fix: positive-frame all references for 100% joy-check compliance

agents/database-engineer/references/sql.md

@@ -232,7 +232,7 @@ user = session.query(User).filter(User.id == user_id).first()
 
 ---
 
-### Avoid Functions on Indexed Columns in WHERE
+### Use Raw Columns in WHERE Clauses for Index Hits
 **Detection**:
 ```bash
 # Find function calls wrapping column names in WHERE clauses

agents/golang-general-engineer/references/go-errors.md

@@ -381,7 +381,7 @@ if errors.As(err, &validationErr) {
 - Add context to wrapped errors
 - Use `errors.Is()` to check for specific errors
 - Use `errors.As()` to extract error details
-- Don't wrap errors you want to hide
+- Return errors unwrapped when callers should not match them with errors.Is
 
 ---
 

agents/hook-development-engineer/references/preferred-patterns.md

@@ -37,7 +37,7 @@ for line in f:
         # Stops reading after finding the match -- O(1) best case
 ```
 
-Avoid loading the entire file into memory and iterating through it. Parse only what is needed, and exit as soon as the relevant data is found.
+Parse only what is needed and exit as soon as the relevant data is found. Read files incrementally rather than loading entire contents into memory.
 
 **Why this matters**: Hooks fire on every tool call. A hook that takes 200ms adds 200ms to every Read, Edit, Write, and Bash invocation. At 50+ tool calls per session, that is 10+ seconds of pure overhead.
 

agents/performance-optimization-engineer/references/metrics-and-monitoring.md

@@ -63,7 +63,7 @@ onINP(sendToAnalytics, { reportAllChanges: true }) // INP updates on each intera
 
 ### Sampling Strategy for High-Traffic Sites
 
-Don't send 100% of metric events — use sampling to reduce costs without losing statistical significance.
+Sample metric events (1-10%) to reduce costs while maintaining statistical significance.
 
 ```typescript
 const SAMPLE_RATE = 0.1 // 10% sample

agents/python-general-engineer/references/python-errors.md

@@ -85,7 +85,7 @@ if __name__ == "__main__":
 **Prevention**:
 - Only use `asyncio.run()` at the top level of scripts
 - In notebooks, use `await` directly
-- Don't nest `asyncio.run()` calls
+- Use a single top-level `asyncio.run()` and await coroutines within it
 
 ---
 

agents/testing-automation-engineer/references/preferred-patterns.md

@@ -158,7 +158,7 @@ it('calculates discount', () => {
 
 **Why it's wrong:**
 - Tests give false confidence - they pass even when code broken
-- Don't actually verify correct behavior
+- Fail to verify correct behavior
 - Waste CI/CD time without providing value
 - Can mask regressions and bugs
 

agents/typescript-frontend-engineer/references/typescript-preferred-patterns.md

@@ -61,8 +61,7 @@ interface ApiResponse<T> {
 ```
 
 **When to use**:
-- NEVER use `any` without explicit justification
-- If absolutely necessary, add a comment explaining why
+- Use specific types; when `any` is unavoidable, add a comment explaining why
 - Consider `unknown` instead - forces type checking before use
 
