探索 AgentMaps

Fetch current library documentation and examples directly into coding agents.

Expose browser automation primitives through MCP for web navigation and testing.

Expose selected local directories to agents for read and write file workflows.

Skill profile for reviewing pull requests with correctness, risk, and test coverage rubrics.

Summarize public repository structure without requiring private tokens or write access.

Parse MCP server metadata, tools, prompts, and resources into a normalized interface report.

Skill profile for auditing MCP and agent projects for tokens, scripts, network, and file access.

Normalize an MCP server repo and flag risky permissions, scripts, and auth requirements.

MCP profile for web search tools that return sources, snippets, and fetchable URLs.

Skill profile for extracting structured tables and facts from PDF and office documents.

MCP profile for read-only local file access in constrained project directories.

Skill profile for source-grounded research briefs with claims, citations, and gaps.

Template for browser-operating agents that complete tasks through web UI actions.

Agent template for checking UI implementation against screenshots, spacing, and copy fidelity.

Check whether code changes need docs, changelog, migration notes, or release risk labels.

Classify pull requests by blast radius, missing tests, security exposure, and migration risk.

Extract colors, typography, spacing, and components from design references into implementation notes.

Build migration plans with dependency maps, risk levels, and validation steps.

Review agent traces for tool choice, wasted steps, unsafe actions, and recovery quality.

Convert ad hoc prompts into structured skill profiles with inputs, outputs, and safety notes.

Draft task-specific evaluation rubrics for agent traces, final outputs, and environment outcomes.

Compare rendered UI screenshots against expected layout, copy, and responsive behavior.

Agent scaffold for sandboxed terminal tasks with command logs and end-state checks.

Run product flows as a simulated user and report blockers, UX friction, and evidence.

Turn scattered docs and notes into deduplicated knowledge base entries with source links.

Analyze spreadsheets, verify formulas, and summarize data quality issues.

Turn outlines and source material into structured presentation drafts.

Draft and classify support replies while respecting policy and escalation rules.

Search local notes, summarize clusters, and identify source-backed follow-ups.

Fill controlled browser forms, validate DOM state, and report completion evidence.

Compare multiple projects or sources and produce a grounded decision matrix.

Profile CSV-like inputs, clean obvious issues, and produce a validation report.

Check outputs or support actions against a policy document and escalate uncertain cases.

Extract decisions, owners, due dates, and follow-ups from meeting notes.

Workflow for monitoring sources and producing a weekly synthesized research digest.

Extract obligations, risky clauses, dates, and unresolved questions from legal documents.

Plan and execute documentation migrations with broken-link and heading consistency checks.

Plan dependency upgrades, run checks, and explain compatibility risks.

Convert a small issue into a branch plan, implementation checklist, patch, and review summary.