plugins/guild/skills/discuss/SKILL.md
This skill should be used when the user says "discuss", "let's discuss", "discuss [topic]", "talk about", "let's talk about", "summarize the context", "what are we working on", "break down the topics", "what subjects do we have", "let's review the context", "recap the conversation", or any phrase asking to analyze and discuss the current context or a specific subject within it.
npx skillsauth add hirogakatageri/hirokata discussInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Analyze the current conversation context, surface the subjects it contains, and facilitate a focused discussion with the user — either across all topics or scoped to one specific topic.
/discuss or discuss)No topic argument given. Scan the full conversation context.
Step 1 — Extract subjects
Read the conversation and identify all distinct subjects, decisions, problems, questions, and themes present. Group closely related items under a single subject heading. Aim for 3–8 groups; merge very small or overlapping ones.
Step 2 — Present the topic map
Display a numbered list of subjects with a one-sentence summary for each:
Context Summary
===============
Subjects found in this conversation:
1. <Subject Title>
<One-sentence description of what was discussed or decided.>
2. <Subject Title>
<One-sentence description.>
3. <Subject Title>
<One-sentence description.>
Which subject would you like to discuss? (enter a number, name, or "all")
Step 3 — Enter discussion loop
Wait for the user to pick a subject or say "all". Then enter the discussion loop (see below).
discuss [topic])A topic argument was given. Scope analysis to that topic only.
Step 1 — Find relevant context
Search the conversation for everything related to the given topic: decisions made, problems raised, open questions, referenced files or code, dependencies, and constraints.
Step 2 — Present the topic summary
Discussion: <Topic>
===================
Summary
-------
<2–4 sentence narrative summary of everything the context says about this topic.>
Key points:
• <point>
• <point>
• <point>
Open questions or unresolved items:
• <item> (or "None" if everything is settled)
What would you like to explore or decide about [topic]?
Then enter the discussion loop immediately.
Once a topic is established, drive an active back-and-forth:
Ask a focused question to advance understanding or reach a decision. Base it on gaps, open questions, or unresolved items surfaced during summarization.
Listen and integrate the user's response into the running understanding of the topic.
Offer observations, trade-offs, or recommendations when relevant — connect the topic to other context in the conversation, surface implications, and raise unconsidered factors.
Check satisfaction signals after each exchange: if the user signals they are done ("that's it", "thanks", "good", "we're done"), wrap up.
Proactively check when the discussion reaches a natural conclusion: ask "Is there anything else you'd like to explore about [topic], or are we good?"
Loop until the user is satisfied — never end the loop without confirmation.
If the user says "let's talk about <other topic>" or "actually, discuss X instead" during the loop, summarize the current topic with a one-liner ("Got it — we covered: <brief summary>"), then start a fresh Mode 2 summary for the new topic.
When the discussion ends, present a brief closing summary:
Discussion Complete: <Topic>
============================
What we covered:
• <point>
• <point>
Decisions or conclusions:
• <item> (or "None reached — exploration only")
Next steps (if any):
• <item>
Then ask:
Would you like to discuss another subject, or are we done?
If the user wants another subject, return to Mode 1 Step 2 (show the topic map again) or jump directly to a new topic if the user names one.
development
This skill should be used when the user reports an error, bug, or unexpected behavior and wants it diagnosed and fixed. Trigger on phrases like "check this error", "check this bug", "here's an error", "here's a bug", "I have an error", "I have a bug", "found a bug", "got an error", "debug this", "this is broken", "fix this error", "verify and fix", or any message that includes a stack trace or error output. Runs a structured workflow: gather context, investigate configured log/code sources, report root cause with ranked solutions, then apply a test-driven fix.
testing
This skill should be used when the user says "check svelte env vars", "check environment variables", "validate env vars", "check env var patterns", "audit environment variables", "audit env vars", "check SvelteKit env", "svelte env check", or any phrase asking to audit or validate SvelteKit environment variable usage patterns.
data-ai
Internal skill used by the session-tracker logger agent to append a session entry to .logs/YYYY-MM-DD-log.md, creating the file and directory if needed. Not user-invocable.
data-ai
Internal skill used by the session-tracker logger agent to query git for committed and uncommitted changes in the past 28 hours. Not user-invocable.