 ---
@@ -299,10 +298,8 @@ const edgeCase = complexTypeFunction(input)
 ```
 
 **When to use**:
-- NEVER use `@ts-ignore` or `@ts-nocheck` without thorough justification
-- If used, add detailed comment explaining WHY and what's being ignored
+- Replace `@ts-ignore` with `@ts-expect-error` (provides compile-time checking); add a comment when suppression is necessary
 - Link to GitHub issue if it's a TypeScript bug
-- Prefer `@ts-expect-error` over `@ts-ignore` (fails if error is fixed)
 
 ---
 

scripts/bulk_fix_instruction_joy.py

@@ -23,12 +23,35 @@
     "CLAUDE.md",
 ]
 SAFE_REWRITES = {
-    "## Anti-Patterns": "## Patterns to Detect and Fix",
-    "## Common Anti-Patterns": "## Common Patterns to Detect and Fix",
-    "## Anti-Pattern Catalog": "## Pattern Catalog",
-    "## Operational Anti-Patterns": "## Operational Patterns to Detect and Fix",
+    # Heading-level Anti-Pattern renames (mechanical, domain-independent)
+    "## Anti-Patterns": "## Common Failure Modes",
+    "### Anti-Patterns": "### Common Failure Modes",
+    "## Common Anti-Patterns": "## Common Failure Modes",
+    "## Anti-Pattern Catalog": "## Failure Mode Catalog",
+    "## Operational Anti-Patterns": "## Operational Failure Modes",
+    "## Anti-Patterns in Runbook Authoring": "## Runbook Authoring Failure Modes",
+    "### Anti-Patterns (never produce these)": "### Failure Modes (detect and fix these)",
     "### FORBIDDEN Patterns (HARD GATE)": "### Hard Gate Patterns",
     "### Anti-Pattern Checklist: What NOT To Do": "### Pattern Checklist: What to Detect and Fix",
+    # Domain-specific Anti-Pattern headings
+    "## Vendor Management Anti-Patterns": "## Vendor Management Failure Modes",
+    "## Triage Anti-Patterns": "## Triage Failure Modes",
+    "## Risk Assessment Anti-Patterns": "## Risk Assessment Failure Modes",
+    "### Retro Anti-Patterns with Corrections": "### Retro Failure Modes with Corrections",
+    "### Reorg Anti-Patterns": "### Reorg Failure Modes",
+    "### Question Design Anti-Patterns": "### Question Design Failure Modes",
+    "## Process Documentation Anti-Patterns": "## Process Documentation Failure Modes",
+    "### Planning Anti-Patterns with Corrections": "### Planning Failure Modes with Corrections",
+    "### Org Design Anti-Patterns": "### Org Design Failure Modes",
+    "### Onboarding Anti-Patterns": "### Onboarding Failure Modes",
+    "## KB Quality Anti-Patterns": "## KB Quality Failure Modes",
+    "### Follow-Up Anti-Patterns": "### Follow-Up Failure Modes",
+    "#### Feedback Anti-Patterns": "#### Feedback Failure Modes",
+    "### De-escalation Anti-Patterns": "### De-escalation Failure Modes",
+    "### Dashboard Anti-Patterns": "### Dashboard Failure Modes",
+    "## Critique Anti-Patterns to Avoid": "## Critique Failure Modes",
+    "### Common Funnel Anti-Patterns": "### Common Funnel Failure Modes",
+    "## Change Management Anti-Patterns": "## Change Management Failure Modes",
 }
 PRIMARY_PATTERNS = [
     ("NEVER", re.compile(r"\bNEVER\b")),

scripts/tests/test_bulk_fix_instruction_joy.py

@@ -17,7 +17,7 @@ def test_scan_file_marks_safe_heading_rewrite(tmp_path: Path) -> None:
     path.write_text("## Anti-Patterns\n\nBody\n", encoding="utf-8")
     findings = _mod.scan_file(path)
     assert len(findings) == 1
-    assert findings[0].safe_fix == "## Patterns to Detect and Fix"
+    assert findings[0].safe_fix == "## Common Failure Modes"
 
 
 def test_apply_safe_rewrites_rewrites_heading_only(tmp_path: Path) -> None:
@@ -26,7 +26,7 @@ def test_apply_safe_rewrites_rewrites_heading_only(tmp_path: Path) -> None:
     changed = _mod.apply_safe_rewrites(path)
     assert changed == 2
     text = path.read_text(encoding="utf-8")
-    assert "## Patterns to Detect and Fix" in text
+    assert "## Common Failure Modes" in text
     assert "### Hard Gate Patterns" in text
 
 

scripts/validate_positive_instruction_docs.py

@@ -40,6 +40,10 @@
     # not toolkit operator instructions. Contextual exception per instruction-rubric.md.
     "voice-andy-nemmity",
     "voice-vexjoy",
+    # PHILOSOPHY.md uses negative patterns as quoted examples in a teaching table
+    # (e.g. "NEVER edit code directly" → "Route all code modifications...").
+    # The patterns are the subject, not instructions.
+    "PHILOSOPHY.md",
 )
 
 

skills/business/customer-support/references/knowledge-base.md

@@ -313,7 +313,7 @@ Billing & Account
 | FAQ | Detailed article | "For a full walkthrough, see [Guide]" |
 | Known issue | Workaround | Keep problem-to-solution chain short |
 
-Avoid circular links unless both articles are genuinely useful entry points. Use relative links within the KB -- they survive restructuring.
