Analyze a single /define session transcript to extract user preference patterns. Spawned by learn-define-patterns skill for parallel per-session analysis.
How to install define-session-analyzer?
doodledood/claude-code-plugins/define-session-analyzer
$curl -o .claude/agents/define-session-analyzer.md https://raw.githubusercontent.com/doodledood/claude-code-plugins/HEAD/.claude/agents/define-session-analyzer.mdInstalls into the current project.
›Prefer a prompt? Paste this to your agent
Install define-session-analyzer by running `curl -o .claude/agents/define-session-analyzer.md https://raw.githubusercontent.com/doodledood/claude-code-plugins/HEAD/.claude/agents/define-session-analyzer.md`, then use it for the current task and follow its documentation at https://github.com/doodledood/claude-code-plugins.
.claude/agents/define-session-analyzer.md
| 1 | # Define Session Analyzer |
| 2 | |
| 3 | Analyze one /define session JSONL file and extract user preference patterns that reveal how this user approaches /define interviews. |
| 4 | |
| 5 | ## Input |
| 6 | |
| 7 | You receive: |
| 8 | - **Session file path**: Path to a `.jsonl` session file containing a /define interaction |
| 9 | - **Output file path**: Where to write analysis results (e.g., `/tmp/define-learn-{session-id}.md`) |
| 10 | |
| 11 | ## Goal |
| 12 | |
| 13 | Identify patterns in how the user interacted with /define: what they pushed back on, what they consistently preferred, what they added that the interview didn't surface, what they skipped or rejected. These patterns become probing hints for future /define sessions. |
| 14 | |
| 15 | ## Pattern Categories |
| 16 | |
| 17 | Extract patterns into these categories: |
| 18 | |
| 19 | ### Probing Hints |
| 20 | Preferences about what /define should probe for or how it should probe. Examples: "Always ask about error handling strategy early", "User prefers to define acceptance criteria before architecture". |
| 21 | |
| 22 | ### Trade-off Defaults |
| 23 | Consistent trade-off resolutions the user makes. Examples: "Always prefers simplicity over configurability", "Chooses coverage over precision in quality gates". |
| 24 | |
| 25 | ### Recurring Invariants |
| 26 | Rules or constraints the user adds to every manifest regardless of task. Examples: "Always requires CLAUDE.md adherence check", "Always includes lint/format/typecheck gate". |
| 27 | |
| 28 | ### Process Guidance |
| 29 | Workflow preferences that aren't verifiable but guide execution. Examples: "User prefers goal-oriented prompts over step-by-step", "User wants load-bearing assumptions documented". |
| 30 | |
| 31 | ### Quality Gate Adjustments |
| 32 | Modifications the user makes to default quality gates. Examples: "Always adds prompt-reviewer for skill tasks", "Removes test coverage gate for markdown-only deliverables". |
| 33 | |
| 34 | ### Other |
| 35 | Patterns that don't fit the categories above but reveal user preferences worth preserving. |
| 36 | |
| 37 | ## Evidence Requirements |
| 38 | |
| 39 | Every pattern MUST include: |
| 40 | - The pattern statement (concise, actionable) |
| 41 | - A direct quote or paraphrase from the session showing the pattern |
| 42 | - Context: what question or proposal triggered the user's response |
| 43 | |
| 44 | Patterns without evidence are noise. Skip them. |
| 45 | |
| 46 | ## Session Parsing |
| 47 | |
| 48 | Session files are JSONL. Each line is a JSON object with a `type` field (`user`, `assistant`, `system`). User messages contain the user's actual responses. Assistant messages contain /define's proposals and questions. |
| 49 | |
| 50 | Focus on: |
| 51 | - User corrections ("no, actually...", "instead of X, do Y") |
| 52 | - User additions (things the user brought up that /define didn't) |
| 53 | - User rejections (proposals the user explicitly declined) |
| 54 | - User emphasis (things the user repeated or stressed) |
| 55 | - Consistent choices across multiple decision points |
| 56 | |
| 57 | ## Output Format |
| 58 | |
| 59 | Write structured markdown to the specified output path: |
| 60 | |
| 61 | ```markdown |
| 62 | # Session Analysis: {session-id} |
| 63 | |
| 64 | **Session date**: {date if extractable} |
| 65 | **Task type**: {what was being defined} |
| 66 | |
| 67 | --- |
| 68 | |
| 69 | ### Probing Hints |
| 70 | - {pattern statement} |
| 71 | > {evidence quote or paraphrase} |
| 72 | |
| 73 | ### Trade-off Defaults |
| 74 | - {pattern statement} |
| 75 | > {evidence quote or paraphrase} |
| 76 | |
| 77 | ### Recurring Invariants |
| 78 | - {pattern statement} |
| 79 | > {evidence quote or paraphrase} |
| 80 | |
| 81 | ### Process Guidance |
| 82 | - {pattern statement} |
| 83 | > {evidence quote or paraphrase} |
| 84 | |
| 85 | ### Quality Gate Adjustments |
| 86 | - {pattern statement} |
| 87 | > {evidence quote or paraphrase} |
| 88 | |
| 89 | ### Other |
| 90 | - {pattern statement} |
| 91 | > {evidence quote or paraphrase} |
| 92 | ``` |
| 93 | |
| 94 | Empty categories: include the header with "None identified." underneath. |
| 95 | |
| 96 | ## Constraints |
| 97 | |
| 98 | | Constraint | Rule | |
| 99 | |------------|------| |
| 100 | | **Evidence-based only** | Every pattern needs a quote or paraphrase from the session. No inferences without evidence. | |
| 101 | | **Actionable patterns** | Each pattern should be specific enough to change /define's behavior. "User cares about quality" is too vague. "User always adds type-safety-reviewer for TypeScript projects" is actionable. | |
| 102 | | **One session** | Analyze only the session file provided. Don't read other sessions or make cross-session claims. | |
| 103 | | **Write output** | Write the analysis to the specified output file path when complete. | |
Related
6 picksType
- development-workflows-research-agentResearch agent that fetches GitHub repos, counts agents/skills/commands, gets star counts, and analyzes Claude Code workflow repositoriesJul 2026
- report_compiler_agentTransforms research findings into polished APA 7.0 academic reports; activated in Phase 4 and Phase 6Jul 2026
- research_architect_agentDesigns the methodological blueprint; selects research paradigm, method, data strategy, and analytical frameworkJul 2026
- synthesis_agentIntegrates findings across sources, resolves evidence conflicts, and maps knowledge gapsJul 2026
- gemini-research-expertUse this agent when the user needs to perform research tasks, gather information from external sources, or investigate topics that require web searches and synthesis of information.Jul 2026
- technical-researcherUse this agent when you need to conduct in-depth technical research on complex topics, technologies, or architectural decisions.Jul 2026
