@qball-inc/the-bulwark 1.0.0

package/skills/bulwark-research/SKILL.md

@@ -0,0 +1,298 @@
+---
+name: bulwark-research
+description: Structured multi-viewpoint research using 5 parallel Sonnet sub-agents. Use when deep research is needed on a complex topic before implementation planning.
+user-invocable: true
+argument-hint: "<topic, filepath, or directory>"
+skills:
+  - subagent-prompting
+---
+
+# Bulwark Research
+
+Structured multi-viewpoint research on a given topic. Spawns 5 Sonnet sub-agents in parallel, each analyzing from a distinct analytical viewpoint, then synthesizes into a single research document.
+
+---
+
+## When to Use This Skill
+
+**Load this skill when the user request matches ANY of these patterns:**
+
+| Trigger Pattern | Example User Request |
+|-----------------|---------------------|
+| Deep research | "Research agent teams", "Investigate loop detection" |
+| Topic exploration | "What do we know about X?", "Explore approaches to Y" |
+| Pre-planning research | "Before we build X, research the landscape" |
+| Multi-viewpoint analysis | "Analyze X from multiple angles" |
+
+**DO NOT use for:**
+- Evaluating implementation feasibility (use `bulwark-brainstorm`)
+- Quick fact lookup (use web search or codebase exploration)
+- Code review (use `code-review`)
+- Debugging (use `issue-debugging`)
+
+---
+
+## Dependencies
+
+| Category | Files | Requirement | When to Load |
+|----------|-------|-------------|--------------|
+| **Viewpoint definitions** | `references/viewpoint-*.md` | **REQUIRED** | Always load all 5 before spawning agents |
+| **Output templates** | `templates/viewpoint-output.md` | **REQUIRED** | Include in every agent prompt |
+| **Synthesis template** | `templates/synthesis-output.md` | **REQUIRED** | Use when writing synthesis |
+| **Subagent prompting** | `subagent-prompting` skill | **REQUIRED** | Load at Stage 1 for 4-part prompt template |
+
+**Fallback behavior:**
+- If a viewpoint reference file is missing: Note in diagnostic log, reduce to 4 agents, continue
+- If output template is missing: Use the schema from this SKILL.md directly
+
+---
+
+## Usage
+
+```
+/bulwark-research <topic-or-prompt> [--context <file>]
+/bulwark-research --doc <path-to-document>
+```
+
+**Arguments:**
+- `<topic-or-prompt>` - Free-text topic description or problem statement
+- `--context <file>` - Additional context file to provide to all agents
+- `--doc <path>` - Use a document as the topic source instead of free text
+
+**Examples:**
+- `/bulwark-research "agent teams and multi-agent orchestration"` - Research a topic
+- `/bulwark-research --doc plans/proposal.md` - Research from a document
+- `/bulwark-research "loop detection" --context docs/architecture.md` - Research with context
+
+---
+
+## Stages
+
+### Stage 1: Pre-Flight
+
+```
+Stage 1: Pre-Flight
+├── Read problem statement / document
+├── AskUserQuestion if ambiguous (iterative, 2-3 questions per round)
+├── Slugify topic for output directory
+├── Create output directories: $PROJECT_DIR/logs/research/{topic-slug}/ and $PROJECT_DIR/artifacts/research/{topic-slug}/
+├── Load subagent-prompting skill
+├── Load all 5 references/viewpoint-*.md
+├── Load templates/viewpoint-output.md
+└── Token budget check (warn if >30% consumed)
+```
+
+**AskUserQuestion Protocol (Pre-Spawn):**
+
+If the problem statement is ambiguous, under-specified, or could benefit from scope boundaries:
+
+1. Ask 2-3 clarifying questions using AskUserQuestion
+2. Assess whether the answers provide sufficient clarity to construct high-quality prompts
+3. If not, ask up to 3 more questions in a follow-up round
+4. Repeat until clarity is achieved (no hard cap on rounds, but each round is 2-3 questions max)
+5. If the problem statement is clear and well-scoped from the start, skip this step and note in diagnostics: `pre_flight_interview: skipped (problem statement sufficient)`
+
+### Stage 2: Viewpoint Analysis (5 Sonnet, Parallel)
