mimetics-dsh-weekly-report: A DSH plugin that generates four-section weekly reports in DeepSeek Harness

What Does It Do?

It is a DSH plugin for DeepSeek Harness maintained by mimetics-ai that turns scattered work materials into a Markdown weekly report with four fixed sections. It solves the problem of gathering notes from many places and sorting them by hand, letting you simply say "generate this week's report" in the harness chat. The report is produced by a built-in pipeline behind a single registered tool, released under the MIT license and ready to use once enabled.

Core Features

• Multi-source intake: work logs, chat excerpts, tickets, spreadsheets, meeting notes and logs can be mixed, each given either as a file path or as raw text.
• Fixed four-section output: content is sorted into Completed This Week, In Progress, Risks and Blockers, and Next Week's Plan, producing Markdown you can paste directly.
• Duplicate merging: the same item appearing across sources is merged into one entry, with the merge rationale explained in the diagnostics.
• Conflicts are never silently resolved: contradictory statuses or mismatched numbers are placed under Risks and Blockers with the conflict spelled out.
• No fabrication: numbers and dates come from the source material, and entries missing data are flagged rather than estimated.
• Local and offline: materials stay in local processes, with no external calls and no telemetry, and you decide where output is saved.

How to Use This Plugin?

After enabling it in DSH and restarting, describe what you need in the chat and attach your materials, for example noting what was finished this week and what is still being integrated. The plugin runs its built-in pipeline through the generation tool and returns the four-section report; to keep an extra copy, supply an additional output path and the directory is created automatically. To confirm it is mounted, check the plugin list in settings or turn on debug logging and watch the registration line at startup. It requires Python 3.9+ locally and uses only the standard library, so no extra dependencies are needed; the interpreter path, pipeline root and per-call timeout can all be overridden in configuration, though defaults usually suffice.

How to Troubleshoot This Plugin?

• No plugin marker in the startup log: confirm it is installed and the debug switch is on, then check the plugin list.
• A message that no usable Python interpreter was found: set the interpreter path explicitly, or point to it through the environment variable.
• A message that the report pipeline entry was not found: your custom pipeline root lacks the entry file, so clearing that setting falls back to the bundled pipeline.
• A material file does not exist error: this is expected behavior, since path inputs fail outright instead of being treated as text and producing a fake report.
• Garbled Chinese output: this should not happen, as the plugin forces UTF-8 encoding.