context-recall

Recall relevant past context before making new decisions or starting strategic work.

Vision to Value Phase

Cross-phase - This skill is used before work in any phase to surface relevant history.

Prerequisites: Context registry populated (decisions, bets, learnings saved) Outputs used by: Decision-making at any phase

Purpose

Before making a decision that may have been made before, or starting work in an area with existing strategic context, query the context registry to surface relevant history.

When to Use

Invoke /context-recall [topic] when:

• Starting work on a new decision
• Beginning strategic planning in an area
• Checking if similar decisions exist
• Understanding constraints from past decisions
• Finding relevant learnings to apply

Process

1. Parse the Query

Accept a topic or keyword from the user, with optional filters:

• /context-recall pricing → Search for pricing-related context
• /context-recall enterprise → Search for enterprise-related context
• /context-recall pricing product:AXIA → Search filtered to AXIA product
• /context-recall pricing --include-demo → Include demo data even if production data exists
• /context-recall pricing --demo-only → Show only demo data

Filters:

• product:[name] - Filter to specific product
• --include-demo - Include demo data (marked with [DEMO])
• --demo-only - Show only demo data (for testing/learning)

1b. Check for Production Data (Demo Filtering)

Before searching, determine if production data exists:

1. Check main context folders (NOT context/demo/):

   • context/decisions/ - Any non-demo decision files?
   • context/bets/ - Any non-demo bet files?
   • context/feedback/index.md - Any non-demo entries?

2. Apply demo filtering rule:

   Production Data?	Flag	Behavior
   No	(any)	Include demo with [DEMO] markers
   Yes	(none)	Exclude demo data, show excluded count
   Yes	--include-demo	Include demo with [DEMO] markers
   (any)	--demo-only	Only demo data

3. Demo data is identified by:

   • Path contains context/demo/
   • ID contains "DEMO" (e.g., DR-DEMO-001)
   • JSON entry has demo: true

2. Search JSON Index First (Fast Path)

Primary search method: Use context/index.json for fast topic-based retrieval.

// Read context/index.json
{
  "topicIndex": {
    "pricing": ["DR-2026-001", "DOC-2026-015"],
    "api": ["DR-2026-003", "DOC-2026-008"]
  }
}

1. Check topicIndex for exact topic match
2. Check typeIndex if user specifies a type (e.g., "prd", "decision")
3. Check phaseIndex if user specifies a phase
4. Retrieve matching entries by ID
5. Get file paths from entries for full document access

If JSON index doesn't exist or has no matches, fall back to markdown index search.

3. Search Markdown Indexes (Fallback)

If JSON index has no matches, search across markdown indexes:

Decisions (context/decisions/index.md)

• Search titles and tags
• Return matching decision summaries

Strategic Bets (context/bets/index.md)

• Search titles and key assumptions
• Return matching bet summaries

Assumptions (context/assumptions/registry.md)

• Search assumption text
• Return matching assumptions with status

Learnings (context/learnings/index.md)

• Search learning text and tags
• Return matching learnings

Active Portfolio (context/portfolio/active-bets.md)

• Check if topic relates to active bets
• Surface current strategic context

Strategic Documents (context/documents/index.md)

• Search titles, type, skill, and auto-generated tags
• Match on document type (PRD, roadmap, GTM strategy, etc.)
• Return matching documents with file location

3. Read Full Records if Needed

For highly relevant matches:

• Read the full decision record or bet document
• Extract key rationale and constraints
• Note re-decision triggers

4. Synthesize Results

Present findings in order of relevance. Apply demo data formatting:

## Context Recall: [Topic]

### Relevant Decisions
| ID | Title | Status | Key Rationale |
|----|-------|--------|---------------|
| DR-2026-001 | [Title] | Accepted | [Why this was decided] |
| `[DEMO]` DR-DEMO-001 | Demo Title | Demo | Demo rationale |

**Constraints from these decisions:**
- [Constraint 1]
- [Constraint 2]

### Related Strategic Bets
| ID | Title | Status | Relevance |
|----|-------|--------|-----------|
| SB-2026-002 | [Title] | Active | [How it relates] |

### Tracked Assumptions
| ID | Assumption | Status |
|----|------------|--------|
| A-005 | [Assumption text] | Pending |

### Applicable Learnings
| ID | Learning | Confidence |
|----|----------|------------|
| L-012 | [Learning text] | High |

### Related Strategic Documents
| ID | Title | Type | Date | Location |
|----|-------|------|------|----------|
| DOC-2026-001 | [Title] | [Type] | [Date] | [File path] |

*Tip: Read the document at [location] for full details*

### Recommendations
Based on past context:
- [Recommendation 1]
- [Recommendation 2]

---
*[If demo excluded]: X demo results excluded. Use `--include-demo` to show.*
*[If demo included]: Results include demo data marked with `[DEMO]`.*

💬 Related conversations may exist. Use `/interaction-recall [topic]` to search past interactions.

Demo Display Rules:

• Demo entries: Prefix ID with [DEMO] marker
• If demo excluded: Show count at bottom with flag hint
• If demo included: Note that demo data is present

5. Flag Conflicts or Risks

If the current work might conflict with past decisions:

• Highlight the potential conflict
• Reference the re-decision triggers
• Suggest whether a re-decision process is needed

Instructions

1. Accept topic/keyword from user (required)
2. Parse optional product:[name] filter from query
3. Read all context index files (decisions, bets, assumptions, learnings, portfolio, documents)
4. Search for matches (case-insensitive, partial match OK)
5. If product filter specified, filter results to that product only
6. For strong matches, read full records or referenced documents
7. Synthesize and present findings
8. Highlight constraints that apply to current work
9. Note any potential conflicts with past decisions
10. If no matches found, say so clearly

Example Usage

User: /context-recall pricing

Agent: Searches context/decisions/index.md, context/bets/index.md,
       context/assumptions/registry.md, context/learnings/index.md,
       context/portfolio/active-bets.md, context/documents/index.md

       Returns relevant matches with summaries and recommendations.

User: /context-recall roadmap product:AXIA

Agent: Searches all context indexes filtered to AXIA product only.
       Returns AXIA-specific roadmap documents, decisions, and bets.

Enhanced Recall (v3)

When querying the context registry:

1. Check auto-context: See if rules/auto-context.md triggers apply
2. Search topic indexes: Use topicIndex, productIndex, phaseIndex from context/index.json
3. Follow cross-references: For each direct match, include linked items one level deep (see rules/context-graph.md)
4. Attribute links: Show why related items appear: [via: DR-2026-003]
5. Limit expansion: Max 3 cross-referenced items per direct match