From playbook-dev
This skill should be used when the user asks to "design cross-stage enrichment", "create bi-directional updates", "enrich earlier outputs", "link later findings to earlier analysis", "update Stage 1 from Stage 3", or needs patterns for later stages to enrich earlier stage outputs.
How this skill is triggered — by the user, by Claude, or both
Slash command
/playbook-dev:bi-directional-enrichmentThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Design patterns for later stages to enrich earlier outputs, creating a connected knowledge graph.
Design patterns for later stages to enrich earlier outputs, creating a connected knowledge graph.
Traditional pipelines flow forward only:
Stage 1 → Stage 2 → Stage 3 → Stage 4
Bi-directional enrichment adds backward connections:
Stage 1 outputs ←── Enriched by ──→ Stage 3 findings
↑ │
└─────────── Validated by ───────────┘
Reserve sections in early stage outputs for later stages to populate:
# Stage 1 Template
## Original Analysis
{Stage 1 content}
## Code Bug Analysis (Enriched by Stage 3)
> **Note:** This section is initially empty. Stage 3 will add details.
**Bug identified:** [To be added by Stage 3]
**Location:** [File:line to be added by Stage 3]
**Problem:** [Description to be added by Stage 3]
**Fix:** [Recommendation to be added by Stage 3]
**Expected impact:** [Estimate to be added by Stage 3]
Add attribution confidence from later analysis:
# Stage 1: Comparison Analysis
## Performance Delta
Current: 75%
Baseline: 60%
Delta: +15 points
## Attribution (Added by Stage 3)
> **Note:** Added by Stage 3 Change Impact Analysis
**Primary attribution:** Change X (Strong confidence)
**Evidence:** {Trace marker found in artifacts}
**Secondary factors:** {Other contributing changes}
Add links to related later-stage analysis:
# Stage 1: Capability Gaps
## Gap: Missing database inspection
...
## Related Analysis (Added by Stage 3)
**Opportunity:** See `opportunity-analysis/1-tool-capabilities.md#database-inspection`
**Impact estimate:** +12 points across 8 items
**Priority:** P1
Mark whether hypotheses were validated:
# Stage 1: Diagnosis
## Failure Hypothesis
Suspected code bug in error handling.
## Validation Status (Added by Stage 3)
**Status:** ✅ Confirmed
**Evidence:** Code inspection found bug at `handler.go:245`
**Fix PR:** #1234
Identify where later stages add value to earlier outputs:
## Enrichment Map
| Stage 1 File | Enriched By | Section Added |
|--------------|-------------|---------------|
| diagnosis.md | Stage 3 | Code bug details |
| capability-gaps.md | Stage 3 | Validation status, priority |
| comparison-analysis.md | Stage 3 | Attribution confidence |
Add placeholder sections in Stage 1 templates:
## {Section Name} (Enriched by Stage {N})
> **Note:** This section is initially empty. Stage {N} ({Name}) will add {what}.
**Field 1:** [To be added by Stage {N}]
**Field 2:** [To be added by Stage {N}]
In Stage 3 instructions, specify how to enrich:
## Stage 1 Enrichment
After completing {analysis}, enrich Stage 1 files:
### diagnosis.md Enrichment
**When:** After Vector 3 (Code Fixes) identifies code bugs
**Add to "Code Bug Analysis" section:**
- Bug identified: {from Vector 3 analysis}
- Location: {file:line from code inspection}
- Problem: {description of the bug}
- Fix: {recommended fix}
- Expected impact: {from impact estimation}
**Process:**
1. For each item with identified code bug
2. Read item's diagnosis.md
3. Fill in the Code Bug Analysis section
4. Write updated file
Check that enrichment occurred:
# Verify enrichment sections populated
grep -l "To be added by Stage" items/*/analysis/*.md
# Should return empty if all enriched
## Code Bug Analysis (Enriched by Stage 3)
**Bug identified:** Yes
**Location:** `{file}:{line}`
**Problem:** {Description of the bug}
**Fix:** {How to fix it}
**Expected impact:** +{N} points on quality score
**Reference:** See `opportunity-analysis/3-check-improvements.md#{anchor}`
## Change Attribution (Added by Stage 3)
**Attribution confidence:** {Strong/Moderate/Weak/None}
**Primary change:** {Change name}
- Commit: {hash}
- Evidence: {What trace markers found}
**Secondary factors:**
- {Factor 1}
- {Factor 2}
**Reference:** See `change-impact-analysis/investigation-impact.md#{item-id}`
## Validation Status (Added by Stage 3)
**Hypothesis:** {Original Stage 1 hypothesis}
**Status:** {✅ Confirmed / ❌ Refuted / ⚠️ Partially confirmed}
**Evidence:**
- {Evidence point 1}
- {Evidence point 2}
**Updated assessment:** {Revised conclusion if needed}
## Priority Assessment (Added by Stage 3)
**Aggregate priority:** P{1/2/3}
**Reasoning:**
- Impact: +{N} points
- Items affected: {count}
- Effort: {Low/Medium/High}
**Roadmap placement:** {Quarter/Phase}
**Reference:** See `opportunity-analysis/cross-vector-analysis.md#roadmap`
Always mark enrichment sections clearly:
## {Section} (Enriched by Stage N)
> **Note:** {Explanation of what this section contains and who adds it}
Include references to where the enrichment came from:
**Reference:** See `{path-to-source}#{anchor}`
Never modify original Stage 1 content - only add to enrichment sections.
If Stage 3 doesn't have relevant information:
## Code Bug Analysis (Enriched by Stage 3)
**Bug identified:** No code bug identified
**Notes:** Issue appears to be reasoning/synthesis related, not code bug.
Before finalizing enrichment design:
For enrichment patterns:
${CLAUDE_PLUGIN_ROOT}/references/playbook-pattern.md - Section 2.6npx claudepluginhub tgvashworth/agent-plugins --plugin playbook-devCross-references structured review findings against an implementation plan, classifies each finding into an action category, applies concrete edits, and produces a traceability summary.
Generates adaptive-depth session retrospective reports (retro.md) from plan.md and lessons.md, converting outcomes into persistent process improvements. Supports deep/light modes and directory resolution logic.
Proposes research-backed changes to knowledge systems by analyzing health reports, friction patterns, and derivation history. Requires approval before implementation.