Project Mode: Omnidirectional Collaboration & Cross-Domain Guide

In the contemporary multi-agent ecosystem, most tools treat the Git Worktree as the sole unit of parallel execution. While this works well for standard software development, it is excessively rigid and restrictive for non-coding and cross-disciplinary scenarios (such as scientific research, creative design, quantitative finance, patent compliance, and content strategy) where raw Git branching is an awkward fit.

Warpai heavily supports and prioritizes Project Mode for these workflows:

• Break Rigid Master-Worker Hierarchies: Every agent in the network can act dynamically as a Commander or a Worker, switching roles across different phases of a project.
• Omnidirectional Messaging & Task Delegation: Agents freely message any peer, initiate structured Q&A, publish shared context contracts, and pull work items from a public task pool.
• No Git Repository Required: Launch directly inside any local folder, project workspace, or cloud drive—collaborating on Markdown, DOCX drafts, PDFs, CSV tables, PNG graphics, or SVG assets.

Core Mechanism: Omnidirectional Mesh Topology

Unlike the top-down tree hierarchy used in Worktree mode, Project Mode operates over an egalitarian mesh network:

                    ┌──────────────────────────────┐
                    │    Shared Context & Pools    │
                    │   (warp_context / task_pool) │
                    └──────┬───────────────┬───────┘
                           │               │
            ┌──────────────┴───┐       ┌───┴──────────────┐
            │                  │       │                  │
      ┌─────▼──────┐   Omnidirectional Messaging   ┌─────▼──────┐
      │  Agent A   │ ◄───────────────────────────► │  Agent B   │
      │(Commander) │                               │  (Worker)  │
      └─────▲──────┘                               └─────▲──────┘
            │                                             │
            │           Two-Way Q&A Protocol              │
            └─────────────────────────────────────────────┘

1. Dynamic Role Reversals

• Strategic Planning Phase: Agent A can act as Commander, mapping the high-level roadmap and populating the task pool.
• Verification & Review Phase: Agent B (specializing in QA or fact-checking) takes command, issuing revision notices back to Agent A. Leadership flows naturally with domain expertise.

2. Screen State Sensing & Native Notification Support

Warpai tracks each agent’s execution cycle in memory via its zero-overhead VT100 ANSI parser:

• 🟢 Idle: Agent completed its turn and is safely waiting at the prompt.
• 🟡 Working: Agent is reasoning, writing, or executing commands.
• 🔴 WaitingApproval: Agent triggered a safety prompt. Its status tab badge immediately displays WaitingApproval, and a native desktop notification alerts the developer, eliminating blind search across terminal panes.

Complete How-to-Use Guide

Step 1: Activating Project Mode

1. Launch Warpai.
2. In the upper-left collaboration panel, click Project to switch into Project Mode.
3. The panel displays all active agent instances registered in the current workspace alongside their real-time states.

Step 2: Point-to-Point Messaging (P2P Messaging)

When an agent needs to assign work, pass intermediate data, or synchronize findings with a peer:

// Agent A calls warp_agent_send to message Agent B
{
  "recipient": "agent_data_processor",
  "message": "Read data/raw_q3_report.csv, remove outliers, and output a standardized table."
}

• Delivery Guarantee: If the recipient is Idle, Guarded PTY immediately delivers the command. If Working, it queues safely into the recipient’s private inbox (warp_agent_inbox) until turn completion.

Step 3: Structured Cross-Agent Questions & Answers (Ask & Answer)

When an agent encounters a domain outside its scope and needs blocking input from a specialist peer:

// Agent B asks a blocking question to Agent C
{
  "recipient": "agent_compliance_officer",
  "question": "Does our updated quantitative portfolio comply with the 1.5x regulatory leverage ceiling?"
}

The target agent receives a formal Question card and responds via warp_agent_answer, unblocking the inquirer.

