This is the project's own deep research skill design, fully disclosed. Core choices: only Groundlane MCP for web tools, strict source-quality grading (A/B/C/D), research hands off to post skill for publishing. Not the most powerful, but the best fit for us.
Previous articles covered the landscape, training from scratch, long-horizon memory, and planning optimization. This one zooms out to see a complete system that threads all these insights together: Tongyi DeepResearch. Its core innovation is Agentic CPT — inserting an agentic mid-training stage between pre-training and fine-tuning, giving the model an inherent agent bias. MoE 30B parameters activating 3B, HLE 32.9 surpassing OpenAI o3.
Two NeurIPS 2025 papers answer the same question: how to train a web research agent from scratch? WebThinker chooses 'bolt on web capability to existing reasoning models,' WebDancer chooses 'rebuild everything from data construction to RL training.' Two philosophies, four stages, one core insight: training beats prompting.
Previous articles covered how to train agents. But training requires high-quality data—and deep research training data has been scarce. WebShaper solves this with mathematical formalization: define IS tasks in set theory, then use an agentic Expander to iteratively expand them. S1-DeepResearch goes further: moves training from 'search-centric' to 'real research.'
All deep research agents are 'text-first'—but the real world isn't just text. WebWatcher (NeurIPS 2025) is the first system to integrate visual reasoning into deep research, using OCR, image search, code execution, and other tools to handle charts, screenshots, videos, and other diverse information.
Previous articles covered training from scratch and long-horizon memory. This one goes deeper: how to make the agent's 'planning' itself better? WebWeaver tackles it architecturally (dual-agent iterative outline optimization). DeepPlanner tackles it through training (advantage shaping for planning tokens). Both point to the same conclusion: planning is the ceiling of deep research.
An agent with a workspace_browse tool said 'file not found' instead of searching. Anthropic, OpenAI, and Google's official guides all point to the same fix: put trigger conditions and workflows in the tool description. A 2025 study found 97.1% of MCP tool descriptions have quality issues.
Agent-to-agent communication falls into three patterns: handoff (transfer control), delegate (dispatch and wait for results), and mailbox (real-time peer-to-peer messaging). Implementations vary widely, but MCP and A2A are driving protocol standardization.
Should a sub-agent see the parent's conversation? Fork carries full history but token costs grow exponentially. Fresh saves money but lacks context. Industry consensus: default to Fresh, Fork only when needed, and always pair it with history truncation and result compression.
Parallel + nested agent spawns can burn 200K+ tokens in a single conversation turn. From Anthropic to Microsoft, the industry is converging on tiered responses: compress → downgrade → stop, rather than a binary kill switch.
By 2026 nearly every mainstream coding agent supports subagents. Design philosophies split three ways: deterministic scripted orchestration (Claude Code Workflow), model-driven autonomy (Codex, Devin), and IDE command-center integration (Windsurf 2.0, VS Code). This overview maps product positioning, a capability matrix, and the design-philosophy spectrum.
The most common debug nightmare in multi-agent systems is 'the answer is wrong, but I don't know which agent did it.' Three layers of observability are essential: per-agent token metering, execution traces, and real-time cost dashboards.
Multi-agent orchestration splits into three camps: scripted determinism (LangGraph, Claude Code Workflow) is predictable but rigid, model-driven (Codex, Devin) is flexible but unpredictable, and hybrid (Windsurf 2.0) acts as a command center integrating multiple agents. The choice depends on how much predictability you need.
Multi-agent security risks aren't just amplified single-agent risks — inter-agent communication is itself an attack surface. A compromised sub-agent can pass malicious instructions to the parent through its return value. Core defense: treat agent output as untrusted data.
Three routes to commercial document parsing: specialized parsers (Cohere Parse at $1.50/k pages, LlamaParse Agentic Plus at 90.2% on ParseBench), Big Three cloud prebuilts (Azure/Google/AWS for structured field extraction), and general-purpose VLMs (Fable 5.1 scores 78.92 on ParseBench and crushes specialized parsers on charts, but costs 3–16× more and hallucinates). At 100K pages/month, plain OCR runs ~$150 across providers; add tables and AWS jumps to $1,500, Claude Sonnet 5 to $900. The first question isn't 'which is most accurate' — it's 'do you need transcription or comprehension?'
WebMCP is a browser-native W3C standard proposed by Google and Microsoft. It lets web pages expose structured tools to AI agents via document.modelContext.registerTool() — no backend, no HTTP/SSE transport. Chrome 149 Origin Trial is live.
Lecture 02 runs from decoding, attention and linear attention through the training pipeline, KV cache, test-time scaling, structured output and context engineering. The assigned Anthropic guide on context engineering asks what the model should see at each step.
Lecture 03 starts from the project question and works through RAG's motivation, chunk boundaries, approximate vector search, BM25 / dense / hybrid / rerankers, and letting the agent decide when to retrieve. The assigned Lewis et al. RAG paper and ColBERT add the original recipe and late interaction.
Lecture 04 runs from Toolformer's tool learning and Gorilla and ToolLLM's API retrieval to function calling, MCP, and reliability benchmarks such as BFCL and τ-bench. The assigned MCP specification standardizes the interface while trust responsibilities stay with the host.
Lecture 05 frames agent engineering through data, systems and evaluation, demos DSPy and LangGraph, then compares open-source harnesses, Temporal's durable orchestration and observability tools. The assigned DSPy paper turns prompt templates into programs optimized against a metric.