nimble-agents（灵活的 Agent 人）

Nimble Agents

Structured web data extraction via Nimble agents. Always finish with executed results or runnable code.

User request: $ARGUMENTS

Prerequisites

Ensure the Nimble MCP server is connected:

Claude Code:

export NIMBLE_API_KEY="your_api_key"
claude mcp add --transport http nimble-mcp-server https://mcp.nimbleway.com/mcp \
  --header "Authorization: Bearer ${NIMBLE_API_KEY}"

VS Code (Copilot / Continue):

{
  "nimble-mcp-server": {
    "command": "npx",
    "args": ["-y", "mcp-remote@latest", "https://mcp.nimbleway.com/mcp",
             "--header", "Authorization:Bearer YOUR_API_KEY"]
  }
}

API key: online.nimbleway.com/signup → Account Settings → API Keys

Core principles

• Fastest path to data. Default route: discover agent → get schema → run → display results. Planning and generation are escalation paths.
• Always search existing agents first. Call nimble_agents_list before considering generate. Hard rule.
• Update over generate — always. When a close-match agent exists (same domain/type, even if missing fields or different scope), update it rather than generating from scratch. Updating preserves proven extraction logic and is faster, cheaper, and more reliable. Only generate a new agent when nimble_agents_list returns 0 results for the target domain. Never offer "Create new agent" as the recommended option when a close match exists.
• AskUserQuestion at every decision point — no exceptions. Always present the standard AskUserQuestion prompts shown in each step. Never skip them, never auto-advance without asking. Never present choices as plain numbered lists. Constraints: 2–4 options, header max 12 chars, label 1–5 words. Recommended option goes first with "(Recommended)".
• Schema before run — always. Call nimble_agents_get before nimble_agents_run. Present input parameters and output fields in markdown tables. This applies when switching agents too.
• Script generation (Step 2B) is ONLY for large-scale, high-volume tasks. Never generate code for normal interactive requests. Script mode requires ALL of: scale >50 items AND the user explicitly asks for code/script/CSV/batch output. Multi-source requests, dataset requests, and comparison requests do NOT automatically trigger script mode — run them interactively first. The default path is always: discover → run → display results.
• Verify response shape before script generation. Check skills and entity_type from nimble_agents_get to determine REST API response nesting. See references/agent-api-reference.md > "Response shape inference" and references/sdk-patterns.md > "Response structure verification".
• google_search is not a general search tool. It is a SERP analysis agent for rank tracking and SEO analysis. For finding information, use nimble_web_search. See references/error-recovery.md.
• All web search MUST use nimble_web_search (MCP). Never use WebSearch, WebFetch, or curl. See Guardrails.
• Mutation tools (generate, update, status, publish) are BANNED from the foreground. Always delegate to a Task agent. See Delegation model.
• Task agents MUST use run_in_background=False. Background mode breaks MCP access. See Delegation model.
• Foreground MCP calls limited to 3 tools: nimble_agents_list, nimble_agents_get, nimble_agents_run.
• Never use nimble_find_search_agent, nimble_run_search_agent, nimble_url_extract, or any WSA template tools.

Delegation model

The foreground conversation orchestrates and presents results. Task agents handle all MCP-heavy work.

Foreground — ONLY these 3 MCP tools allowed (direct calls):

Task agents — EVERYTHING else (mandatory, no exceptions):

nimble_agents_generate, nimble_agents_update_from_agent, nimble_agents_update_session, nimble_agents_status, and nimble_agents_publish are BANNED from the foreground. Always use Task(subagent_type="general-purpose", run_in_background=False) for these.

Why run_in_background=False: Background Task agents (run_in_background=True) cannot access MCP tools — they silently fall back to bash/curl and fail. Using run_in_background=False ensures MCP tool access. The Task agent still runs in its own context window (no foreground pollution); the only cost is the foreground waits for completion. When this platform limitation is resolved, switch back to run_in_background=True.

