Blindspot Detection Directions¶
How to scan for active work-pattern blindspots — intervention levels, cascade checking, rewriting template, output format
When to use¶
user wants to check their current work for active cognitive or behavioral patterns
Slash command: /wb-meta-blindspots
Related capabilities¶
context/context_bundletasks/task_briefingcontracts/contract_health
Directions¶
This framework helps the user spot patterns they have chosen to be accountable for. It applies to any self-accountability domain — research workflow, coding habits, health signals, focus, communication patterns — not just work patterns narrowly.
Step 1 — Load the user's documented patterns¶
The specific patterns live in personal knowledge, not in this unit. Load all categories the user has documented:
`mcp__work-buddy__wb_run("knowledge_personal", {"category": "work_pattern"})`
`mcp__work-buddy__wb_run("knowledge_personal", {"category": "self_regulation"})`
// or any other category they track — see the `knowledge_personal` capability for the full category list
If no personal pattern units exist yet, tell the user: "No documented patterns yet — knowledge_mint one first, or we'll be guessing." Do not invent pattern names.
Step 2 — Gather current context¶
`mcp__work-buddy__wb_run("context_bundle", {"hours": 24})`
`mcp__work-buddy__wb_run("task_briefing")`
`mcp__work-buddy__wb_run("contract_health")`
Or run the full workflow: mcp__work-buddy__wb_run("detect-blindspots").
Step 3 — Name what's active¶
Use only pattern names the user has documented. Be specific with evidence: "This looks like
Intervention levels¶
- Level 1 (first detection): Name and redirect — name the pattern, give one next action, park nonessential work.
- Level 2 (pattern persists): Force a decision — require the user to classify items, pick one path, or set an explicit stop rule.
- Level 3 (repeated): Tighten environment — reduce degrees of freedom. The exact form depends on the pattern domain (fewer branches for work patterns, fewer obligations for overcommitment, tighter routines for habit patterns).
Cascade checking¶
Patterns can chain. If the user's personal pattern unit documents cascade relationships, check downstream patterns after detecting an upstream one, and address the upstream first.
Rewriting template¶
For vague global complaints ("I'm bad at this"), rewrite:
"When [trigger] happens, I tend to [habitual response], which causes [cost], so the best immediate intervention is [specific response]."
Output format¶
- Pattern detected (with evidence)
- Intervention level
- One next action
- One thing to park or defer
Don'ts¶
- Don't diagnose personality — name the pattern, not the person.
- Don't list every possible pattern — name the 1–2 actually active.
- Don't pile on interventions — one next action is enough.
- Don't invent patterns the user hasn't documented. If the signal is real but unnamed, suggest minting a new pattern unit rather than using ad-hoc labels.