Fix an agent's charset decoding of fetched web pages
A contributor fixed OpenClaw's link-understanding step, which decoded Shift_JIS pages as replacement characters, by reusing the existing charset-aware decoder, and verified it end to end. ·
Done withOpenClaw
- What they did
- The contributor found OpenClaw's link-understanding step decoding fetched pages with a plain UTF-8 decoder, ignoring declared charsets. The codebase already had a charset-aware decoder in web_fetch, so they exported and reused it rather than writing another. They tested through the real production path with the SSRF guard enabled.
- How it went
- Replacement characters on an Aozora Bunko Shift_JIS page fell from 208,953 to 0, and on a second legacy site from 13 to 0. The maintainer re-ran it through a built Gateway before merging.
- Worth knowing
- An earlier attempt stalled because its strict decoder would reject pages declaring no charset; the reused decoder falls back to UTF-8 for unknown labels, avoiding that.
Try it yourself with OpenClaw
In my OpenClaw checkout, find why fetched web pages in [language or encoding, e.g. Shift_JIS] come out as replacement characters in the link-understanding step. Fix it by reusing the existing charset-aware decoder rather than writing new code, add a test using a sample page in that encoding, and run it end to end against [a real URL]. Stop when the test passes and show me the diff before committing anything.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Oct 31, 2026
Related use cases
Manage 18 agents from a Notion workspace
Notion used as the control plane for a fleet of 18 OpenClaw agents.
Triage failed GitHub Actions runs and self-heal flaky tests
A QA engineer's OpenClaw agent reads failed CI logs, classifies failures, posts analysis to GitHub and Slack and opens issues; flakiness fell from 8% to 0.8%, though it once wrongly altered an assertion.
Build and maintain a local ops dashboard with agents
A solo business owner had OpenClaw agents build and iterate on a local mission-control dashboard with Kanban, approvals, cron and social scheduling, replacing Asana; open-ended overnight runs burned tokens.
Submitted by @jennabrinning
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Monitor tasks and open pull requests with long-running agents
A commenter runs a persistent script that watches tasks and uses Cursor to open pull requests, building a documentation summarizer and a data pipeline.
Turn GitHub issues into draft pull requests with long-running agents
The author used Dots for read-heavy long-horizon tasks like reading GitHub issues and making PRs, which were typically poor but gave a starting point.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.