
Full-Content Retrieval
When replying /save to an assistant (cloudybot) message, the skill retrieves the full assistant response from session history instead of using the truncated context snippet. This is done by:
- Detecting
[reply target] cloudybotin the context - Fetching session history via
sessions_history - Matching the assistant message by its truncated prefix
- Passing the full text via
handler.py save-reply --full-content "..."
This ensures long assistant responses are saved completely, not cut off at the truncation boundary.
Storage
memory/bookmarks.md— append-only markdown log- Entry format:
## Bookmark #N — ISO timestampwith metadata fields