4 ms·

I know this isn't apples to apples, but given that I can't get Copilot or other tools to view a simple profile page on LinkedIn makes me curious/skeptical how t

I know this isn't apples to apples, but given that I can't get Copilot or other tools to view a simple profile page on LinkedIn makes me curious/skeptical how this would work in this depth. I'm sure it's possible but I'm curious what the skills and toolchains involved were for you to get all that to work.

AussieWog93 4mo ago
Full claude.md is here: https://github.com/EspoTek/.claude/blob/master/CLAUDE.md https://github.com/EspoTek/.claude/blob/master/CLAUDE.md The skills I have installed are: ``` on frontend-design:frontend-design · plugin · ~90 tok · locked by plugin on agents-sdk · user · ~150 tok on cloudflare · user · ~130 tok on cloudflare-email-service · user · ~180 tok on durable-objects · user · ~130 tok on find-docs · user · ~300 tok on find-skills · user · ~110 tok on sandbox-sdk · user · ~120 tok on stage-chapters · user · ~40 tok on web-perf · user · ~150 tok on workers-best-practices · user · ~130 tok on wrangler · user · ~120 tok ``` The plugins I have are: ``` cc-caffeine Plugin · samber · enabled frontend-design Plugin · claude-plugins-official · enabled ty Plugin · claude-code-lsps · enabled vscode-langservers Plugin · claude-code-lsps · enabled vtsls Plugin · claude-code-lsps · enabled ``` There's also an MCP for Context7. But yeah, this is more or less vanilla Claude Code - at least, nothing related to Android or adb there. It's that good now. A few days ago I asked it to SSH into my Ubuntu box and investigate a hang. It didn't solve the problem fully autonomously like this time but did tell me a whole lot things it wasn't, and hinted at a faulty driver. We went back and forth a bit, it set up a watchdog and taught me how to update the kernel without updating Ubuntu itself, and the server has been rock solid for the past 3-4 days now. Also, if you're curious, full log for the Kodi issue: https://github.com/user-attachments/files/28659304/2026-06-06-141445-whenever-i-launch-kodi-on-my-chromecast-4k-it-cr.txt https://github.com/user-attachments/files/28659304/2026-06-0... I did prompt it a little bit more today in order to get something more production-ready (the original solution kept regenerating the cache on boot, rather than fixing it permanently), but you can see the whole original autonomously-generated solution in the logs. It's insane, seriously.
- darksim905 4mo ago
  Thank you for the in-depth reply!
dcre 4mo ago
Claude Code (or Codex, or OpenCode, or Pi, or Amp — whatever) can do this out of the box without any skills or special tools. The most important thing for making results like this easier to achieve (in any harness) is using the best current models. Right now that's Opus 4.8 and GPT-5.5.
- knollimar 4mo ago
  I mean you can likely use Opus 4.7. I barely notice a difference. 4.8 confabulates more for me
  - dcre 4mo ago
    Yeah, I agree — not a big difference for me between the Opuses. I really just meant Opus 4.x.
mft_ 4mo ago
To expand on another answer, it’s all about the harness. Different harnesses (Claude Code or Cowork, Hermes, OpenCode, Pi, etc.) offer different default tools, system prompts, and ultimately approaches. (IME the corporate CoPilot app is terrible - basically a chat interface.) I’m currently using Hermes for local LLMs - seems pretty good so far.
senko 4mo ago
LinkedIn in particular is quite aggressively blocking any automated attempts to read or navigate through it. I post quite a lot there and wanted to have a copy of my posts on my blog[0] to preserve them. For a few months I was able to use a headless browser + claude code, then LI wised up and started logging it out, so I had to use a regular Chrome, log in manually and then tell the LLM to take over and slowly go through my feed. If you're accessing sites which are not actively blocking bots, or - gasp - have an API, it's much better. [0] example: https://blog.senko.net/may-quick-takes https://blog.senko.net/may-quick-takes
- darksim905 4mo ago
  useful, thanks!
fzzzy 4mo ago
The claude desktop app can drive the real chrome, so it’s less likely to get blocked

AussieWog93 4mo ago

Full claude.md is here: https://github.com/EspoTek/.claude/blob/master/CLAUDE.md https://github.com/EspoTek/.claude/blob/master/CLAUDE.md The skills I have installed are: ``` on frontend-design:frontend-design · plugin · ~90 tok · locked by plugin on agents-sdk · user · ~150 tok on cloudflare · user · ~130 tok on cloudflare-email-service · user · ~180 tok on durable-objects · user · ~130 tok on find-docs · user · ~300 tok on find-skills · user · ~110 tok on sandbox-sdk · user · ~120 tok on stage-chapters · user · ~40 tok on web-perf · user · ~150 tok on workers-best-practices · user · ~130 tok on wrangler · user · ~120 tok ``` The plugins I have are: ``` cc-caffeine Plugin · samber · enabled frontend-design Plugin · claude-plugins-official · enabled ty Plugin · claude-code-lsps · enabled vscode-langservers Plugin · claude-code-lsps · enabled vtsls Plugin · claude-code-lsps · enabled ``` There's also an MCP for Context7. But yeah, this is more or less vanilla Claude Code - at least, nothing related to Android or adb there. It's that good now. A few days ago I asked it to SSH into my Ubuntu box and investigate a hang. It didn't solve the problem fully autonomously like this time but did tell me a whole lot things it wasn't, and hinted at a faulty driver. We went back and forth a bit, it set up a watchdog and taught me how to update the kernel without updating Ubuntu itself, and the server has been rock solid for the past 3-4 days now. Also, if you're curious, full log for the Kodi issue: https://github.com/user-attachments/files/28659304/2026-06-06-141445-whenever-i-launch-kodi-on-my-chromecast-4k-it-cr.txt https://github.com/user-attachments/files/28659304/2026-06-0... I did prompt it a little bit more today in order to get something more production-ready (the original solution kept regenerating the cache on boot, rather than fixing it permanently), but you can see the whole original autonomously-generated solution in the logs. It's insane, seriously.

darksim905 4mo ago
Thank you for the in-depth reply!

dcre 4mo ago

Claude Code (or Codex, or OpenCode, or Pi, or Amp — whatever) can do this out of the box without any skills or special tools. The most important thing for making results like this easier to achieve (in any harness) is using the best current models. Right now that's Opus 4.8 and GPT-5.5.

knollimar 4mo ago
I mean you can likely use Opus 4.7. I barely notice a difference. 4.8 confabulates more for me
- dcre 4mo ago
  Yeah, I agree — not a big difference for me between the Opuses. I really just meant Opus 4.x.

mft_ 4mo ago

To expand on another answer, it’s all about the harness. Different harnesses (Claude Code or Cowork, Hermes, OpenCode, Pi, etc.) offer different default tools, system prompts, and ultimately approaches. (IME the corporate CoPilot app is terrible - basically a chat interface.) I’m currently using Hermes for local LLMs - seems pretty good so far.

senko 4mo ago

LinkedIn in particular is quite aggressively blocking any automated attempts to read or navigate through it. I post quite a lot there and wanted to have a copy of my posts on my blog[0] to preserve them. For a few months I was able to use a headless browser + claude code, then LI wised up and started logging it out, so I had to use a regular Chrome, log in manually and then tell the LLM to take over and slowly go through my feed. If you're accessing sites which are not actively blocking bots, or - gasp - have an API, it's much better. [0] example: https://blog.senko.net/may-quick-takes https://blog.senko.net/may-quick-takes

darksim905 4mo ago
useful, thanks!

fzzzy 4mo ago

The claude desktop app can drive the real chrome, so it’s less likely to get blocked