Pattern Harvester

Use this when you want to scan across engagement journals to find emerging patterns worth codifying into your knowledge system. Run monthly or quarterly.

---

How it works

1. You trigger the skill (optionally scoping to specific engagements or a time range)
2. The skill reads all engagement journals and cross-references the existing pattern index
3. It produces a pattern report with new patterns to add, existing patterns with new evidence, and a watch list

---

Prompt

You are mining Kate Makrigiannis's engagement journals for reusable patterns. Your job is to find the signal that crosses engagement boundaries: things that happen more than once, in different contexts, and might become part of Kate's methodology.

Inputs I will provide:

• Scope (optional): {{SCOPE}} (specific engagement slugs, or "all" for everything. Defaults to all.)
• Time range (optional): {{RANGE}} (e.g., "last 90 days," "Q1 2026." Defaults to all time.)
• Focus (optional): {{FOCUS}} (a specific theme to look for, e.g., "prioritization failures," "stakeholder alignment")

Step 1: Read the journals

Read all files in journals/ that match the scope and time range. Skip language ledger files (those prefixed with language-). Extract all entries, noting:

• Entry type (decision-made, pattern-observed, outcome-achieved, etc.)
• Engagement slug
• Date
• Content

Step 2: Read the existing pattern index

Read the reusable patterns section of knowledge/engagement-history.md. Build a list of already-codified patterns with their names and descriptions.

Step 3: Find patterns

Analyze journal entries across engagements looking for:

Cross-Engagement Patterns (appear in 2+ engagements)

Things that show up independently in different contexts. These are the most valuable because they suggest something structural, not coincidental.

For each:

• Pattern name: (give it a clear, reusable name)
• Description: What keeps happening?
• Evidence: Which engagements, which journal entries?
• Why it matters: What does Kate do about it when she sees it?
• Already indexed? Does this pattern already exist in engagement-history.md?

Existing Patterns with New Evidence

Patterns already in the index that now have additional supporting data from recent journals.

For each:

• Pattern name: (from the existing index)
• New evidence: Which engagement, what happened?
• Update snippet: Ready-to-paste addition to the existing pattern entry

Watch List (single-occurrence patterns)

Things that appeared in only one engagement but seem significant enough that Kate should watch for them again.

For each:

• Observation: What happened?
• Engagement: Where?
• Why it might be a pattern: What would make this significant if it appeared again?

Step 4: Produce the pattern report

Pattern Harvest | {{Date}}

Scope: {{what was scanned}} | Journals scanned: {{count}} | Entries analyzed: {{count}}

New Patterns to Codify

For each new cross-engagement pattern, provide a ready-to-paste snippet in the format used by engagement-history.md's reusable patterns section:

[ADD TO PATTERNS]
### {{Pattern Name}}
{{Description}}
**Seen in:** {{engagement1}}, {{engagement2}}
**What to do:** {{Kate's response when she encounters this}}

Pattern Updates

For each existing pattern with new evidence:

[UPDATE PATTERN: {{Pattern Name}}]
Add to evidence: {{engagement}} — {{what happened}}

Watch List

Observation	Engagement	Why It Might Matter
{{observation}}	{{slug}}	{{reasoning}}

Meta-Observations

1-2 sentences on the overall health of Kate's pattern library:

• Is it growing? Stagnating?
• Are there domains or service types that are well-covered vs. blind spots?
• Should any existing patterns be retired or merged?

---

[ADD TO PATTERNS]
### The Proxy Sponsor Problem
A senior stakeholder nominates a direct report to serve as project sponsor,
then resurfaces during delivery to override decisions made in their absence.
This creates whiplash for the team and erodes trust with the named sponsor.
**Seen in:** meridian-health, corven-retail
**What to do:** In kickoff, explicitly contract with the named sponsor on decision
rights and escalation paths. If a "silent executive" is in the background,
name them and establish a single point of contact protocol before work begins.

[ADD TO PATTERNS]
### Data Confidence Mismatch
The client presents existing data assets as "ready to use" during scoping,
but early discovery reveals significant quality, access, or ownership issues
that were unknown to the business stakeholder who made the claim.
**Seen in:** softlayer-dx, nbc-logistics
**What to do:** Build a 3–5 day data audit into every engagement that involves
analytics or reporting deliverables. Do not begin analysis work until the audit
closes. Flag scope risk to sponsor immediately if issues surface.

[UPDATE PATTERN: Scope Creep via Slack]
Add to evidence: corven-retail — On March 3, the VP of Merchandising added three
"quick questions" to a Slack thread that collectively represented ~12 hours of
unscoped analysis. Kate redirected to a formal change request; client accepted
without friction once the ask was made explicit. Reinforces that the channel
matters: informal tools invite informal asks.

[UPDATE PATTERN: Workshop Energy Collapse (Day 2 Afternoon)]
Add to evidence: meridian-health — Four-hour working session on Feb 19 hit a wall
at the 2:45 PM mark after a catered lunch. Two key decision-makers disengaged.
Subsequent reframe exercise fell flat. Restructuring Day 2 to front-load decisions
before lunch continues to be the right mitigation.