perf-feedback-rehearsal

/perf-feedback-rehearsal

Load SIP.md, VOICES.md, agents/starlight-performance.md, skills/people-intelligence/feedback-conversations.md, and the manager's Genius Profile if available. Produce a Feedback Rehearsal — multiple openings, SCARF-aware adjustments, three reaction trees with the manager's response to each, and aftercare. Save to people-intelligence/performance/rehearsal-<topic-slug>-<YYYY-MM-DD>.md.

Disclaimer (non-waivable)

This is rehearsal architecture for a high-stakes feedback conversation. It is NOT legal counsel — if this feedback is part of a documentation chain leading to termination, demotion, or formal discipline, route to /perf-difficult-conversation and have the documentation reviewed by employment counsel before delivery. It is NOT psychotherapy — if the recipient is in clinical-level distress, route to clinical support, not feedback. The rehearsal goes stale after 7 days; deliver soon or rehearse again.

Input

$ARGUMENTS

When this command fires

• A manager has feedback to deliver and the conversation is non-trivial
• The feedback addresses a recurring pattern, not a one-off
• The recipient is likely to react strongly (history of defensiveness, status threat, tenure imbalance)
• The manager has not delivered feedback like this before and wants a rehearsed structure

When this command does NOT fire

• The feedback is trivial ("hey, the deck title was misspelled") — no rehearsal needed; just SBI in the moment
• The conversation is termination, demotion, or formal discipline → route to /perf-difficult-conversation
• The conversation is conflict between two parties → route to /perf-conflict-mediation
• The recipient is in clinical-level distress → refuse rehearsal and route to clinical
• The "feedback" is really a PIP scaffolding for termination → refuse and route to honest termination conversation

Process

1. Recall prior rehearsals (memory handshake).
   • Before generating, check the SIS memory layer for prior rehearsals on the same topic or with the same manager/recipient pair. The substrate-aware orchestrator routes via the People Intelligence namespace.
   • Run from the SIS repo root:

     cd private/voice-operator && MSYS_NO_PATHCONV=1 python -m service.memory.cli recall \
       --query "<topic-slug> <manager-slug> <recipient-slug>" \
       --k 3 \
       --namespace people-intelligence/perf \
       --source /perf-feedback-rehearsal
     

   • Parse the JSON output. If hits surface, treat them as continuity context: reference the prior pattern, note what changed, avoid repeating dead-ends. If no hits, proceed without prior context — first rehearsal in this lineage.
   • This is the People Intelligence dog-food gate for v0.1 of the SIS memory orchestrator (see skills/memory/sis-memory-orchestrator/SKILL.md). The recall is logged to memory/_audit/<date>.jsonl.