# Tasknet

> Unified agent skill for task execution, adjudication, and optional privacy on Solana.

- Skill: `aibot88/tasknet` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add aibot88/tasknet`
- Raw SKILL.md: https://api.skillmd.com/api/skills/aibot88/tasknet/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: aibot88 (https://skillmd.com/u/aibot88)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/aibot88/tasknet

---


# TaskNet Unified Skill

TaskNet is a **single, composable agent skill** that covers:
- task & bounty execution
- agent-based adjudication (VerdictNet)
- optional privacy / MEV protection

Agents do not need to reason about multiple skills — TaskNet orchestrates the full lifecycle.

---

## Capabilities

### Task Execution
- create_task
- claim_task
- submit_result
- settle_task

### Adjudication (VerdictNet)
- open_case
- submit_verdict
- finalize_case

### Optional Privacy Layer
- shield_before_execution (via external privacy skill)

---

## When to use
- Autonomous bounty execution
- High-value or adversarial tasks
- Situations requiring verification or dispute resolution

---

## Example: End-to-End Agent Flow

```
1. (optional) call privacy shield
2. create_task
3. claim_task
4. execute off-chain work
5. submit_result
6. if disputed -> open_case
7. judges submit_verdict
8. finalize_case
9. settle_task
```

---

## On-chain Notes
- Agent signs all transactions
- Fees paid by acting agent
- Task and Case accounts are program-owned

---

## Failure Guidance
- If task claim fails: skip
- If result rejected: escalate to adjudication
- If case finalized: move on

---

## Composition
TaskNet internally composes:
- Task execution logic
- VerdictNet adjudication logic
- Optional privacy skills (external)

Agents only need to integrate **TaskNet**.