+Link articles in one direction unless both are genuinely useful entry points. Use relative links within the KB — they survive restructuring.
 
 ---
 
@@ -397,7 +397,7 @@ Transform the ticket-specific resolution into a general guide:
 
 ---
 
-## KB Quality Anti-Patterns
+## KB Quality Failure Modes
 
 | Anti-Pattern | Problem | Fix |
 |-------------|---------|-----|

skills/business/customer-support/references/response-drafting.md

@@ -75,7 +75,7 @@ When a customer is angry, frustrated, or escalating. The goal is not to win an a
 | "You need to..." | "Here's how to resolve this: [steps]" | Directive language feels adversarial under stress |
 
 <!-- no-pair-required: anti-patterns are self-explanatory failure modes; positive approach is in surrounding context -->
-### De-escalation Anti-Patterns
+### De-escalation Failure Modes
 
 | Anti-Pattern | Why It Fails |
 |-------------|-------------|

skills/business/customer-support/references/triage-methodology.md

@@ -283,7 +283,7 @@ Every triage produces this structured output:
 
 ---
 
-## Triage Anti-Patterns
+## Triage Failure Modes
 
 | Anti-Pattern | Problem | Correct Approach |
 |-------------|---------|-----------------|

skills/business/design/references/design-critique.md

@@ -194,7 +194,7 @@ Simulate a new user attempting a specific task. For each step:
 
 ---
 
-## Critique Anti-Patterns to Avoid
+## Critique Failure Modes
 
 | Anti-pattern | Do instead |
 |-------------|-----------|

skills/business/finance/references/variance-analysis.md

@@ -157,7 +157,7 @@ Every narrative must be:
 | Actionable | Identifies required follow-up or decision |
 | Concise | 2-4 sentences. No filler |
 
-### Anti-Patterns (never produce these)
+### Failure Modes (detect and fix these)
 
 | Anti-Pattern | Why It Fails | Fix |
 |-------------|-------------|-----|

skills/business/hr/references/org-planning.md

@@ -35,7 +35,7 @@ Headcount modeling, org design principles, capacity planning, and people analyti
 | Single points of failure | 0 critical | 1-2 | >3 (bus factor risk) |
 | Skip-level ratio | Every IC has skip-level access | Some don't | No skip-levels happening |
 
-### Org Design Anti-Patterns
+### Org Design Failure Modes
 
 | Anti-Pattern | Symptom | Root Cause | Fix |
 |-------------|---------|-----------|-----|
@@ -230,7 +230,7 @@ Composite risk score based on multiple factors:
 | **Team kickoff** | Week 1 | New teams | Mission, priorities, working agreements |
 | **30-day check-in** | Day 30 | All affected | How's it going, what needs adjustment |
 
-### Reorg Anti-Patterns
+### Reorg Failure Modes
 
 | Anti-Pattern | Risk | Prevention |
 |-------------|------|-----------|

skills/business/hr/references/performance-management.md

@@ -97,7 +97,7 @@ Purpose: Document performance assessment, provide development guidance, justify
 | Gap identification | "You're strong at execution. The gap I see is in proactive communication — stakeholders are sometimes surprised by delays." | Names the specific gap between current and expected. |
 | Frequency + evidence | "In the last quarter, 3 of 5 project updates were late by more than a day. Let's discuss what's causing this." | Data-driven. Not "you're always late." |
 
-#### Feedback Anti-Patterns
+#### Feedback Failure Modes
 
 | Anti-Pattern | Problem | Better |
 |-------------|---------|--------|

skills/business/hr/references/recruiting.md

@@ -100,7 +100,7 @@ For each role, define 4-6 competencies. Each competency gets dedicated interview
 | Trained interviewers | Everyone on the panel has completed structured interviewing training. |
 | Consistent questions | Same question set per competency across all candidates for the same role. |
 
-### Question Design Anti-Patterns
+### Question Design Failure Modes
 
 | Anti-Pattern | Problem | Better Approach |
 |-------------|---------|-----------------|
@@ -182,7 +182,7 @@ Date: [Date] | Panel: [Names]
 | **Days 31-60** | Contribute | Delivers independently on scoped work. Participates in reviews. Identifies one improvement. |
 | **Days 61-90** | Own | Owns a workstream. Contributes to planning. Can onboard the next new hire. |
 
-### Onboarding Anti-Patterns
+### Onboarding Failure Modes
 
 | Anti-Pattern | Problem | Fix |
 |-------------|---------|-----|

skills/business/legal/references/legal-writing.md

@@ -223,7 +223,7 @@ to any other legal holds, those obligations remain in effect.
 
 ## Escalation Triggers
 
-Before generating ANY templated legal response, check these triggers. If any fire, do NOT generate a templated response -- recommend escalation instead.
+Before generating ANY templated legal response, check these triggers. If any fire, recommend escalation to qualified counsel instead of generating a template.
 