+
+```
+Stage 2: Viewpoint Analysis
+├── Construct 5 prompts using 4-part template (GOAL/CONSTRAINTS/CONTEXT/OUTPUT)
+├── Each prompt includes:
+│   ├── Viewpoint definition from references/viewpoint-{name}.md
+│   ├── Output template from templates/viewpoint-output.md
+│   ├── Topic description + any user-provided context
+│   └── Output path: $PROJECT_DIR/logs/research/{topic-slug}/{NN}-{viewpoint-slug}.md
+├── Spawn all 5 agents in parallel via Task tool
+│   ├── subagent_type: general-purpose
+│   ├── model: sonnet
+│   └── All 5 in a single message (parallel)
+└── Token budget check after all 5 complete (checkpoint if >55%)
+```
+
+**CRITICAL**: Spawn all 5 agents in a single message with 5 Task tool calls. Do NOT spawn sequentially.
+
+### Stage 3: Synthesis
+
+```
+Stage 3: Synthesis
+├── Read ALL 5 agent output files (MANDATORY — do not skip any)
+├── If any output is missing or empty → re-spawn that agent once (max 1 retry)
+├── If retry fails → document gap in synthesis under "Incomplete Coverage"
+├── Load templates/synthesis-output.md
+├── Write synthesis to $PROJECT_DIR/artifacts/research/{topic-slug}/synthesis.md
+├── AskUserQuestion for user on open questions (iterative, 2-3 per round)
+├── Critical Evaluation Gate (see below)
+└── Token budget check (must be <65% after synthesis)
+```
+
+**Enforcement**: Do NOT begin writing synthesis until ALL available agent outputs have been read. The orchestrator must reference every agent's output at least once in the synthesis.
+
+#### Critical Evaluation Gate (Post-User Q&A)
+
+After each AskUserQuestion round, do NOT blindly incorporate user responses. Instead:
+
+**Step 1 — Classify each user response:**
+
+| Classification | Definition | Action |
+|---------------|------------|--------|
+| **Factual** | Known, verifiable information (e.g., "We use PostgreSQL") | Incorporate directly into synthesis |
+| **Opinion** | Preference or priority (e.g., "I'd prefer approach A") | Incorporate directly with attribution: "User preference: ..." |
+| **Speculative** | Unvalidated claim or proposed solution (e.g., "I think library X can do this", "What if we used approach Y?") | **Do NOT incorporate.** Trigger Step 2. |
+
+**Step 2 — For Speculative responses, present to user:**
+
+> "Your suggestion about [X] is unvalidated. I recommend a targeted follow-up research phase with 2 focused agents (Direct Investigation + Contrarian) to verify feasibility and surface risks before incorporating this into the synthesis.
+>
+> This will spawn 2 Sonnet agents and consume additional token budget.
+>
+> [Run follow-up research / Incorporate as-is with LOW confidence caveat]"
+
+**Step 3 — If follow-up research approved:**
+
+1. Spawn 2 Sonnet agents in parallel (single message, 2 Task tool calls):
+   - **Direct Investigation** — focused on validating the specific claim/solution
+   - **Contrarian** — focused on finding failure modes and alternatives for the specific claim/solution
+2. Use the same 4-part prompt template (GOAL/CONSTRAINTS/CONTEXT/OUTPUT)
+3. Include the REASONING DEPTH instructions from the viewpoint reference docs
+4. Output to: `$PROJECT_DIR/logs/research/{topic-slug}/followup-{NN}-direct-investigation.md` and `followup-{NN}-contrarian.md`
+5. Read both outputs, then update synthesis with validated findings
+6. Tag follow-up findings in synthesis with: `[Follow-up: validated]` or `[Follow-up: refuted]` or `[Follow-up: mixed — see details]`
+
+**Step 4 — If user declines follow-up:**
+
+Incorporate the user's suggestion into synthesis with an explicit caveat:
+> **[Unvalidated — user suggestion, not research-backed]**: {suggestion}
+
+**Repeat**: After updating synthesis, ask if user has additional questions or input. Apply the same classification gate to each round. There is no limit on follow-up rounds, but each round with Speculative input that triggers research consumes ~10-15% token budget — warn user if approaching 60%.
