WebSearch Overriding Weather
have brave search and clawdhub weather and skills.md. But when i am doing what is the weather at <location>? It's doing web fetch with wrong url instead of the one described in the skills. Technically using the brave search api instead of weather. I checked the skills are eligible . This is the context breakdown
Context breakdown (detailed)
- AGENTS.md: OK | raw 7,804 chars (~1,951 tok) | injected 7,804 chars (~1,951 tok)
- TOOLS.md: OK | raw 850 chars (~213 tok) | injected 850 chars (~213 tok)
Skills list (system prompt text): 2,052 chars (~513 tok) (6 skills)
Skills: discord, healthcheck, skill-creator, sonoscli, tmux, weather
Top skills (prompt entry size):
- healthcheck: 491 chars (~123 tok)
- skill-creator: 296 chars (~74 tok)
- tmux: 255 chars (~64 tok)
- discord: 215 chars (~54 tok)
- sonoscli: 207 chars (~52 tok)
- weather: 202 chars (~51 tok)
Top tools (schema size):
- message: 3,885 chars (~972 tok)
- browser: 1,897 chars (~475 tok)
- nodes: 1,479 chars (~370 tok)
- exec: 1,086 chars (~272 tok)
- process: 961 chars (~241 tok)
- web_search: 753 chars (~189 tok)
- cron: 662 chars (~166 tok)
- canvas: 661 chars (~166 tok)
- edit: 591 chars (~148 tok)
- gateway: 465 chars (~117 tok)
- read: 392 chars (~98 tok)
- web_fetch: 374 chars (~94 tok)
- sessions_spawn: 336 chars (~84 tok)
- write: 313 chars (~79 tok)
- sessions_send: 273 chars (~69 tok)
- tts: 223 chars (~56 tok)
- sessions_list: 212 chars (~53 tok)
- subagents: 191 chars (~48 tok)
- sessions_history: 161 chars (~41 tok)
- memory_search: 139 chars (~35 tok)
- memory_get: 128 chars (~32 tok)
- session_status: 89 chars (~23 tok)
- agents_list: 33 chars (~9 tok)
Top tools (summary text size):
- cron: 2,689 chars (~673 tok)
Session tokens (cached): 16,759 total / ctx=256000
Inline shortcut: a command token inside normal text (e.g. “hey /status”) that runs immediately (allowlisted senders only) and is stripped before the model sees the remaining message.
Context breakdown (detailed)- AGENTS.md: OK | raw 7,804 chars (~1,951 tok) | injected 7,804 chars (~1,951 tok)
- TOOLS.md: OK | raw 850 chars (~213 tok) | injected 850 chars (~213 tok)
Skills list (system prompt text): 2,052 chars (~513 tok) (6 skills)
Skills: discord, healthcheck, skill-creator, sonoscli, tmux, weather
Top skills (prompt entry size):
- healthcheck: 491 chars (~123 tok)
- skill-creator: 296 chars (~74 tok)
- tmux: 255 chars (~64 tok)
- discord: 215 chars (~54 tok)
- sonoscli: 207 chars (~52 tok)
- weather: 202 chars (~51 tok)
Top tools (schema size):
- message: 3,885 chars (~972 tok)
- browser: 1,897 chars (~475 tok)
- nodes: 1,479 chars (~370 tok)
- exec: 1,086 chars (~272 tok)
- process: 961 chars (~241 tok)
- web_search: 753 chars (~189 tok)
- cron: 662 chars (~166 tok)
- canvas: 661 chars (~166 tok)
- edit: 591 chars (~148 tok)
- gateway: 465 chars (~117 tok)
- read: 392 chars (~98 tok)
- web_fetch: 374 chars (~94 tok)
- sessions_spawn: 336 chars (~84 tok)
- write: 313 chars (~79 tok)
- sessions_send: 273 chars (~69 tok)
- tts: 223 chars (~56 tok)
- sessions_list: 212 chars (~53 tok)
- subagents: 191 chars (~48 tok)
- sessions_history: 161 chars (~41 tok)
- memory_search: 139 chars (~35 tok)
- memory_get: 128 chars (~32 tok)
- session_status: 89 chars (~23 tok)
- agents_list: 33 chars (~9 tok)
Top tools (summary text size):
- cron: 2,689 chars (~673 tok)
Session tokens (cached): 16,759 total / ctx=256000
Inline shortcut: a command token inside normal text (e.g. “hey /status”) that runs immediately (allowlisted senders only) and is stripped before the model sees the remaining message.
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
134,827Members
Resources
Similar Threads
Was this page helpful?
