All skills
simota avatar

/lore

@35ffd55
by shingo imotasimota/agent-skills85 stars
15

Curating cross-agent knowledge and institutional memory: extracts patterns from agent journals into METAPATTERNS.md, detects knowledge decay, propagates best practices. Use for memory curation.

Use this Skill: https://skilld.dev/gh/simota/agent-skills/lore

This session only. Nothing lands on disk.

referenceofficial-pattern-taxonomy.md

≈1.6k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Official Pattern Taxonomy Reference

Source: "The Complete Guide to Building Skills for Claude" (Anthropic, 2025)

Official pattern integration reference that Lore consults during the CATALOG / PROPAGATE phases.


1. Integrated Mapping of Official Patterns to Lore Pattern Classification

Official 5 Patterns → Lore Taxonomy Conversion

Official Pattern Lore Domain Lore Type Default Confidence Scope
Sequential Workflow Orchestration PROCESS SUCCESS ESTABLISHED ECOSYSTEM
Multi-MCP Coordination INFRA SUCCESS ESTABLISHED ECOSYSTEM
Iterative Refinement PROCESS HEURISTIC ESTABLISHED ECOSYSTEM
Context-Aware Tool Selection APP HEURISTIC ESTABLISHED ECOSYSTEM
Domain-Specific Intelligence DESIGN SUCCESS ESTABLISHED ECOSYSTEM

Official patterns start at the ESTABLISHED level based on Anthropic's observations. They can be promoted to FOUNDATIONAL once enough evidence accumulates within the ecosystem.

Official 3 Use-Case Categories → Lore Domain Conversion

Official Category Primary Lore Domain Secondary Domain
Document & Asset Creation APP DESIGN
Workflow Automation PROCESS INFRA
MCP Enhancement INFRA APP

2. Integrating Official Quality Signals with Lore Evidence Classification

Quantitative Signals

Official Metric Lore Evidence Type Threshold Mapping
Trigger rate 90%+ Execution evidence ≥ 90% auto-load SUCCESS pattern if met; FAILURE if consistently below
0 failed API calls Execution evidence 0 failures per workflow SUCCESS if met; ANTI if consistently failing
Workflow efficiency (token reduction) Performance evidence Baseline comparison TRADEOFF pattern with quantitative data

Qualitative Signals

Official Metric Lore Evidence Type Assessment
No next-step prompting needed User behavior evidence SUCCESS when observed; FAILURE as anti-pattern
Correction-free execution Consistency evidence 3-5 identical runs → PATTERN confidence
First-try accessibility Usability evidence New user feedback → EMERGING then PATTERN

3. Integrating Official Iteration Signals with Lore Decay Detection

Undertriggering → Knowledge Gap Detection

Official Signal Lore Mapping Action
Skill doesn't load when expected FAILURE pattern candidate Register as META-FAILURE-NNN
Users manually enabling skills Usability gap evidence Propagate to Sigil (description improvement)
Support questions about usage Knowledge gap signal Propagate to Architect (trigger guidance review)

Overtriggering → Anti-Pattern Detection

Official Signal Lore Mapping Action
Skill loads for irrelevant queries ANTI pattern candidate Register as META-ANTI-NNN
Users disabling skills Negative evidence Propagate to Sigil (negative trigger addition)
Confusion about purpose Design gap signal Propagate to Architect (scope clarification)

Execution Issues → Failure Pattern Detection

Official Signal Lore Mapping Action
Inconsistent results FAILURE pattern candidate Cross-reference with other agent journals
API call failures INFRA-FAILURE-NNN Propagate to Mend (remediation pattern)
User corrections needed Instruction quality gap Propagate to Sigil (instruction improvement)

4. Official Pattern Cross-Check Rules During the CATALOG Phase

Official Cross-Check When Registering a New Pattern

  1. Pre-classification check: Confirm whether the new pattern candidate matches one of the official 5 patterns
  2. If it matches: Register it as a variant of the official pattern (append a -V suffix to the ID)
  3. If it doesn't match: Register it as a normal new pattern
  4. If it conflicts: Explicitly record the difference from the official pattern, and promote it to an independent pattern once evidence count ≥ 3

Pattern Quality Assessment Against Official Criteria

Periodically assess whether registered patterns satisfy the following official criteria:

Criterion Check Source
Progressive Disclosure compliance Does the pattern reflect the 3-level structure? Official Guide §1
Description quality Does it include a WHAT+WHEN structure? Official Guide §2
Testability Can it be verified across the 3 Areas (Triggering/Functional/Performance)? Official Guide §3
Error handling Does it address the 6 troubleshooting categories? Official Guide §5

5. Official Criteria Distribution Rules During the PROPAGATE Phase

Relationship Between Distribution Targets and Official Criteria

Consumer Agent Relevant Official Knowledge Propagation Trigger
Sigil Description writing rules, instruction structure, test methodology When a decline in skill-generation quality patterns is detected
Architect The 5 patterns, success-criteria framework, Progressive Disclosure When a decline in new-agent design quality patterns is detected
Gauge Frontmatter validation spec, 6 troubleshooting categories When a decline in audit accuracy patterns is detected
Darwin Quality signals (quantitative/qualitative), iteration signals When an opportunity to improve EFS assessment accuracy is detected
Nexus The 3 use-case categories, pattern selection guide When a decline in routing accuracy patterns is detected

Official Criteria Reference Format in LORE_INSIGHT

LORE_INSIGHT:
  Pattern: [ID]
  Official_Alignment: [ALIGNED | VARIANT | NOVEL | CONTRADICTS]
  Official_Reference: "The Complete Guide to Building Skills for Claude" §[section]
  Evidence: [agent, date, context]
  Implication: [what this means for the consumer]

Source: SKILL.md on GitHub

No alerts13d5 checks · Risk SAFE
  • Gen Agent Trust Hub13d

    Lore is a sophisticated knowledge curation and institutional memory skill designed to extract, synthesize, and manage patterns from agent journals and logs. It incorporates advanced defensive mechanisms against knowledge decay and information corruption, including confidence-based promotion rules and bi-temporal validity tracking, ensuring that the curated knowledge base remains accurate and actionable.

  • Socket13d

    No alerts

  • Snyk13d

    Risk: LOW · No issues

  • Runlayer6mo

    1/6 files flagged

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 35ffd55. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 3 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/lore