+
+### Stage 4: Diagnostics (REQUIRED)
+
+```
+Stage 4: Diagnostics
+├── Write diagnostic YAML to $PROJECT_DIR/logs/diagnostics/bulwark-research-{YYYYMMDD-HHMMSS}.yaml
+└── Verify completion checklist
+```
+
+---
+
+## Viewpoints (Sections)
+
+Each viewpoint is a distinct analytical lens. All 5 run in parallel — they do not see each other's output.
+
+### Viewpoint 1: Direct Investigation
+
+**Core Question**: What is this? How does it work? State of the art?
+
+**Focus Areas**:
+- Precise definition — what it is and what it is not
+- Mechanical operation (architecture, data flow, lifecycle)
+- Current state of the art — tooling, adoption, standards
+- Key terminology and taxonomy
+
+**Reference**: `references/viewpoint-direct-investigation.md`
+
+### Viewpoint 2: Practitioner Perspective
+
+**Core Question**: How do teams use this in production? What works?
+
+**Focus Areas**:
+- Real-world adoption patterns
+- Common implementation approaches and trade-offs
+- Practical gotchas documentation doesn't cover
+- Operational concerns (debugging, monitoring, maintenance)
+- Team skill requirements and learning curves
+
+**Reference**: `references/viewpoint-practitioner.md`
+
+### Viewpoint 3: Contrarian Angle
+
+**Core Question**: What failure modes do most people overlook?
+
+**Focus Areas**:
+- Failure modes advocates rarely mention
+- Scenarios where this is the wrong choice
+- Hidden costs (complexity, maintenance burden, cognitive load)
+- Alternatives that might be simpler
+- When NOT to use this
+
+**Reference**: `references/viewpoint-contrarian.md`
+
+### Viewpoint 4: First Principles
+
+**Core Question**: What core problem does this solve? Minimal viable version?
+
+**Focus Areas**:
+- Fundamental problem being addressed (stripped of buzzwords)
+- Why existing approaches are insufficient
+- Minimal set of capabilities for value
+- Essential vs. deferrable
+- Decomposition into independent sub-problems
+
+**Reference**: `references/viewpoint-first-principles.md`
+
+### Viewpoint 5: Prior Art / Historical
+
+**Core Question**: What similar patterns exist? Lessons from predecessors?
+
+**Focus Areas**:
+- Historical predecessors and analogous patterns
+- Evolution trajectories — what succeeded, what failed, why
+- Hype vs. foundational patterns
+- Lessons applicable to current topic
+
+**Reference**: `references/viewpoint-prior-art.md`
+
+---
+
+## Token Budget Management
+
+| Checkpoint | Threshold | Action |
+|------------|-----------|--------|
+| After constructing all prompts | >30% consumed | Warn user: "5 agents will consume significant context" |
+| After reading 3 of 5 outputs | Running tally | If approaching 55%, checkpoint with user |
+| After synthesis | Must be <65% | Leave room for session closing |
+| Synthesis complete at >65% | Immediate | Create handoff, do not start additional work |
+
+If token budget is insufficient to complete all 5 agents + synthesis, inform the user and suggest splitting (e.g., "3 agents this session, 2 + synthesis next session").
