playbook-ops

name	playbook-ops
description	Generate task playbooks and explain patterns using learned strategies from episodic memory

Playbook Operations

When to Use

• Generating step-by-step playbooks for a new task
• Getting actionable recommendations based on past success
• Explaining what a pattern means in human-readable form
• Understanding when/when-not to apply a strategy

CLI Commands

Command	Purpose
do-memory-cli playbook recommend [TASK]	Generate playbook for a task
do-memory-cli playbook explain <PATTERN_ID>	Explain a pattern

MCP Tools

Tool	Parameters	Purpose
recommend_playbook	task, domain, task_type, max_steps, language, framework, tags	Generate actionable playbook
explain_pattern	pattern_id	Human-readable pattern explanation
search_patterns	query, domain, limit	Find relevant patterns
recommend_patterns	task_type, domain, context	Get pattern recommendations

Generating a Playbook

# Basic recommendation
do-memory-cli playbook recommend "implement rate limiting middleware"

# With context
do-memory-cli playbook recommend "add caching layer" \
  --domain "rust-backend" \
  --task-type "code_generation" \
  --language "rust" \
  --framework "axum" \
  --max-steps 7

Playbook Output Includes

• Task match score: How well the playbook fits your request
• Confidence: Based on past episode success rates
• Steps: Ordered actions with tool hints and expected results
• Pitfalls: Known failure modes to avoid
• When to apply / When NOT to apply: Applicability conditions
• Expected outcome: What success looks like

Explaining Patterns

do-memory-cli playbook explain "pat-uuid-here"

Returns a natural-language explanation of:

• What the pattern does
• When it was discovered
• Success rate and contexts where it works

Task Types

Valid --task-type values:

• code_generation — Building new features
• debugging — Finding and fixing bugs
• refactoring — Restructuring code
• testing — Writing or fixing tests
• analysis — Code review and analysis
• documentation — Writing docs

Integration

Playbooks work best when the memory system has episodes:

1. Complete episodes with outcomes and steps
2. System extracts patterns from successful episodes
3. recommend_playbook uses patterns to generate actionable steps
4. Feedback loop (via recommendation-feedback skill) improves recommendations