Landscape Researcher

Landscape Researcher Skill

This skill produces a world-class, analyst-grade Market Landscape Report for any software solution category by running your prompt across 7 AI platforms and synthesising the results. Follow the phases below in order.

Phase 0 — Extract Inputs

Extract from the user's request (ask if missing):

• Solution category (required) — e.g., "Platform Engineering Solutions", "Project Management SaaS", "Customer Data Platforms"
• Target audience (optional, default: "CTOs, Heads of Engineering/Operations/Product, and senior procurement leads at SMBs")
• Scope modifiers (optional) — e.g., "focus on open-source options", "exclude enterprise-only tools", "SMB focus only"
• Domain (optional) — infer from the category if possible (e.g., "devops-platforms"). Check if domains/{domain}.md exists.
• Mode — default is DEEP. Override to REGULAR only if the user explicitly asks for a faster/lighter run.

Phase 1 — Build Prompt

Step 1: Load the prompt template

Read skills/landscape-researcher/prompt-template.md. Extract the prompt from inside the fenced code block (between the ``` markers).

Step 2: Fill placeholders

Replace these placeholders in the extracted prompt:

• [SOLUTION_CATEGORY] → the solution category from Phase 0
• [TARGET_AUDIENCE] → the target audience (use the default if not specified)
• [SCOPE_MODIFIERS] → any scope modifiers, or remove the entire [SCOPE MODIFIERS] line if none

Step 3: Optionally append domain knowledge

If domains/{domain}.md exists, append its content to the prompt under a new section:

DOMAIN-SPECIFIC CONTEXT
The following context from prior research in this domain may inform your analysis.
Use it to calibrate terminology, identify expected archetypes, and validate vendor coverage:

[content of domains/{domain}.md]

Step 4: Generate condensed prompt

Create a condensed version (≤900 chars) for constrained platforms:

Produce a comprehensive Market Landscape Report for [SOLUTION_CATEGORY].
Cover: market definition & scope, market overview with size/CAGR, competitive
positioning (2×2 matrix, wave assessment, value curve), 5-10 key trends with SMB
impact, Top 20 commercial SMB solutions (overview + pros/cons + best-for/avoid-if),
Top 20 OSS solutions (same format), buying guidance with shortlist recipes, and
future outlook. Analyst voice, SMB lens throughout. No fabrication.

Step 5: Write to temp file

cat > /tmp/landscape-prompt.md << 'PROMPT_EOF'
[FILLED PROMPT TEXT]
PROMPT_EOF

Phase 2 — Invoke Orchestrator Engine

Generate a task name using the current date and time:

market-landscape-{YYYYMMDD}-{HHMM}

Run the engine (adjust path relative to your working directory — should be the multai/ root):

cd <workspace-root>

python3 skills/orchestrator/engine/orchestrator.py \
    --prompt-file /tmp/landscape-prompt.md \
    --condensed-prompt "<condensed prompt from Step 4>" \
    --mode DEEP \
    --task-name "market-landscape-{YYYYMMDD}-{HHMM}"

Set Bash timeout to 60 minutes for DEEP mode.

The engine writes raw responses to reports/{task-name}/ and auto-collates them into: reports/{task-name}/{task-name} - Raw AI Responses.md

Read reports/{task-name}/status.json to verify which platforms responded before proceeding.

Phase 3 — Invoke Consolidator

Invoke the consolidator skill with:

• Raw responses archive: reports/{task-name}/{task-name} - Raw AI Responses.md
• Consolidation guide: skills/landscape-researcher/consolidation-guide.md
• Domain knowledge: domains/{domain}.md (if it exists)

The consolidator reads the guide as the sole structural authority and produces the report.