+
+---
+
+## Error Handling
+
+| Scenario | Action |
+|----------|--------|
+| Agent returns empty output | Re-spawn once. If still empty, document gap in synthesis. |
+| Agent returns truncated output | Accept as-is, note in diagnostics. |
+| Agent fails to spawn | Re-spawn once. If still fails, reduce to 4 agents, document. |
+| Token budget exceeded mid-session | Stop spawning, synthesize from available outputs, note incomplete. |
+| User-provided document unreadable | AskUserQuestion for alternative source. |
+
+---
+
+## Diagnostic Output (REQUIRED)
+
+**MANDATORY**: You MUST write diagnostic output after every invocation. This is Stage 4 and cannot be skipped.
+
+Write to: `$PROJECT_DIR/logs/diagnostics/bulwark-research-{YYYYMMDD-HHMMSS}.yaml`
+
+**Template**: Use `templates/diagnostic-output.yaml` for the schema. Fill in actual values from the session.
+
+---
+
+## Completion Checklist
+
+**IMPORTANT**: Before returning to the user, verify ALL items are complete:
+
+- [ ] Stage 1: Pre-flight complete (topic defined, directories created, skills loaded)
+- [ ] Stage 1: AskUserQuestion used if topic was ambiguous
+- [ ] Stage 2: All 5 viewpoint agents spawned in parallel
+- [ ] Stage 2: All agent outputs written to `$PROJECT_DIR/logs/research/{topic-slug}/`
+- [ ] Stage 3: ALL 5 outputs read before writing synthesis
+- [ ] Stage 3: Synthesis written using `templates/synthesis-output.md`
+- [ ] Stage 3: AskUserQuestion used for post-synthesis review
+- [ ] Stage 3: Critical Evaluation Gate applied to all user responses (classified as Factual/Opinion/Speculative)
+- [ ] Stage 3: Follow-up research spawned for Speculative responses (or user declined with caveat added)
+- [ ] Stage 3: Synthesis written to `$PROJECT_DIR/artifacts/research/{topic-slug}/synthesis.md`
+- [ ] Stage 4: Diagnostic YAML written to `$PROJECT_DIR/logs/diagnostics/`
+
+**Do NOT return to user until all checkboxes can be marked complete.**

package/skills/bulwark-research/references/viewpoint-contrarian.md

@@ -0,0 +1,63 @@
+# Viewpoint: Contrarian Angle
+
+## Core Question
+
+What failure modes and risks do most people overlook?
+
+## Focus Areas
+
+- Failure modes that advocates rarely mention
+- Scenarios where this approach is the wrong choice
+- Hidden costs (complexity, maintenance burden, cognitive load)
+- Alternatives that might be simpler or more appropriate
+- When NOT to use this
+
+## Prompt Template
+
+```
+GOAL: Research [{topic}] from the Contrarian Angle. Identify failure modes,
+hidden costs, and scenarios where this approach is the wrong choice. Challenge
+the prevailing narrative.
+
+CONSTRAINTS:
+- Focus exclusively on the Contrarian lens — other viewpoints are handled
+  by parallel agents
+- Be genuinely critical, not performatively contrarian — ground critiques in evidence
+- Flag confidence levels: HIGH (verified/multiple sources), MEDIUM (single
+  source/strong reasoning), LOW (inference/limited data)
+- Do not pad findings — "I couldn't find evidence for X" is a valid and valuable finding
+- Target 1000-1500 words
+
+REASONING DEPTH — Research-Evaluate-Deepen:
+You MUST follow this multi-pass process (do not skip to writing the final output):
+
+1. INITIAL RESEARCH: Conduct your first pass — search for criticisms, failure cases,
+   abandoned projects, and dissenting opinions. Gather raw contrarian evidence.
+2. EVALUATE: Review what you found. For each finding, explicitly state:
+   - The critique
+   - Supporting evidence (specific failures, measurable costs, real examples)
+   - Counterargument (how advocates respond to this critique)
+   - Your net assessment (is the critique valid despite the counterargument?)
