Pattern Index
Consolidated from pattern-catalog.md and PATTERN-INDEX.md
Format: ADR-style numbered patterns
Total Patterns: 74 patterns (001-074), plus a template (000). Recent additions (2026-05-09 → 2026-05-24); all are now integrated into the category index below (this highlights box is a recent-filing snapshot, not the authoritative list):
- Pattern-066 (Stacked Silent Failures) Emerging — filed 2026-05-09 (CIO Pattern Sweep #1025 Phase 2C TRUE EMERGENCE finding)
- Pattern-067 (Issue-Body Reality Mismatch) Emerging — filed 2026-05-09 (Lead Dev; 3-of-5 evidence from M2f Group A+B dead-code dispositions same day)
- Pattern-068 (Silent State Mutation in Shared Working Tree) Emerging — filed 2026-05-11 (CIO; parent meta-pattern over P-13 branch-drift + P-15 residue-drift + P-16 index-drift + P-17 working-tree-path-fragmentation)
- Pattern-069 (Coarse Triggers Causing False-Positive Triage Cost) Emerging — filed 2026-05-11 (CIO; hook-design meta-pattern from PreCompact-hook two-incident thread)
- Pattern-070 (Cleanup-Job-with-Cancellation-Hygiene) Emerging — filed 2026-05-15 (Architect; three-instance convergence May 2–5 across #1018 audit cleanup / #1035 composting / #1052 standup; Anthropic Dreams Type 1 consolidation as prospective Proven trigger; CIO co-signs methodology sidecar 12o “Pattern Formation via Successful Imitation”)
- Pattern-071 (Audit Logs as Attack Surface) Emerging — filed 2026-05-15 (Lead Dev; sibling shape of Pattern-064 “Alive Scaffolding” — compliance-shaped scaffolding that actively amplifies leaks vs. passive scaffolding that does nothing; reference implementation #1017
OutputFilterDecision hash-only invariant + write-time guard; promotion to Proven contingent on 4-6 week cohort exposure surfacing no required exception path)
- Pattern-072 (Registries that Grow into Architectural Shapes) Proven — filed Emerging 2026-05-15 (Lead Dev); promoted to Proven 2026-05-16 (CIO) after #1094 ENGINE-DELETION close-out landed Slack handler dispatch as the fourth behavior-deciding consumer of the
task_type registry, with all four formalization-discipline invariants intact. First sub-day Emerging-to-Proven promotion in the catalog (~6 hours between recognition trigger and Proven trigger).
- Pattern-073 (Documentation-Asserted-Behavior Drift) Proven (promoted 2026-05-18) — filed 2026-05-16 (Lead Dev; CIO methodology cosign). Six independent reference instances in ≤48 hours across five narrative-artifact layers (methodology docs / code docstrings / templated user-facing copy / orphan dependency / test fixture vs. classification / incomplete pattern translation). Pattern-064-adjacent sibling at the narrative layer: where Pattern-064 names code-that-looks-live-but-does-nothing, Pattern-073 names narrative-that-asserts-a-contract-the-code-doesn’t-honor. Operational recognition discipline:
doc-sync-sweep v0.1 skill at .claude/skills/doc-sync-sweep/. Promotion-to-Proven contingent on one more independent instance within 14 days + skill operating cleanly on a fresh-fix flow by an agent who didn’t draft it.
CIO carries the catalog-management authority; full index updates per category live below.
How to read this catalog: A code-side citation count alone does not determine whether a pattern is load-bearing. Many foundational patterns (Repository, Service, Factory) are followed implicitly throughout the codebase but rarely cited by name in code. Diagnostic-vocabulary patterns (e.g., Pattern-062 Assembly Assumption) are used in conversation and review work, not in code annotations. See the ADR catalog citation framework for the full status taxonomy (Load-bearing-interface / Load-bearing-decision / Internalized / Archival / Genuinely decorative). Same framework applies here.
Pattern Categories
Core Architecture Patterns
Foundational patterns for system architecture and infrastructure
Data & Query Patterns
Data access, session management, and query abstraction patterns
AI & Intelligence Patterns
AI coordination, spatial intelligence, and language model integration patterns
External system integration, plugin architecture, and platform connectivity patterns
Infrastructure & Scheduling Patterns
Background job execution, periodic tasks, and infrastructure coordination patterns
Development & Process Patterns
Development methodology, quality assurance, and workflow management patterns
Grammar Application Patterns
Patterns for maintaining grammar awareness (Entity/Moment/Place) throughout features
Pattern Development
Creating New Patterns
See pattern-000-template.md for the standard template when creating new patterns.
Pattern Status Levels
- Proven: Established patterns with extensive usage
- Emerging: New patterns with initial validation
- Experimental: Patterns under active development
- Deprecated: Legacy patterns being phased out
- Methodology-Elevated: A Proven pattern whose underlying principle has graduated out of the pattern catalog and into the methodology corpus as a named principle or practice. The pattern entry remains as the catalog record of origin; the methodology entry becomes the canonical, actively-cited form. This is a promotion, not a deprecation — it marks a diagnostic vocabulary item maturing into a standing principle the cohort applies by default. Exemplar: Pattern-062 (Assembly Assumption) → Excellence Flywheel v2.0 Practice 5 (“Audit the Composition”). Retroactive candidates: Pattern-045 (Green Tests, Red User), Pattern-049 (Audit Cascade) — both diagnostic-vocabulary patterns that now function as standing methodology principles. (Term coined in the Pattern Sweep 2.0 report, 2026-05-09; formalized 2026-05-28 per PM ratification. Recognition criterion: the pattern’s principle appears as a named entry/practice in
docs/internal/development/methodology-core/ or the Excellence Flywheel, and the methodology form is the one the cohort cites.)
Product Relevance Classification
Every pattern should include a Product Relevance classification:
- Process-only: Useful for building Piper, not applicable to Piper’s users
- Portable: Methodology that could become user-facing capability
- Converged: Pattern already implemented as product feature
When in doubt, default to “Process-only” — it can be reclassified later.
Pattern Sweep Process
Pattern Sweeps occur every 6 weeks per the staggered audit calendar.
Sweep Checklist (Product Relevance)
In addition to standard sweep activities, each sweep should:
Sweep Output Template Addition
Each Pattern Sweep results document should include:
## Product Relevance Summary
- Portable patterns awaiting product review: [list]
- Patterns reclassified this sweep: [list]
- Converged patterns (methodology → product): [list]
Pattern Family Index
The Pattern Family Index organizes patterns by how they work together. While this index (README.md) groups patterns by category (what they’re about), the family index groups them by family (patterns that reinforce each other). Apply patterns in family units for best results.
Proto-Pattern Registry
The Proto-Pattern Registry tracks emerging pattern candidates that need additional evidence before formalization. Proto-patterns are evaluated during 6-week pattern sweeps and promoted when they meet the evidence threshold.
Anti-Pattern Index
The Anti-Pattern Index provides bidirectional navigation between patterns and their associated anti-patterns:
- 43 anti-patterns across 5 categories (Grammar, Testing, Architecture, Process, Integration)
- Links from anti-patterns → patterns that address them
- Reverse index: patterns → anti-patterns they prevent
- Updated during 6-week pattern sweeps
Use the anti-pattern index to:
- Understand common pitfalls related to a pattern
- Find patterns that address a known problem
- Discover traps to avoid when implementing a pattern
Navigation
Legacy References
Last updated: March 3, 2026
Added Human-AI Collaboration Referee (061); total now 62 patterns