performance

name	performance
description	(beta) Manage engineers with yeaboi: 1:1 prep from real delivery data, 1:1 completion summaries with tracked action items, periodic performance reviews, and quick notes. Use when the user wants to prepare for a 1:1, write up a held 1:1, draft a performance review, or record an observation about an engineer.

Performance workflows with yeaboi

Beta. Performance mode is in beta — its output is not yet verified against real delivery data. Treat every 1:1 prep, summary and review as a draft to edit, not a verdict, and say so when you present it to the user.

1. Know the roster. Call perf_roster to list the engineers derived from recent Jira/Azure DevOps assignees — these are the names every other tool accepts. If it's empty, no tracker is configured (point at yeaboi --setup).

2. Pick the workflow from what the user asked:

   • Prep a 1:1 → perf_one_on_one_prep with the engineer's name. Present the talking points, feedback, goals and growth areas conversationally — it already folds in the open action items from their previous 1:1.
   • Write up a held 1:1 → ask for the notes/transcript (pasted text), then perf_one_on_one_complete. Leave deliver false — you present the summary; only set deliver: true (emails it via the configured SMTP) if the user explicitly asks. The action items are tracked and carried into the next prep automatically.
   • Draft a review → perf_six_month_review (set period_months if they name a different window). It synthesises past 1:1s, delivery history and the competency framework. Frame the output as a draft for the lead to edit, never a verdict.
   • Record an observation → perf_note_add with the engineer and the note. Notes feed future preps and reviews.

3. Surface warnings (the beta caveat, tracker 401s, missing SMTP, LLM fallback) so the user knows what informed — or didn't inform — the output. Every perf tool now returns the beta caveat as its first warning.

4. Sensitive data. This is personnel material: keep it in the conversation, don't post it anywhere external unless explicitly asked. Exports auto-save under ~/.yeaboi/exports/performance/<engineer>/.

Error handling

Every tool returns {ok, llm_mode, warnings, data}. If ok is false, relay error.message and its hint (usually credentials — yeaboi --setup); don't retry blindly. llm_mode: "fallback" means no LLM was reachable and the artifact is a deterministic skeleton — suggest yeaboi --setup.