+3. IDENTIFY GAPS: What are the 2-3 failure modes or risks that NEITHER advocates
+   NOR critics seem to discuss? What blind spots exist in the discourse?
+4. DEEPEN: Conduct a second targeted research pass focused on those blind spots.
+   Look for adjacent domains where similar patterns failed for non-obvious reasons.
+5. RECONCILE: Document what changed between your initial critiques and your deepened
+   analysis. Did any critiques turn out to be weaker than expected? Did new risks
+   emerge that are more important than the commonly cited ones?
+
+Only after completing all 5 steps, write your final output using the template below.
+
+CONTEXT:
+{topic_description}
+{user_provided_context}
+{scope_boundaries}
+
+OUTPUT:
+Write findings to: {output_path}
+Use the output template provided below for document structure.
+Use YAML header with: viewpoint, topic, confidence_summary, key_findings (3-5 bullets)
+Follow with detailed analysis organized by the focus areas above.
+
+{viewpoint_output_template}
+```

package/skills/bulwark-research/references/viewpoint-direct-investigation.md

@@ -0,0 +1,62 @@
+# Viewpoint: Direct Investigation
+
+## Core Question
+
+What is this? How does it work? What is the current state of the art?
+
+## Focus Areas
+
+- Define the concept precisely — what it is and what it is not
+- How it works mechanically (architecture, data flow, lifecycle)
+- Current state of the art — who uses it, what tooling exists
+- Official documentation, specifications, or standards
+- Key terminology and taxonomy
+
+## Prompt Template
+
+```
+GOAL: Research [{topic}] from the Direct Investigation perspective. Produce a
+comprehensive technical analysis covering definition, mechanics, state of the art,
+and key terminology.
+
+CONSTRAINTS:
+- Focus exclusively on the Direct Investigation lens — other viewpoints are handled
+  by parallel agents
+- Be evidence-based: cite sources, examples, or reasoning for each claim
+- Flag confidence levels: HIGH (verified/multiple sources), MEDIUM (single
+  source/strong reasoning), LOW (inference/limited data)
+- Do not pad findings — "I couldn't find evidence for X" is a valid and valuable finding
+- Target 1000-1500 words
+
+REASONING DEPTH — Research-Evaluate-Deepen:
+You MUST follow this multi-pass process (do not skip to writing the final output):
+
+1. INITIAL RESEARCH: Conduct your first pass — web searches, codebase exploration,
+   document reads. Gather raw findings.
+2. EVALUATE: Review what you found. For each finding, explicitly state:
+   - The claim
+   - Supporting evidence
+   - Counterevidence or caveats
+   - Your net assessment
+3. IDENTIFY GAPS: What are the 2-3 most important questions your initial research
+   did NOT answer? What uncertainties remain?
+4. DEEPEN: Conduct a second targeted research pass focused specifically on those
+   gaps. Search for counterexamples, edge cases, or missing context.
+5. RECONCILE: Document what changed between your initial findings and your deepened
+   findings. Did any initial conclusions shift? Flag these explicitly.
+
+Only after completing all 5 steps, write your final output using the template below.
+
+CONTEXT:
+{topic_description}
+{user_provided_context}
+{scope_boundaries}
+
+OUTPUT:
+Write findings to: {output_path}
+Use the output template provided below for document structure.
+Use YAML header with: viewpoint, topic, confidence_summary, key_findings (3-5 bullets)
+Follow with detailed analysis organized by the focus areas above.
+
+{viewpoint_output_template}
+```

package/skills/bulwark-research/references/viewpoint-first-principles.md

@@ -0,0 +1,65 @@
+# Viewpoint: First Principles
+
+## Core Question
+
+What core problem does this solve? What is the minimal viable version?