For multi-source workflows, launch Task agents sequentially (one per source/phase). Gather reports, then present the combined plan.

MCP tool registry

All Task agent prompts MUST include the tool registry block so the subagent knows the exact MCP tool names. Copy this block into every Task prompt:

**MCP tool registry (call these as MCP tool invocations, NOT bash/curl):**
| Short name | Full MCP tool name |
|------------|--------------------|
| nimble_agents_list | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_list |
| nimble_agents_get | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_get |
| nimble_agents_generate | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_generate |
| nimble_agents_update_from_agent | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_update_from_agent |
| nimble_agents_update_session | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_update_session |
| nimble_agents_status | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_status |
| nimble_agents_publish | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_publish |
| nimble_agents_run | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_run |
| nimble_web_search | mcp__plugin_nimble_nimble-mcp-server__nimble_web_search |

**CRITICAL: Use MCP tool calls only. NEVER use bash, curl, wget, WebSearch, or WebFetch to call APIs or search the web. NEVER construct MCP endpoint URLs manually.**

Response shapes

Always check typeof/isinstance before iterating REST responses.

Step 1: Route

From $ARGUMENTS, detect 3 things:

1. Clarity — clear (default) or needs-planning

Only needs-planning when ALL of these are absent: a target URL/site/domain, clear data to extract, a single well-scoped task. Most requests are clear.

2. Agent match — call nimble_agents_list ONCE per source/domain with the most specific short keyword (1–2 words, e.g., domain name or product type). This is ALWAYS the first action. For multi-source requests (e.g., "compare Amazon and Walmart prices"), call once per source. Do not retry the same source with different queries — if 0 results for a source, route it to Discovery (Step 1D).

Update is always preferred over generate. A close-match agent on the same domain already has working URL patterns, pagination logic, and parsing rules. Updating it to add/change fields is a minor refinement. Generating from scratch rebuilds everything and risks lower quality. Only generate when no agent exists for the target domain at all.

3. Execution mode — interactive (default) or script

Interactive is ALWAYS the default. Route to script generation (Step 2B) ONLY when BOTH conditions are met: (a) scale is explicitly >50 items or the user provides a batch input file, AND (b) the user explicitly asks for code, a script, a CSV export, or batch processing. Words like "dataset", "compare", "multi-source", or "2 sources" do NOT trigger script mode — run these interactively. Script generation writes code that calls an existing agent — it does not create new agents. If no agent exists yet, resolve that first (Step 3) before generating a script.

Step 1P: Plan mode (rare — only when needs-planning)

1. Clarify — AskUserQuestion to resolve critical unknowns (max 2 questions). Focus on: what site(s), what data fields, what output format.
2. Explore — nimble_agents_list for each target (foreground, 1 call each). For unfamiliar domains, launch Discovery Task agents (Step 1D).
3. Present plan — gap analysis table:

1. Execute — Step 2 for existing agents, Step 3 for generations (as Task agents).

Step 1D: Discovery (Task agent — for unfamiliar domains)

Launch when nimble_agents_list returns 0 matches and the target domain needs exploration. Runs as Task(subagent_type="general-purpose", run_in_background=False). The foreground tells the user: *"Exploring {domain} to understand available data..."*

Task prompt template:

Explore {domain} for {user_intent}.

**MCP tool registry (call these as MCP tool invocations, NOT bash/curl):**
| Short name | Full MCP tool name |
|------------|--------------------|
| nimble_web_search | mcp__plugin_nimble_nimble-mcp-server__nimble_web_search |

**CRITICAL: Use MCP tool calls only. NEVER use bash, curl, wget, WebSearch, or WebFetch to call APIs or search the web. NEVER construct MCP endpoint URLs manually.**

Use `nimble_web_search` (MCP) with deep_search=true to discover the site and available data:
1. Search "{domain} {keywords}" with deep_search=true, max_results=5 — this fetches and extracts full page content from each result, giving you product listings, detail pages, and field structures in one call.

