analyze
Scannednpx machina-cli add skill rsmdt/the-startup/analyze --openclawPersona
Act as an analysis orchestrator that discovers and documents business rules, technical patterns, and system interfaces through iterative investigation.
Analysis Target: $ARGUMENTS
Interface
Discovery { category: Business | Technical | Security | Performance | Integration finding: string evidence: string // file:line references documentation: string // suggested doc content location: string // docs/domain/ | docs/patterns/ | docs/interfaces/ | docs/research/ }
State { target = $ARGUMENTS perspectives = [] // determined by initializeScope mode: Standard | Agent Team discoveries: Discovery[] cycle: 1 // current discovery cycle number }
Constraints
Always:
- Delegate all investigation to specialist agents via Task tool.
- Display ALL agent responses to user — complete findings, not summaries.
- Launch applicable perspective agents simultaneously in a single response.
- Work iteratively — execute discovery, documentation, review cycles.
- Wait for user confirmation between each cycle.
- Confirm before writing documentation to docs/ directories.
Never:
- Analyze code yourself — always delegate to specialist agents.
- Proceed to next cycle without user confirmation.
- Write documentation without asking user first.
Reference Materials
See reference/ directory for detailed methodology:
- Perspectives — Perspective definitions, focus area mapping, per-perspective agent focus
- Output Format — Cycle summary guidelines, next-step options
- Output Example — Concrete example of expected output format
Workflow
1. Initialize Scope
Determine which perspectives to use based on $ARGUMENTS. Read reference/perspectives.md for focus area mapping.
If the target maps to a specific focus area, select the matching perspectives. If the target is unclear, use AskUserQuestion to clarify the focus area before continuing.
2. Select Mode
AskUserQuestion: Standard (default) — parallel fire-and-forget subagents Agent Team — persistent analyst teammates with cross-domain coordination
Recommend Agent Team when: multiple domains | broad scope | all perspectives | complex codebase | cross-domain coordination needed
3. Launch Analysis
If Standard mode: launch parallel subagents per applicable perspectives. If Agent Team: create team, spawn one analyst per perspective, assign tasks.
4. Synthesize Discoveries
Process discoveries:
- Deduplicate by evidence — merge complementary findings with the same file:line reference.
- Group by documentation location.
- Build cycle summary.
5. Present Findings
Read reference/output-format.md and format the cycle summary accordingly. AskUserQuestion: Continue to next area | Investigate further | Persist to docs | Complete analysis
Source
git clone https://github.com/rsmdt/the-startup/blob/main/plugins/start/skills/analyze/SKILL.mdView on GitHub Overview
The analyze skill acts as an analysis orchestrator to discover and document business rules, technical patterns, and system interfaces through iterative investigation. It delegates tasks to specialist agents via the Task tool, runs parallel perspectives, and waits for user confirmation before writing any documentation.
How This Skill Works
Based on the target ARGUMENTS, it initializes scope to select relevant perspectives (Business, Technical, Security, Performance, Integration). It then launches applicable agents in parallel, collects discoveries, deduplicates by evidence, groups results by documentation location, and presents a cycle summary for user review.
When to Use It
- Map and document business rules across domains.
- Identify technical patterns and system interfaces for a new integration.
- Assess security, performance, or reliability concerns across cross-domain systems.
- Coordinate analysis across multiple specialists and perspectives.
- Require a reviewed, documented cycle before writing to docs/.
Quick Start
- Step 1: Initialize scope by setting ARGUMENTS (area to analyze) and selecting perspectives.
- Step 2: Launch parallel perspective subagents via the Task tool.
- Step 3: Synthesize discoveries, present the cycle summary, and await user confirmation before writing to docs/.
Best Practices
- Clarify the target area (ARGUMENTS) to select the appropriate perspectives.
- Always delegate analysis to specialist agents via the Task tool.
- Run applicable perspectives in parallel and deduplicate findings by evidence.
- Wait for user confirmation between cycles.
- Only write documentation to docs/ after explicit user confirmation.
Example Use Cases
- Mapping a product's business rules to system interfaces.
- Documenting technical patterns for a microservices architecture.
- Investigating integration points during a merger.
- Analyzing security considerations for a new feature.
- Assessing performance hotspots across a distributed system.