+
+## Focus Areas
+
+- The fundamental problem being addressed (stripped of buzzwords)
+- Why existing approaches are insufficient
+- The minimal set of capabilities needed to solve the core problem
+- What can be deferred vs. what is essential
+- Decomposition into independent sub-problems
+
+## Prompt Template
+
+```
+GOAL: Research [{topic}] from First Principles. Break it down to the fundamental
+problem it solves, identify the minimal viable version, and decompose into
+independent sub-problems.
+
+CONSTRAINTS:
+- Focus exclusively on the First Principles lens — other viewpoints are handled
+  by parallel agents
+- Strip away buzzwords and marketing — focus on the underlying problem
+- Flag confidence levels: HIGH (verified/multiple sources), MEDIUM (single
+  source/strong reasoning), LOW (inference/limited data)
+- Do not pad findings — "I couldn't find evidence for X" is a valid and valuable finding
+- Target 1000-1500 words
+
+REASONING DEPTH — Research-Evaluate-Deepen:
+You MUST follow this multi-pass process (do not skip to writing the final output):
+
+1. INITIAL RESEARCH: Conduct your first pass — strip the topic to its core problem.
+   What fundamental need does this address? Research the problem space, not just
+   the proposed solution.
+2. EVALUATE: Review your decomposition. For each sub-problem identified:
+   - The sub-problem statement
+   - Why existing approaches fail to solve it
+   - The minimal capability needed to address it
+   - Whether this sub-problem is truly essential or a nice-to-have
+3. IDENTIFY GAPS: What are the 2-3 assumptions in your decomposition that you
+   haven't validated? Are you sure the problem is what it appears to be?
+4. DEEPEN: Conduct a second targeted research pass. Look for cases where the
+   "obvious" problem was actually a symptom of a different root cause. Check
+   whether simpler framings of the problem exist.
+5. RECONCILE: Document what changed between your initial decomposition and your
+   deepened analysis. Did the problem turn out to be simpler or more complex than
+   initially framed?
+
+Only after completing all 5 steps, write your final output using the template below.
+
+CONTEXT:
+{topic_description}
+{user_provided_context}
+{scope_boundaries}
+
+OUTPUT:
+Write findings to: {output_path}
+Use the output template provided below for document structure.
+Use YAML header with: viewpoint, topic, confidence_summary, key_findings (3-5 bullets)
+Follow with detailed analysis organized by the focus areas above.
+
+{viewpoint_output_template}
+```

package/skills/bulwark-research/references/viewpoint-practitioner.md

@@ -0,0 +1,62 @@
+# Viewpoint: Practitioner Perspective
+
+## Core Question
+
+How do teams actually use this in production? What works and what doesn't?
+
+## Focus Areas
+
+- Real-world adoption patterns — who uses this and how
+- Common implementation approaches and their trade-offs
+- Practical gotchas that documentation doesn't cover
+- Operational concerns (debugging, monitoring, maintenance)
+- Team skill requirements and learning curves
+
+## Prompt Template
+
+```
+GOAL: Research [{topic}] from the Practitioner Perspective. Describe how teams
+actually use this in production — what works well, what's harder than expected,
+and what operational concerns arise.
+
+CONSTRAINTS:
+- Focus exclusively on the Practitioner lens — other viewpoints are handled
+  by parallel agents
+- Draw on real-world usage patterns, not theoretical capabilities
+- Flag confidence levels: HIGH (verified/multiple sources), MEDIUM (single
+  source/strong reasoning), LOW (inference/limited data)
+- Do not pad findings — "I couldn't find evidence for X" is a valid and valuable finding
+- Target 1000-1500 words
+
+REASONING DEPTH — Research-Evaluate-Deepen:
+You MUST follow this multi-pass process (do not skip to writing the final output):
+
+1. INITIAL RESEARCH: Conduct your first pass — web searches, community discussions,
+   blog posts, production case studies. Gather raw practitioner experiences.