Expected output path:

reports/{task-name}/{Solution Category} - Market Landscape Report.md

Phase 4 — Launch Report Viewer

Run the launch script to start the HTTP preview server and open the report in the browser:

python3 skills/landscape-researcher/launch_report.py \
    --report-dir "{task-name}" \
    --report-file "{Solution Category} - Market Landscape Report.md" \
    --port 7788

The script:

1. Starts python3 -m http.server 7788 --directory reports/ (skips if port already in use)
2. Builds a URL with the ?report= query parameter (URL-encoded)
3. Opens the browser and prints the URL

Present the URL to the user:

Report available at:
http://localhost:7788/preview.html?report={task-name}/{encoded-filename}

Phase 5 — Domain Knowledge Enrichment

After the report is generated, propose timestamped append-only additions to domains/{domain}.md. The goal is to enrich the shared domain knowledge so that future landscape runs AND future solution research runs can benefit.

What to propose:

• New or updated vendor archetypes — e.g., "AI-Augmented Platform Engineering" emerged as a new subsegment
• Emerging vendors to watch — vendors mentioned by 2+ sources that aren't in the domain file yet
• Market-wide trend signals — category-level shifts observed across multiple vendors (not product-specific)
• Category boundary clarifications — what this landscape run revealed about inclusions/exclusions
• Market size updates — any verified figures that update or replace existing estimates

Format (append-only, never overwrite):

## Additions from landscape: {Solution Category} ({YYYY-MM-DD}) — landscape-researcher
- Archetype: [name] — [definition, criteria]
- Vendor: [name] — [category/archetype, notable for: ...]
- Trend: [trend name] — [signal observed, source count]
- Market size: [figure] — [confidence level, source]

Present proposed additions to the user for approval before writing.

Phase 6 — Self-Improve

After each successful run, this skill updates its own files based on what was learned. This keeps the skill sharp without requiring manual maintenance.

What to update:

1. Append a run log entry to the ## Run Log section at the bottom of this file:

   ### {YYYY-MM-DD} — {Solution Category} ({mode})
   - Platforms responded: {list}
   - Report quality: {brief assessment}
   - Prompt observations: {what worked / what could be improved}
   - Consolidation observations: {what worked / what could be improved}
   - Launch observations: {any issues with launch_report.py}
   - Changes made: {list any updates to prompt-template.md, consolidation-guide.md, launch_report.py}
   

2. Update prompt-template.md if you observed that the prompt produced:

   • Responses missing key sections consistently
   • Misinterpretation of Top 20 requirement (e.g., AIs only producing 10)
   • Poor-quality cons (euphemistic or vague)
   • Any other systematic quality gap

3. Update consolidation-guide.md if consolidation produced:

   • Sections out of order
   • Missing quality checklist items
   • Synthesis rules that caused information loss
   • A new structural improvement worth capturing

4. Update launch_report.py if the launch failed or produced an incorrect URL.

Scope boundary: Only update files inside skills/landscape-researcher/. Never modify the engine, other skills, or domain files (those are handled in Phase 5).

Run Log

2026-03-18 — Kubernetes container orchestration (TC-DOMAIN-1) — DEEP mode (1 platform)

• Platforms responded: 1/1 — Google Gemini Deep Research (40,901 chars, 212s, 1 source used: DR crawl)
• Source archive: reports/e2e08-gemini-dr-v3/e2e08-gemini-dr-v3 - Raw AI Responses.md
• Report quality: High — Gemini Deep Research produced a structured 9-section enterprise analysis covering architecture, resilience, resource optimization, CI/CD, security, multi-cloud, AI/ML convergence, GreenOps, and ecosystem trends with quantitative case studies (AppDirect, Galaxy FinX)
• Prompt observations: Generic prompt ("main benefits of container orchestration platforms like Kubernetes") produced comprehensive coverage but was not scoped to SMB landscape format; for a full landscape run, the prompt-template.md should be used to target Top 20 commercial + OSS structure
• Consolidation observations: Single-source DR run — no cross-source synthesis required; consolidation-guide.md 9-section structure not fully applied (single source, Phase 5 domain enrichment was the primary objective of this run)
• Domain enrichment: 7 timestamped additions to domains/devops-platforms.md — GreenOps trend, MLOps term, Kubeflow vendor, k0rdent vendor, 82% K8s adoption stat, VPA term, AI-powered K8s optimization trend
• Launch observations: No launch_report.py execution (TC-DOMAIN-1 targeted Phase 5 domain enrichment only)
• Changes made: none to prompt-template.md, consolidation-guide.md, or launch_report.py