 ### Universal Triggers (All Categories)
 

skills/business/legal/references/llm-legal-failure-modes.md

@@ -31,7 +31,7 @@ Legal analysis has properties that amplify LLM weaknesses:
 - LLM provides citation with unusual confidence (no hedging language)
 
 **Mitigation:**
-- NEVER present a case citation, statute number, or regulatory reference as authoritative without instructing the user to verify
+- Mark every case citation, statute number, and regulatory reference as "requires verification" and instruct the user to confirm with primary sources
 - Use hedging: "regulations such as..." or "provisions similar to..." rather than citing specific sections
 - For well-known, foundational statutes (GDPR, CCPA, HIPAA), cite the act name and general provision (e.g., "GDPR Article 28") but avoid citing specific subsections unless highly confident
 - When the user needs specific citations, recommend legal research tools (Westlaw, Lexis, regulatory authority websites)

skills/business/marketing/references/content-strategy.md

@@ -282,14 +282,14 @@ The voice stays constant. Tone adjusts to context.
 #### Inclusive Language Rules
 
 - Gender-neutral: they/them for unknown individuals
-- Avoid ableist language ("crazy", "blind spot", "lame")
+- Use inclusive alternatives for ableist terms ("unexpected" for "crazy", "oversight" for "blind spot", "weak" for "lame")
 - "Simple" or "straightforward" instead of "easy" (easy varies by person)
-- Avoid culturally specific idioms that don't translate
+- Use universally understood phrasing over culturally specific idioms
 
 #### Competitor and Category Terms
 
 - Use your preferred category framing, not the competitor's
-- Avoid terms competitors coined (reinforces their positioning)
+- Use original terminology over competitor-coined terms (which reinforce their positioning)
 - Refer to competitors generically unless direct comparison is the point
 - Define your differentiation language and use it consistently
 

skills/business/marketing/references/email-sequences.md

@@ -128,7 +128,7 @@ Map each email to a specific stage:
 - Under 50 characters (35-40 optimal for mobile)
 - No ALL CAPS words
 - No excessive punctuation (!!!, ???)
-- Avoid spam triggers: free, act now, limited time, congratulations, click here
+- Replace spam-trigger words (free, act now, limited time, congratulations, click here) with specific value statements
 - Preview text complements -- never repeats -- the subject line
 - Always provide 2-3 options per email for A/B testing
 

skills/business/operations/references/change-management.md

@@ -327,7 +327,7 @@ Rollback procedures must be tested before the change. An untested rollback is a
 
 ---
 
-## Change Management Anti-Patterns
+## Change Management Failure Modes
 
 | Anti-Pattern | Symptom | Fix |
 |-------------|---------|-----|

skills/business/operations/references/process-documentation.md

@@ -364,7 +364,7 @@ Annual value: 3hr × 48 weeks × $75/hr loaded cost = $10,800
 
 ---
 
-## Process Documentation Anti-Patterns
+## Process Documentation Failure Modes
 
 | Anti-Pattern | Symptom | Fix |
 |-------------|---------|-----|

skills/business/operations/references/risk-assessment.md

@@ -256,7 +256,7 @@ Maintain evidence that risk management is active, not just documented.
 
 ---
 
-## Risk Assessment Anti-Patterns
+## Risk Assessment Failure Modes
 
 | Anti-Pattern | Symptom | Fix |
 |-------------|---------|-----|

skills/business/operations/references/runbook-authoring.md

@@ -349,7 +349,7 @@ Review the runbook when any of these occur:
 
 ---
 
-## Anti-Patterns in Runbook Authoring
+## Runbook Authoring Failure Modes
 
 | Anti-Pattern | Example | Fix |
 |-------------|---------|-----|

skills/business/operations/references/vendor-management.md

@@ -258,7 +258,7 @@ Every vendor relationship should have an exit plan. Not because you plan to leav
 
 ---
 
-## Vendor Management Anti-Patterns
+## Vendor Management Failure Modes
 
 | Anti-Pattern | Symptom | Fix |
 |-------------|---------|-----|

skills/business/product-management/references/metrics-review.md

@@ -177,7 +177,7 @@ KR3: 3 core workflows with >80% task completion rate
 - **Beware averages**: Mean time-to-activate can be misleading if the distribution is bimodal. Use median or percentiles.
 - **Attribution**: If you changed multiple things, you cannot attribute funnel changes to one change. Run A/B tests for causal claims.
 
-### Common Funnel Anti-Patterns
+### Common Funnel Failure Modes
 
 | Anti-Pattern | Problem | Fix |
 |-------------|---------|-----|
@@ -327,7 +327,7 @@ Possible explanations:
 └────────────────────────────────────────┘
 ```
 
