/pm-skill-discover

Detect workflow patterns from AI sessions and suggest reusable skills.

---

What It Does

Analyzes your recent AI development sessions to detect repeated workflow patterns and suggests new skills you could create to automate them. Includes:

• Claude Code session transcript analysis (tool usage patterns)
• Git history analysis (file-group patterns, commit frequency)
• jaan-to learning file analysis (skill usage frequency)
• 4-dimension scoring rubric (frequency, time saved, parameterizability, risk)
• Matching against 10 known workflow archetypes from research
• Optional auto-invocation of /skill-create for selected patterns

---

Usage

/pm-skill-discover
/pm-skill-discover --days=7
/pm-skill-discover --days=30 --min-frequency=5 --max-suggestions=3

---

Parameters

Flag	Default	Description
--days=N	14	Number of days to analyze
--min-frequency=N	3	Minimum pattern occurrences to surface
--max-suggestions=N	5	Maximum suggestions to present

---

Data Sources

Source	What It Extracts
Claude Code sessions	Tool names, result status, timestamps (structural metadata only)
Git history	File groups, commit frequency, message categories
jaan-to learn files	Skill usage frequency, accumulated lesson counts

Privacy: Only structural metadata is extracted. No raw code, prompts, file paths, or variable values are stored or displayed.

---

Scoring Rubric

Each candidate pattern is scored on 4 dimensions:

Dimension	Weight	What It Measures
Frequency	30%	How often the pattern occurs per week
Time Saved	30%	Estimated duration x frequency
Parameterizability	25%	Ratio of variable to fixed steps
Risk	15%	Inverse of destructive operations

Patterns scoring above 40/100 are surfaced as candidates.

---

Known Archetypes

Detected patterns are matched against 10 research-backed workflow archetypes:

1. Error diagnosis and fix cycle
2. Red-green-refactor loop
3. CI pipeline repair
4. Dependency update workflow
5. Code review response pattern
6. Feature scaffolding
7. Migration execution
8. API integration
9. Merge conflict resolution
10. Post-deployment verification

---

Output

Path: jaan-to/outputs/pm/skill-discover/{id}-{slug}/{id}-{slug}.md

Contains:

• Executive summary with total estimated time savings
• Data source statistics
• Discovered patterns with full scoring breakdown
• Archetype matches
• Pre-filled /skill-create commands for each candidate

---

Example

Input:

/pm-skill-discover --days=14

Preview at HARD STOP:

SKILL DISCOVERY REPORT
══════════════════════
Period: 14 days | Sessions: 23 | Episodes: 67
Patterns detected: 12 | Above threshold: 3

TOP SUGGESTIONS
───────────────
1. [Score: 87] "Error Diagnosis Cycle"
   Frequency: 12x/week | Est. savings: ~40 min/week
   Archetype: Error diagnosis and fix cycle
   Suggested skill: qa-error-fix

2. [Score: 72] "Feature File Scaffolding"
   Frequency: 5x/week | Est. savings: ~25 min/week
   Archetype: Feature scaffolding
   Suggested skill: dev-feature-scaffold

---

Tips

• Run after at least 2 weeks of active development for meaningful patterns
• Combine with /skill-create for end-to-end discovery-to-creation pipeline
• Re-run monthly to detect new patterns as workflow evolves
• Use --min-frequency=5 for high-activity repos to reduce noise
• For a full workflow-reliability conversion plan (evals, safety, autonomy, phased build order), run /pm-workflow-audit — this skill only surfaces individual skill candidates. Both share the cross-tool session reader (scripts/lib/session-reader.sh).

---

Learning

This skill reads from:

jaan-to/learn/jaan-to-pm-skill-discover.learn.md

Add feedback:

/learn-add pm-skill-discover "Check for monorepo context mixing"