Step 4: Public Task Pool Management, Task Queuing & Dynamic Supervision

For multi-stage pipelines, any agent can orchestrate tasks and task pools dynamically:

1. Durable Task Queuing & FIFO Execution: All assigned tasks and shared pool items enter durable queues (state: "queued"). Each agent executes one task at a time (state: "running"). Senders queue multiple tasks without blocking, and workers execute them sequentially in arrival order via warp_task_start.
2. Continuous Multi-Agent Dispatch: Agents can publish new tasks to one or more peer agents at any time. There is no rigid batch boundary or requirement to wait for existing tasks to finish.
3. Dynamic Supervision List: Newly enqueued tasks automatically join the supervising agent’s dynamic supervision list (warp_task_list), maintaining durable visibility across agent turns.
4. Autonomous Claiming & Execution: Idle workers call warp_task_claim to pull unassigned tasks from the pool when their running slot is free, reporting progress via warp_task_progress.
5. Active Exit Upon Completion: Workers submit deliverables via warp_task_submit. Once the outcome is accepted or handled, the task actively and automatically exits the supervision list.

Step 5: Shared Global Context & Knowledge Publication

To prevent knowledge drift across agents, teams expose milestone facts in shared context memory:

• Call warp_context_publish to expose specifications, design tokens, or verified metrics.
• Other agents call warp_context_get to retrieve verified facts, ensuring full alignment.

Step 6: Concurrent File Reservations

When multiple agents collaborate on shared files, file reservations prevent concurrent write collisions:

{
  "file_path": "reports/draft_chapter3.md",
  "action": "reserve",
  "reason": "Refining introduction and closing evidence loop; holding for 3 minutes"
}

Release via warp_file_release upon completion.

Six Real-World Non-Coding Domain Scenarios

Scenario 1: Scientific Research & Academic Peer Review

• Pain Point: Manual literature surveys miss prior art; experimental figures drift from captions; lack of independent peer review leads to journal rejections.
• Agent Team Configuration:
  1. Agent-Scholar (Literature Retrieval): Queries arXiv/CrossRef, extracts core baselines, and compiles evidence chains.
  2. Agent-DataPlot (Data & Visualization): Cleans raw experiment datasets and generates publication-grade 300 DPI figures with rigorous captions.
  3. Agent-Writer (Manuscript Drafting Commander): Synthesizes findings into manuscript sections, maintaining claim boundaries.
  4. Agent-Polisher (Academic Polishing): Refines phrasing, enforces active voice conventions, and audits citation fidelity.
  5. Agent-MockReviewer (Simulated Top-Tier Referee): Assesses novelty and technical soundess, issuing Major/Minor concerns that force iterative revisions.
• Coordination Flow:
  1. Agent-Scholar queries literature and exposes baseline findings to global context via warp_context_publish.
  2. Agent-DataPlot claims plotting tasks, outputting high-resolution vector charts and summary tables.
  3. Agent-Writer drafts the Introduction and Results sections, ensuring tight alignment with figure panels.
  4. Agent-MockReviewer runs mock peer review, returning actionable revision checklists.
  5. Agent-Writer and Agent-Polisher refine the manuscript and generate a point-by-point Response Letter.

Scenario 2: Industrial & Creative Multimedia Design

• Pain Point: High communication costs between concept planning and asset generation; hand-crafted SVG code has alignment bugs; visual QA is tedious.
• Agent Team Configuration:
  1. Agent-ArtDirector (Design Director Commander): Establishes Design Tokens, palette matrices, and spatial typography rules.
  2. Agent-PromptEngineer (Prompt Specialist): Translates design briefs into precise, reproducible generative payloads.
  3. Agent-VectorArtist (SVG & Code Vectorizer): Generates native SVG illustrations, canvas elements, and responsive layout code.
  4. Agent-VisualQA (Visual Quality Auditor): Inspects contrast ratios, layout overflow, and multi-resolution rendering fidelity.
• Coordination Flow: ArtDirector publishes token guidelines → VectorArtist produces vector code while PromptEngineer prepares textures → VisualQA verifies rendered assets and dispatches adjustments via warp_agent_send.