-### Dashboard Anti-Patterns
+### Dashboard Failure Modes
 
 | Anti-Pattern | Why It Fails |
 |-------------|-------------|

skills/business/product-management/references/spec-writing.md

@@ -77,7 +77,7 @@ Non-Goals:
 | **S**mall | Completable in one sprint |
 | **T**estable | Clear verification path |
 
-### Anti-Patterns
+### Common Failure Modes
 
 | Pattern | Problem | Fix |
 |---------|---------|-----|

skills/business/productivity/references/daily-weekly-planning.md

@@ -56,7 +56,7 @@ Match task difficulty to energy state. Most knowledge workers have predictable e
 - If more than 3 hours of meetings, only 2 Top 3 items are realistic. Acknowledge this explicitly.
 - Buffer time is not optional — unplanned work always appears. Plans without buffers fail by noon.
 
-### Planning Anti-Patterns with Corrections
+### Planning Failure Modes with Corrections
 
 | Anti-Pattern | What Goes Wrong | Do Instead |
 |-------------|----------------|------------|

skills/business/productivity/references/status-updates.md

@@ -172,7 +172,7 @@ The simplest effective retro format.
    - What can we change? (specific action, not aspiration)
 5. **Decide** (5 min): Pick 1-2 action items. Each has an owner and a deadline. More than 2 actions from a single retro dilutes focus.
 
-### Retro Anti-Patterns with Corrections
+### Retro Failure Modes with Corrections
 
 | Anti-Pattern | What Goes Wrong | Do Instead |
 |-------------|----------------|------------|

skills/business/sales/references/outreach-patterns.md

@@ -166,7 +166,7 @@ Total email: under 150 words. 5 sentences is ideal. 7 sentences is the hard max.
 | Follow-up 3 | Day 14 | Break-up email (creates urgency through scarcity) |
 | LinkedIn touch | Day 5-10 | Parallel channel. Connection request or engage with their content. |
 
-### Follow-Up Anti-Patterns
+### Follow-Up Failure Modes
 
 | Bad Follow-Up | Why It Fails | Instead |
 |--------------|-------------|---------|

skills/code-quality/code-linting/references/ruff-rules-reference.md

@@ -113,7 +113,7 @@ if not items:
 
 ---
 
-### Avoid Mutable Default Arguments (B006)
+### Use Immutable Default Arguments (B006)
 
 **Detection**:
 ```bash