Evidence gathering for performance review cycles. Gathers goal completion evidence, peer feedback, development progress, scope changes, and values alignment, organised along the org's performance framework dimensions, with organizational values as the 'how' lens. Surfaces evidence gaps. Never suggests ratings, only organises evidence for the manager's judgment.
Principle: "You are responsible." This skill gathers and organises evidence. Rating decisions and development assessments are the manager's alone.
Helps managers prepare evidence-based assessments for performance review cycles. The org's performance framework dimensions measure what was achieved and how the person developed. Organizational values measure how they showed up while doing it.
When to Use
During full review cycles (per the org's review cadence)
During lighter check-ins between full reviews
When the manager says "help me prep [name]'s review", "gather evidence for [name]'s performance"
Can be run for one team member or all reports in batch
Context: Performance Framework
Load the org's performance framework from (created during ). This defines:
相关技能
manager-context/performance-framework.md
/setup
Framework dimensions and sub-dimensions
Rating scale
Promotion readiness labels (if tracked)
Review cadence
If manager-context/performance-framework.md doesn't exist, ask the manager to run /setup first.
Instructions
If any MCP connector is unavailable, follow the connector unavailability protocol in ../../references/operating-principles.md.
1. Identify Scope
Determine who to prepare for:
Single team member: "prep [name]'s review"
Whole team: "prep all reviews" (runs sequentially for each report)
Determine the review period:
Default: last 6 months (for bi-annual review) or last 3 months (for check-in)
Can be customised: "since [date]"
2. Load Context
For the target team member, read from manager-context/team/[name].md:
Their goals (locations in Notion/Drive)
Their development areas from last review
Their role and level (from Job Architecture)
Their projects and responsibilities
Also load:
manager-context/performance-framework.md: org-specific framework dimensions and rating descriptors (falls back to ../../references/performance-framework.md defaults)
manager-context/management-framework.md: org-specific management dimensions (falls back to ../../references/management-framework.md defaults)
../../references/values-guide.md: values definitions and signal guidance
manager-context/values.md: the organization's specific values
3. Gather Evidence Along Each Dimension
For each dimension and sub-dimension in the org's performance framework (from manager-context/performance-framework.md), gather evidence from connected sources.
For each sub-dimension:
Notion: Pull goals, project pages, status updates, metrics relevant to this dimension
Slack: Search for messages showing activity, feedback, recognition, or friction related to this dimension
Google Drive: Look for deliverables, reports, documents tied to this dimension
Calendar: Check for activities that signal growth or scope changes (new meetings, new stakeholders)
Compile: what evidence was found, with links
Common evidence patterns by dimension type:
Results/delivery dimensions: goal completion, shipped work, quality feedback, business outcomes
Collaboration/leadership dimensions: cross-team activity, mentoring, influence in discussions
For dimensions that are hardest to assess digitally (e.g., behavioural growth, leadership presence), explicitly flag that the manager's direct observations carry more weight.
4. Gather Values Evidence
Values are the "how": how this person delivered their results and showed up for the team. Search for evidence across the organization's values (from manager-context/values.md). See ../../references/values-guide.md for guidance on finding value signals.
For each value defined in manager-context/values.md, search for evidence using the signal guidance stored there. Common evidence sources by value type:
Collaboration / teamwork values:
Slack: cross-team collaboration, helping unblock others, participating in team decisions
DMs (with manager): conversations about team dynamics, commitment to group decisions
Ambition / ownership values:
Slack/Notion: volunteering for stretch work, proposing ideas, driving outcomes
Evidence of taking things to completion without being pushed
Innovation / resourcefulness values:
Slack: creative problem-solving, finding workarounds, learning from obstacles
Evidence of unblocking themselves or the team under constraints
Calendar: sustainable work patterns or concerning overwork patterns
For each value, compile evidence as observations (not judgments):
Strong signal: Multiple visible examples
Some signal: 1-2 examples
Gap: No evidence found (note: absence of evidence ≠ absence of the behaviour)
5. Check Peer Recognition
Search Slack for recognition this person received during the review period:
Direct shoutouts from teammates
Recognition in team channels
Reactions on their messages (high-reaction messages = valued contributions)
6. Identify Evidence Gaps
For each dimension, assess evidence strength:
Strong evidence: Multiple sources corroborate
Some evidence: 1-2 data points
Gap: No evidence found, manager needs to gather this manually
7. Produce the Evidence Summary
Read references/output-template.md for the full output template structure (individual and batch mode).
8. For Batch Mode (All Reports)
If preparing for the whole team, produce individual evidence summaries for each team member plus a team-level comparison view. See the batch mode template in references/output-template.md.
9. Present
Here's the evidence I gathered for [name]'s review. I've flagged gaps where you'll want to add your own observations.
Remember: this is evidence gathering only. Rating decisions and promotion assessments are yours to make based on the full picture, including things I can't see.
10. Sub-Agent Review
Spawn a sub-agent to review the evidence summary with fresh eyes. The reviewer should:
Check for recency bias: is most evidence from the last few weeks, or spread across the review period?
Check for dimension imbalance: are some dimensions well-evidenced while others are thin? Flag under-covered areas.
Check for interpretive language: flag any phrasing that crosses from evidence ("shipped X on time") into interpretation ("demonstrated strong execution").
Verify evidence gaps are honestly flagged, not papered over with weak data.
Check that values evidence is presented as "examples I found", not "the full picture."
Incorporate the reviewer's feedback before presenting the final summary to the manager.
Important Notes
Read ../../references/operating-principles.md for shared operating principles (data scope, DM flagging, signals vs diagnoses, connector unavailability).
Additional notes specific to this skill:
NEVER suggest a rating. Not even hints. The manager decides ratings. Period.
NEVER compare team members to each other. Evidence is per-person. Calibration is the manager's job.
Evidence, not interpretation. "Shipped feature X on time" is evidence. "Demonstrated strong execution" is interpretation. Stick to evidence.
Flag gaps honestly. "I found no evidence for behavioral growth" is more useful than padding with weak data.
Recency bias warning. Note if most evidence is from the last month vs. spread across the review period.
Values evidence is hardest to gather digitally. Many values are lived in person, not in Slack. The manager's direct observations are more authoritative.
This supplements, not replaces. The manager has 6 months of direct observation.