feed
Execute all steps without asking for user confirmation at intermediate stages. Report results…
Deep Q&A on a paper already analyzed. Actively records valuable insights to the note file during the discussion, without waiting for the user to prompt it.
How it fires
How this command gets triggered: by you, by Claude, or both.
/discussContext preview
What this command does when you run it.
Deep Q&A on a paper already analyzed. Actively records valuable insights to the note file during the discussion, without waiting for the user to prompt it.
Deep Q&A on a paper already analyzed. Actively records valuable insights to the note file during the discussion, without waiting for the user to prompt it.
---
**[Tier C / B]** Read `config.json` to get `data_dir` (`notes_dir` = `data_dir/notes`).
The user can specify the paper by:
Recursively search `notes_dir/` for matching `.md` files. Read the matching note in full. If a same-named PDF exists, read it as well for supplementary detail.
If no match is found, inform the user and ask whether they want to provide the path directly or run `/read` first.
**[Tier A]** Ask the user to paste the note content from a previous session, or describe which paper they want to discuss.
---
Briefly recap the paper's core content in 2–3 sentences. Ask what the user wants to discuss.
In the note file, create a discussion log section (or use the existing one for today if it already exists):
## Discussion Log ### YYYY-MM-DD
---
**Record immediately when any of the following occurs — no user prompt needed:**
**Do not record:**
**Quantity limit:** If a single session generates more than 5 records, consolidate related entries rather than appending indefinitely.
**Format:**
### YYYY-MM-DD **Q: [core of user's question, one sentence]** [Answer in 3–5 sentences, preserving key terms and conclusions. If there is a specific implication for the user's own paper, add a separate line:] → Implication for your paper: ...
Multiple records are appended in sequence. Never overwrite existing content.
---
Confirm the note has been updated. Tell the user how many new records were added in this session.
If the session produced no recordable content, do not create an empty discussion section.
---
Count the number of valid records in the "Discussion Log" section of the **current** paper note.
Also read `MEMORY.md` Reading Progress to count how many papers in the reading list have discussion log entries.
If **either** condition is met:
Then append to the end of the conversation:
> 💡 **You have accumulated substantial discussion on your literature.** > This is a good time to synthesize your research proposal. Type `/propose` to start building your Research Foundation Document. > (No need to do it now — you can run it at any time.)
Show this prompt only once per `/discuss` session. Do not repeat it on subsequent turns within the same session.
A curated library of 550+ medical research agent skills created by AIPOCH, designed to work with Claude Code, Codex, and other SKILL.md-compatible agent platforms.
Execute all steps without asking for user confirmation at intermediate stages. Report results…
Synthesizes accumulated literature notes into a structured **Research Foundation Document…
Accepts: local PDF path, DOI, journal URL, or a pasted abstract with basic metadata.
Generates a global view of the current reading state: what has been read, what is missing,…