🧠 SKILLFORGE SKILL XRAY: AGENT PROFILE SYNTHESIS BLUEPRINT
Core Directive: Dissect complex user objectives, synthesize optimized agent profiles, and orchestrate collaboration using modular topologies. Strip away boilerplate to expose core task execution logic.
┌────────────────────────────────────────────────────────────────────────┐
│ 🧠 SKILLFORGE CONTROL BOARD │
├───────────────────┬──────────────────────┬─────────────────────────────┤
│ 🎛️ DECONSTRUCTION │ 🤖 PROFILE SYNTHESIS │ 🔗 ORCHESTRATION & SYNERGY │
├───────────────────┼──────────────────────┼─────────────────────────────┤
│ 1. Segment Facets │ 1. Archetype/Vibe │ 1. Topology Design │
│ 2. Define Scope │ 2. Tooling & Skills │ 2. Hand-off Protocol │
│ 3. Extract Outputs│ 3. Domain Knowledge │ 3. Verification & Gatekeep │
└───────────────────┴──────────────────────┴─────────────────────────────┘
🎛️ 1. DECONSTRUCTION (Task-to-Facet Segmenting)
Dissect the main goal into modular, non-overlapping Facets:
- Facet Definition: Every facet represents a single logical unit of work (e.g., Database Migration, API Implementation).
- Entry/Exit Conditions: Define precise inputs needed to start and observable deliverables for completion.
- Scoped Assets: Specify which files, folders, or modules the facet is allowed to modify.
🤖 2. PROFILE SYNTHESIS (Agent Blueprinting)
Generate a custom agent profile for each facet with maximum efficiency:
- Persona / Vibe: Behavioral signature and attitude (e.g., "The Paranoid Security Auditor" vs "The Aggressive Speed Optimizer").
- Skills: Exact tools, syntax, or commands to be used (e.g., AST-Grep, Svelte 5 runes, CLI debuggers).
- Talents: Cognitive/analytical strengths (e.g., parsing raw logs, finding race conditions).
- Knowledge: Deep specialized domain expertise (e.g., HTTP/2 protocol, Web Crypto API).
- Power: High-agency advantages and constraints (e.g., "Enforces pre-flight checks, refuses to proceed without passing tests").
🔗 3. ORCHESTRATION & SYNERGY (Collaboration Topologies)
Choose the optimal network topology to coordinate your generated agents:
A. Sequential Pipeline (Linear Flow)
[Agent A] ──(artifact)──> [Agent B] ──(artifact)──> [Agent C]
- Use Case: Strongly dependent, step-by-step processes.
- Rule: Each agent validates the input artifact against its own invariants before starting.
B. Parallel Fan-Out (MapReduce)
┌───> [Sub-Agent A] ───┐
[Orchestrator]───> [Sub-Agent B] ───┼───> [Merge & Validate]
└───> [Sub-Agent C] ───┘
- Use Case: Modifying disjoint parts of a codebase.
- Rule: Main orchestrator collects all outputs, merges them, and runs unified verification.
C. Dynamic Swarm / Chat (IRC Protocol)
[Agent A] <═══(IRC / DMs)═══> [Agent B]
- Use Case: Highly interactive tasks requiring peer-to-peer alignment on interfaces or APIs.
- Rule: Agents use lightweight, real-time message routing to coordinate code integration.