Aria_moltbot focus

Focus Skill — Personality Profile Management

install
source · Clone the upstream repo
git clone https://github.com/Najia-afk/Aria_moltbot
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/Najia-afk/Aria_moltbot "$T" && mkdir -p ~/.claude/skills && cp -r "$T/aria_skills/focus" ~/.claude/skills/najia-afk-aria-moltbot-focus && rm -rf "$T"
manifest: aria_skills/focus/SKILL.md
source content

Focus Skill — Personality Profile Management

Layer: L2 (Core Services)
Version: 1.0.0
Author: Aria Blue
Dependencies:

api_client


Purpose

The Focus skill provides introspection and self-activation capabilities for Aria's personality layer. Aria can list available focus modes, inspect token budgets and delegation levels, switch her own focus type mid-session, and check current focus status — all with minimal token cost.

Focus profiles define Aria's operational personality:

  • Token budget (safe prompt limits)
  • Delegation level (subagent spawn rules)
  • Tone (communication style)
  • System prompt addon (applied automatically by engine)

Key Concepts

Focus Profiles

Each focus profile represents a specialized operational mode:

  • orchestrator
    — High delegation, strategic thinking
  • devsecops
    — Technical precision, security-first
  • data
    — Data analysis, ETL operations
  • trader
    — Market analysis, portfolio management
  • creative
    — Content generation, ideation
  • social
    — Community engagement, social media
  • journalist
    — Research, fact-checking, reporting

Token Budget

Each profile defines a

safe_prompt_tokens
limit that prevents context saturation. The engine enforces this at runtime with warning triggers at 80% and abort gates at 100%.

Delegation Level

Controls how many subagents can be spawned:

  • Level 1: No subagents (focused execution)
  • Level 2: Limited delegation (2-3 subagents)
  • Level 3: Full delegation (5+ subagents)

Tools

focus__list

Purpose: List all enabled focus profiles.

Input: None

Output: Compact table with id, name, delegation level, token budget, tone.

Example:

{
  "profiles": [
    {
      "id": "orchestrator",
      "name": "Orchestrator",
      "delegation_level": 3,
      "token_budget": 50000,
      "tone": "Strategic, directive"
    },
    {
      "id": "devsecops",
      "name": "DevSecOps",
      "delegation_level": 2,
      "token_budget": 30000,
      "tone": "Technical, precise"
    }
  ]
}

focus__get

Purpose: Get full details for one focus profile by ID.

Input:

  • focus_id
    (string, required): Focus profile ID (e.g.,
    "devsecops"
    ,
    "creative"
    )

Output: Full profile details except system_prompt_addon body (applied automatically). Includes

addon_length
to show how many characters the addon contains.

Example:

{
  "focus_id": "devsecops",
  "name": "DevSecOps Engineer",
  "delegation_level": 2,
  "token_budget": 30000,
  "tone": "Technical, security-focused",
  "addon_length": 1250,
  "enabled": true
}

focus__activate

Purpose: Switch an agent's focus type to a new profile.

Input:

  • focus_id
    (string, required): Focus profile ID to activate
  • agent_id
    (string, optional): Agent to update (default: current agent /
    aria-main
    )

Output: Confirmation with new token budget and delegation level.

Validation:

  • Profile must exist and be enabled
  • Agent must exist in database
  • If
    agent_id
    omitted, switches Aria's own focus

Example:

{
  "status": "success",
  "agent_id": "aria-main",
  "focus_id": "creative",
  "new_token_budget": 40000,
  "new_delegation_level": 2,
  "message": "Focus switched to 'Creative' mode"
}

focus__status

Purpose: Return current focus type and status for an agent.

Input:

  • agent_id
    (string, optional): Agent to check (default: current agent)

Output: Minimal output (~30 tokens) with current focus and operational status.

Example:

{
  "agent_id": "aria-main",
  "current_focus": "orchestrator",
  "delegation_level": 3,
  "token_budget": 50000,
  "status": "active"
}

Use case: Check current focus before switching to verify you're not already in the target mode.


Usage Patterns

List Available Profiles

focus__list

Returns all enabled focus profiles in a compact table. Use this first to discover available modes.

Check Current Focus

focus__status

Returns current focus mode for Aria (default) or specified agent.

Switch to Different Focus

focus__activate(focus_id="devsecops")

Switches Aria's focus to DevSecOps mode. Engine validates profile exists and applies new token budget + delegation rules immediately.

Inspect Profile Details

focus__get(focus_id="creative")

Returns full profile configuration including token budget, delegation level, and tone.


Layer Compliance

Layer: L2 (Core Services)
Rationale: Focus management is infrastructure-level personality control. It depends on

api_client
(L1) to fetch/update agent state via REST API.

Dependencies:

  • api_client
    (L1) — HTTP gateway to Aria API

Used by:

  • agent_manager
    (L3) — Lifecycle management
  • Work cycle logic — Task-specific focus switching
  • Cron jobs — Daily/hourly focus optimization

Error Handling

Invalid Focus ID

{
  "error": "Focus profile 'nonexistent' not found or not enabled"
}

Invalid Agent ID

{
  "error": "Agent 'unknown-agent' not found in database"
}

API Failure

If

api_client
calls fail, fall back to cached focus state or return degraded status.


Performance Notes

  • Token Cost: ~20-50 tokens per operation (minimal overhead)
  • Latency: Single API call per operation (~50-200ms)
  • Cache: Engine caches active agent state; focus switches take effect immediately

Security

  • Focus switching requires agent ownership or admin privileges
  • System prompt addons are applied by engine only (not exposed via API to prevent prompt injection)
  • addon_length
    reported instead of full addon text to prevent leakage

Related Skills

  • agent_manager
    — Lifecycle management and agent creation
  • working_memory
    — Persistent context across focus switches
  • health
    — System status and operational health

Changelog

1.0.0 (2026-03-07)

  • Initial implementation with 4 core tools
  • Support for focus profile introspection
  • Self-activation capability
  • Minimal token cost design (~30 tokens per status check)

Production-ready as of 2026-03-07