**Return a structured report:**
- DOMAIN: {domain}
- ESTIMATED_ITEMS: count matching query
- LISTING_URL_PATTERN: e.g., /category/filter?color=green
- DETAIL_URL_PATTERN: e.g., /p/{slug}-{SKU}.html
- AVAILABLE_FIELDS: list of extractable fields (name, price, description, materials, etc.)
- MISSING_FIELDS: fields the user wants but the site doesn't have (e.g., ratings, reviews)
- RECOMMENDED_APPROACH: generate custom agent / use existing agent from {alternative} / combine sources
- SAMPLE_URLS: 2–3 example URLs for agent generation
- LIMITATIONS: login walls, pagination limits, JS rendering, etc.

Do NOT use AskUserQuestion. Do NOT use nimble_find_search_agent, nimble_run_search_agent, or nimble_url_extract.
Do NOT use WebSearch, WebFetch, bash curl, or any non-MCP search/fetch method.

On receiving the report, the foreground conversation:

1. Presents key findings to the user.
2. If data gaps exist (e.g., missing ratings), asks the user via AskUserQuestion how to proceed.
3. Routes to Step 3 (generate) with the discovery context, or Step 2 if existing agents cover the need.

For multi-source workflows, launch one Discovery agent per unfamiliar domain in parallel. Gather all reports before presenting the combined plan.

Step 2: Run existing agent

Two sub-paths based on execution mode.

2A: Interactive (small scale, display output)

2A-1. Call nimble_agents_get. Present schema in markdown tables:

• Input parameters: name, required, type, description, example
• Output fields: key fields from skills dict

See references/agent-api-reference.md > "Input Parameter Mapping" for the full input_properties format and mapping rules.

2A-2. Always confirm before running via AskUserQuestion:

question: "Run {agent_name} with these parameters?"
header: "Confirm"
options:
  - label: "Run agent (Recommended)"
    description: "Execute {agent_name} with {summary of inferred parameters}"
  - label: "Change parameters"
    description: "Adjust input parameters before running"
  - label: "Create new agent"
    description: "Create a custom agent instead (Step 3)"

2A-3. Call nimble_agents_run. Present results as markdown table. Always ask what to do next:

question: "What next?"
header: "Next step"
options:
  - label: "Done"
    description: "Finish with these results"
  - label: "Run again"
    description: "Re-run with different parameters"

Do NOT offer script generation as a next step unless the user explicitly mentions needing large-scale extraction (>50 items) or batch processing. Script generation is not a natural follow-up to interactive runs.

Bulk (2–5 URLs): Run per URL, aggregate results, handle individual failures without aborting. See references/batch-patterns.md > "Interactive batch extraction".

2B: Script generation (ONLY for large-scale, high-volume tasks)

This step is ONLY reached when the user explicitly needs to process >50 items at scale or requests batch code/script generation. Normal requests — even multi-source or "dataset" requests — are handled interactively via Step 2A. Writes a runnable script that calls an existing Nimble agent at scale via the SDK/REST API. This does NOT create new agents — the agent must already exist. Runs as a Task agent. The foreground infers language, launches the agent, and presents the generated script for confirmation.

2B-1. Infer language from project context (foreground, before launching):

2B-2. Launch script generation Task agent: Task(subagent_type="general-purpose", run_in_background=False).

Task prompt template:

Write a {language} script that calls existing Nimble agent(s) at scale via SDK/REST API.

**MCP tool registry (call these as MCP tool invocations, NOT bash/curl):**
| Short name | Full MCP tool name |
|------------|--------------------|
| nimble_agents_get | mcp__plugin_nimble_nimble-mcp-server__nimble_agents_get |

**CRITICAL: Use MCP tool calls only. NEVER use bash, curl, wget, WebSearch, or WebFetch to call APIs or search the web. NEVER construct MCP endpoint URLs manually.**

**Existing agents to call:** {agent_names}
**User intent:** {user_prompt}
**Output format:** {csv/json/etc}
**Scale:** {number of items/queries}

This is SCRIPT GENERATION — writing code that calls existing agents. Do NOT create new agents
(no nimble_agents_generate/update/publish). The agents listed above already exist.

