← ClaudeAtlas

myna-dev-buglisted

File a bug report from the current session — infers the prompt, output, model, and MCPs from context, asks only what it can't infer, auto-redacts private content, and writes a filled GitHub issue template to tmp/bugs/. Use when: "file a bug", "report this", "/myna-dev-bug the output should have been X".
agentflock/myna · ★ 4 · AI & Automation · score 83
Install: claude install-skill agentflock/myna
# Myna Dev Bug You are filling out a GitHub bug report from the current testing session. Your goal: produce a filled issue template written to `tmp/bugs/`, with minimal input from the user. ## Step 1 — Gather what you already know From the current conversation context, extract: - **Exact prompt** — the user's last substantive prompt before invoking this skill. Copy verbatim. - **Myna's output** — the response that was wrong. Copy verbatim — do not summarize or paraphrase. - **Skill involved** — infer from the prompt or output which Myna skill ran (e.g. sync, capture, prep-meeting). If unclear, leave blank. - **Claude model** — you know what model you are. State it (e.g. Claude Sonnet 4.6). - **MCPs connected** — read Myna config files (e.g. `~/.myna/config/integrations.yaml` or equivalent). Report only the *type* of integration (Email, Slack, Calendar) — never account names, server addresses, or credentials. If config is unreadable or absent, infer from context (did the output reference emails, calendar events, Slack messages?). If unknown, write "Unknown". - **Files affected** — if the output mentioned creating or modifying vault files, note the paths. Otherwise leave blank. ## Step 2 — Parse $ARGUMENTS If `$ARGUMENTS` is provided, it may contain: - A description of the bug - What the output should have been - Both, or neither clearly Read it and extract what you can. Do not assume it's only "what should have happened" — the user may be describing the whole issue. Af