+2. EVALUATE: Review what you found. For each finding, explicitly state:
+   - The claim (e.g., "Teams report X works well")
+   - Supporting evidence (specific examples, team sizes, contexts)
+   - Counterevidence or caveats (who reports it does NOT work?)
+   - Your net assessment
+3. IDENTIFY GAPS: What are the 2-3 most important practical questions your initial
+   research did NOT answer? What operational concerns remain unclear?
+4. DEEPEN: Conduct a second targeted research pass focused on those gaps. Look for
+   failure post-mortems, migration stories, or "lessons learned" content.
+5. RECONCILE: Document what changed between your initial findings and your deepened
+   findings. Did any "best practices" turn out to have significant caveats?
+
+Only after completing all 5 steps, write your final output using the template below.
+
+CONTEXT:
+{topic_description}
+{user_provided_context}
+{scope_boundaries}
+
+OUTPUT:
+Write findings to: {output_path}
+Use the output template provided below for document structure.
+Use YAML header with: viewpoint, topic, confidence_summary, key_findings (3-5 bullets)
+Follow with detailed analysis organized by the focus areas above.
+
+{viewpoint_output_template}
+```

package/skills/bulwark-research/references/viewpoint-prior-art.md

@@ -0,0 +1,66 @@
+# Viewpoint: Prior Art / Historical
+
+## Core Question
+
+What similar patterns have existed before? What can we learn from their trajectory?
+
+## Focus Areas
+
+- Historical predecessors and analogous patterns
+- How similar approaches evolved over time
+- What succeeded and why; what failed and why
+- Patterns that were hyped then abandoned vs. patterns that became foundational
+- Lessons applicable to the current topic
+
+## Prompt Template
+
+```
+GOAL: Research [{topic}] from the Prior Art / Historical perspective. Analyze
+historical predecessors, their trajectories, and lessons applicable to how we
+should approach this topic today.
+
+CONSTRAINTS:
+- Focus exclusively on the Prior Art lens — other viewpoints are handled
+  by parallel agents
+- Draw genuine historical parallels, not superficial analogies
+- Flag confidence levels: HIGH (verified/multiple sources), MEDIUM (single
+  source/strong reasoning), LOW (inference/limited data)
+- Do not pad findings — "I couldn't find evidence for X" is a valid and valuable finding
+- Target 1000-1500 words
+
+REASONING DEPTH — Research-Evaluate-Deepen:
+You MUST follow this multi-pass process (do not skip to writing the final output):
+
+1. INITIAL RESEARCH: Conduct your first pass — identify historical predecessors,
+   analogous patterns from other domains, and evolution trajectories. Cast a wide
+   net across computing history and adjacent fields.
+2. EVALUATE: Review each historical parallel. For each:
+   - The predecessor and why it's relevant
+   - How it succeeded or failed (with specific evidence)
+   - The lesson applicable to the current topic
+   - How strong the analogy is (direct parallel vs. loose similarity)
+3. IDENTIFY GAPS: What are the 2-3 historical patterns you suspect exist but
+   couldn't find? What eras or domains haven't you checked?
+4. DEEPEN: Conduct a second targeted research pass focused on those gaps. Look for
+   less obvious predecessors — patterns from other industries, abandoned research
+   directions, or solutions that were ahead of their time.
+5. RECONCILE: Document what changed between your initial historical survey and your
+   deepened analysis. Did any "new" ideas turn out to have well-documented
+   predecessors? Did any historical "failures" turn out to be timing issues rather
+   than fundamental flaws?
+
+Only after completing all 5 steps, write your final output using the template below.
+
+CONTEXT:
+{topic_description}
+{user_provided_context}
+{scope_boundaries}
+
+OUTPUT:
+Write findings to: {output_path}
+Use the output template provided below for document structure.
+Use YAML header with: viewpoint, topic, confidence_summary, key_findings (3-5 bullets)
+Follow with detailed analysis organized by the focus areas above.
+
+{viewpoint_output_template}
+```