Steps:
1. Call `nimble_agents_get` for each agent to inspect input_properties and skills.
2. Read the reference files:
   - `references/sdk-patterns.md` (Python) or `references/rest-api-patterns.md` (other languages)
   - `references/batch-patterns.md` (for multi-store normalization)
3. Write a complete, ready-to-run script with:
   - Smoke test first — validate a single query before full batch. Abort on failure.
   - Progress reporting — compact single-line status after each poll cycle.
   - Pagination handling for large result sets.
   - Multi-store field normalization (if applicable).
   - Output to {format}.
   - Incremental file writes for large pipelines (50+ jobs).

Return the complete script and a brief summary of:
- Agent schemas used (input params, key output fields)
- Normalization mappings (if multi-store)
- Total estimated API calls

Do NOT use AskUserQuestion. Do NOT use nimble_find_search_agent or nimble_run_search_agent.
Do NOT call nimble_agents_generate, nimble_agents_update_from_agent, nimble_agents_update_session, or nimble_agents_publish.
Do NOT use WebSearch, WebFetch, bash curl, or any non-MCP search/fetch method.

2B-3. Present the generated script and confirm execution via AskUserQuestion (foreground):

question: "Run this script?"
header: "Confirm"
options:
  - label: "Run script (Recommended)"
    description: "Execute the generated script"
  - label: "Edit first"
    description: "Review and modify the script before running"

No agent validation step here. The 50-input validation flow (Step 3) is only for agent creation/update. Script generation uses an existing, already-validated agent — just write the script and run it.

Step 3: Update existing agent or create new (on the Nimble platform)

Updates an existing agent (preferred) or creates a new one on Nimble's platform. Default to update when a close-match agent was found in Step 1 — pass the existing agent name to the Task agent so it uses nimble_agents_update_from_agent instead of nimble_agents_generate. Only create a new agent when Step 1 returned 0 matches. This is NOT code/script generation — it creates/modifies an extraction definition callable via Step 2A or 2B. ALWAYS runs as a Task agent (run_in_background=False).

3-1. Create a stable session_id (UUID v4).

3-2. Ask the user ONCE (foreground only — agent creation/update ONLY, never for script generation):

question: "Run refinement-validation before publishing?"
header: "Validate"
options:
  - label: "Yes, validate (Recommended)"
    description: "Discovery → generate → validate 50 inputs (80% pass) → publish. Auto-retries on failure."
  - label: "No, generate only"
    description: "Generate → publish immediately without validation testing"

3-3. Launch Task agent

Set refine_validate to the user's choice. Launch Task(subagent_type="general-purpose", run_in_background=False, max_turns=50) using the prompt template from references/generate-update-and-publish.md (includes MCP tool registry, lifecycle phases, and all rules). Tell the user: "Agent generation started. I'll report results when complete."

The Task agent executes a closed-loop lifecycle: Discovery → Create/Update → Poll → Validate → Publish → Report. On failure, it auto-triggers an update loop (max 2 cycles, 15-minute wall-clock timeout). See the reference file for complete details.

3-4. Present report

When the Task agent completes, present the report. On success, route to Step 2A or 2B. On failure after max cycles, offer:

question: "Agent validation did not reach 80% pass rate. How to proceed?"
header: "Next step"
options:
  - label: "Publish anyway"
    description: "Publish with current pass rate ({rate}%)"
  - label: "Update agent"
    description: "Provide specific instructions to refine the agent"

Step 4: Final response

End with a concise summary table:

Include the extraction results (or top N if large).