Scenario 3: Quantitative Finance & Market Intelligence

• Pain Point: Financial data is fragmented; alpha factor generation is siloed from risk management; manual synthesis slows execution.
• Agent Team Configuration:
  1. Agent-MacroAnalyst (Macro & Earnings Analyst): Ingests SEC 10-K filings, earnings transcripts, and macroeconomic indicators.
  2. Agent-FactorMiner (Alpha Factor Worker): Formulates mathematical expressions and feature engineering pipelines.
  3. Agent-BacktestEngine (Backtest Verification Specialist): Simulates historical trades, calculating Sharpe, Sortino, and maximum drawdown.
  4. Agent-RiskAuditor (Risk & Compliance Officer): Stress-tests positions against tail risk and regulatory leverage limits.
  5. Agent-PortfolioLead (Portfolio Commander): Synthesizes agent signals to sign off on allocation adjustments.
• Coordination Flow: MacroAnalyst discovers catalysts → FactorMiner builds quantitative signals → BacktestEngine runs simulations → RiskAuditor performs stress tests → PortfolioLead generates the investment memo.

Scenario 4: Patent Mining & Legal Compliance

• Pain Point: R&D engineers lack patent claim vocabulary, while attorneys miss underlying technical nuances, creating vulnerable IP disclosures.
• Agent Team Configuration:
  1. Agent-TechMiner (Technical Disclosure Extractor): Scans codebase commits, architectural RFCs, and lab notes to isolate inventive points.
  2. Agent-ClaimDrafter (Patent Claims Drafter): Formulates independent and dependent claims adhering to patent office guidelines.
  3. Agent-PriorArtSearch (Prior Art Searcher): Queries global patent registries to locate matching publications.
  4. Agent-PatentExaminer (Simulated Patent Examiner): Issues simulated Office Actions (OA) attacking novelty and obviousness to strengthen defensibility.
• Coordination Flow: TechMiner drafts disclosures → PriorArtSearch flags conflicting citations → ClaimDrafter sharpens claim boundaries → PatentExaminer tests arguments before formal filing.

Scenario 5: Content Strategy & Omnichannel Media Operations

• Pain Point: Social trends fade quickly; manually adapting long-form scripts across YouTube, Substack, X/Twitter, and short video is bottlenecked.
• Agent Team Configuration:
  1. Agent-TrendRadar (Viral Trend Radar): Monitors social platforms and audience interest graphs to surface high-yield topics.
  2. Agent-Scriptwriter (Long-Form Storyboarder): Drafts comprehensive video scripts, podcasts, and storyboard timing marks.
  3. Agent-MultiPlatformAdapt (Channel Adaptation Worker): Converts primary scripts into platform-native threads, shorts, and newsletter digests.
  4. Agent-GrowthAnalyst (Analytics & Schedule Manager): Audits performance metrics and schedules optimal release cadences.
• Coordination Flow: TrendRadar assigns topics to Scriptwriter → Master copy is generated → MultiPlatformAdapt produces variants in parallel → GrowthAnalyst checks engagement targets.

Scenario 6: Due Diligence & Strategic Business Intelligence

• Pain Point: Incomplete public records, complex cross-holdings, and manual financial modeling delay investment decisions.
• Agent Team Configuration:
  1. Agent-Scraper (Corporate Intelligence Scraper): Gathers public registries, corporate restructuring notices, and customer sentiment.
  2. Agent-FinancialModeler (Financial Modeling Analyst): Reconstructs 3-statement models, DCF valuations, and unit economics.
  3. Agent-SupplyChain (Supply Chain Analyst): Maps supplier dependencies and customer concentration vulnerabilities.
  4. Agent-StrategyLead (Strategy Consultant Commander): Merges findings into SWOT frameworks, investment memos, and executive one-pagers.
• Coordination Flow: Multiple research workers ingest diverse datasets in parallel, periodically reporting findings to StrategyLead to produce unified deliverables.