LANGUAGES AND TOOLING / JetBrains
Wasm Inspector
Inspects WebAssembly binaries through JetBrains’ Hexana MCP server: imports, exports, memories, globals and tables read from the binary itself, ABI diffs between two builds, crash-frame triage, and host or supply-chain import audits.
“Use when inspecting or reasoning about a WebAssembly binary or component (.wasm) — imports/exports, functions, memories, globals, tables …”
01 / THE REASONING
Why this made the selection.
- Grounds every structural claim in tool output read from the binary itself, explicitly preferring the MCP server over objdump/strings/hexdump shell probing — the discipline that keeps wasm analysis from hallucinating.
- The operation table maps tasks to single focused calls: crash-frame triage passes every stack funcidx in one query, API diffs take a baseline binary, and following a focused operation with generic find/diff/trace probes is forbidden as wasteful.
- Covers the questions other wasm tooling skips: memory-contract limit changes between builds, mutable imported globals, data-segment payload policy checks, and entry-point effects (which exports write memory or call the host).
02 / THE REVIEW RECORD
What we actually inspected.
Source review has boundaries.
A clear record is more useful than a “safe” badge.
Material inspected
- plugins/hexana/skills/wasm-inspector/SKILL.md
- agents/openai.yaml
- plugins/hexana/LICENSE.txt
- .claude-plugin/plugin.json
Our findings
- The plugin manifest declares the skills directory and an mcpServers config, so the skill and the tool it drives ship as one unit.
- The path-field rule ("never write local, current, or baseline into a path field") and the one-call-per-audit rules are concrete anti-footgun guidance, not filler.
- The agents/openai.yaml enables implicit invocation with a short description, matching the skill’s own scope statement.
Not established by this review
- No .wasm binary was inspected; the MCP server was not installed or called.
- Operation outputs (api_diff, crash_frames, data_payload_audit) were read from the SKILL.md, not observed.
The review applies to the material and revision named here. A newer upstream release can change its behavior.
03 / PUT IT TO WORK
Add a skill to your workflow.
- Add the JetBrains Hexana plugin from its marketplace in your Claude Code-compatible client; the skill and its MCP server install together.
- Confirm the MCP server is connected before asking wasm questions; the skill cannot answer without it.
Before you start
- A client that supports Claude Code plugins and MCP servers
- A .wasm binary or component to inspect
04 / FOLLOW THE EVIDENCE
The source trail.
Our notes are separate from the original resource.
Check upstream before adopting a new version.
https://github.com/JetBrains/hexana/blob/2da10ef84c907ddb66f6f0eb980e70e8bf21ef46/plugins/hexana/skills/wasm-inspector/SKILL.md
Supports: summary, upstreamDescription, whySelected, bestFor, limitations, review, access
https://github.com/JetBrains/hexana/blob/2da10ef84c907ddb66f6f0eb980e70e8bf21ef46/plugins/hexana/skills/wasm-inspector/agents/openai.yaml
Supports: review, compatibility
https://github.com/JetBrains/hexana/blob/2da10ef84c907ddb66f6f0eb980e70e8bf21ef46/plugins/hexana/.claude-plugin/plugin.json
Supports: install, compatibility, limitations
https://github.com/JetBrains/hexana/blob/2da10ef84c907ddb66f6f0eb980e70e8bf21ef46/plugins/hexana/LICENSE.txt
Supports: license, access
https://github.com/vercel-labs/skills/blob/7407f3893ad4dceab546ac002c3ef806e4000c73/README.md
Supports: install