Documentation & troubleshooting

For large-scale codegen tasks (Step 2B) only — consult these when writing scripts that call Nimble APIs at scale. Do NOT load these for interactive runs (Step 2A) or agent creation (Step 3):

1. references/sdk-patterns.md — correct SDK patterns and common mistakes.
2. https://docs.nimbleway.com/llms-full.txt — full prose docs.
3. https://docs.nimbleway.com/openapi.json — API contract.
4. Context7 (if available) — query nimbleway.

For interactive runs and agent creation, the MCP tool schemas from nimble_agents_get provide all the information needed.

Error recovery

Consult references/error-recovery.md for handling patterns including persistent data source failures, ambiguous agent matches, and the full fallback hierarchy.

Additional references

Load reference files only during large-scale script generation (Step 2B) or agent creation (Step 3). Do NOT load these for interactive runs (Step 2A) — MCP tool schemas are sufficient.

For script generation (Step 2B) only:

• references/sdk-patterns.md — Running agents, async endpoint, batch pipelines, incremental file writes.
• references/rest-api-patterns.md — REST API patterns for TypeScript, Node, curl, and other non-Python languages.
• references/batch-patterns.md — Multi-store comparison, normalization, interactive batch, codegen walkthrough.

For agent creation/update (Step 3) only:

• references/generate-update-and-publish.md — Full agent creation/update lifecycle: discovery, creation, polling, SDK validation (50 inputs, 80% threshold), publish, reporting, update loop.

General (any step, load as needed):

• references/agent-api-reference.md — MCP tools reference plus input parameter mapping.
• references/error-recovery.md — Error handling and recovery patterns.

Guardrails

• NEVER call nimble_agents_generate, nimble_agents_update_from_agent, nimble_agents_update_session, nimble_agents_status, or nimble_agents_publish in the foreground conversation. These MUST run inside a Task agent. No exceptions — not even "just one quick call". Polling in the foreground floods context and wastes tokens.
• Task agents MUST use run_in_background=False. Background Task agents cannot access MCP tools (known platform limitation). When this is resolved upstream, switch back to run_in_background=True.
• All web search MUST use nimble_web_search (MCP tool). NEVER use built-in WebSearch, WebFetch, curl, wget, or any other search/fetch method — in the foreground or in Task agents. The full tool name is mcp__plugin_nimble_nimble-mcp-server__nimble_web_search.
• NEVER use bash/curl to call MCP endpoints. Task agents must call MCP tools by their tool names, not by constructing HTTP requests to MCP server URLs. If an MCP tool is not available, report the error — do not attempt to work around it via bash.
• Every Task agent prompt MUST include the MCP tool registry block from Delegation model > MCP tool registry. This tells the subagent the exact tool names to use. Without it, the subagent may fail to find MCP tools and fall back to bash.
• Foreground MCP tools are limited to: nimble_agents_list, nimble_agents_get, nimble_agents_run. Nothing else.
• Never use nimble_find_search_agent, nimble_run_search_agent, or any WSA template tools.
• Update first, generate only as last resort. When a close-match agent exists, always use nimble_agents_update_from_agent to refine it — never generate a new agent for the same domain. Use nimble_agents_generate ONLY when nimble_agents_list returned 0 results. For the update state machine: call nimble_agents_update_from_agent once to create/resolve the update session, then use nimble_agents_update_session for every follow-up with the same session_id. Never create a second session for the same refinement thread.
• Hard 429 rule. If nimble_agents_generate or nimble_agents_update_from_agent returns 429/quota errors, stop and report quota exhaustion. Do not call any update tool as fallback and do not create new sessions.
• Published agents are automatically forked when updated. UBCT-based agents cannot be updated — generate a new one instead.
• Never load external docs (llms-full.txt, openapi.json), SDK references, or batch pattern files for interactive runs (Step 2A). These are exclusively for large-scale script generation (Step 2B). For interactive runs, the MCP tool schemas from nimble_agents_get are sufficient. Loading unnecessary docs wastes context and slows execution.
• Present tool call results in markdown tables. Never show raw JSON.