PlectoProxy

Plecto's core architecture — the two halves (native-Rust fast path / WASM extension plane), the WIT type contract between them, the deny-by-default capability boundary, the filter chain, typed decision/short-circuit, init vs per-request hooks, instance lifecycle, and host-held state.

Audits code for security vulnerabilities using OWASP Top 10:2025, OWASP ASVS 5.0, and the OWASP Secure Code Review Cheat Sheet, with Plecto-specific dimensions: WASM sandbox / capability boundary (deny-by-default, epoch metering, CVE-2022-39393 pooling leakage, OCI signature verification, untrusted multi-tenant filters) and L7 proxy / gateway risks (SSRF on upstream construction, TLS termination, request smuggling/splitting, header injection, rate-limit/WAF bypass). Produces a structured findings report (severity + OWASP/CWE/ASVS mapping + evidence + remediation).

Best practices for embedding wasmtime as Plecto's host: InstancePre + pooling allocator for fast per-worker instance reuse, epoch interruption + memory limits for metering, Linker-based deny-by-default host functions, Store-per-request lifecycle, async host calls, pooling zeroization (CVE-2022-39393), and OCI-artifact load + cosign signature verification.

Design and evolve Plecto's `plecto:filter` WIT world and host-API surface — the type contract between the native fast path and untrusted WASM filters. Covers worlds/interfaces, resources, the decision variant, minimal-capability host-API slicing, header bytes (list<u8>), the filter / filter-body world split (header-only vs buffered body), versioning/compat with frozen v0.1/v0.2 load-time adapters, and the projected stream<u8> + wasm32-wasip2 migration.

Rust ベストプラクティス。Plecto の fast path（native Rust）と wasmtime ホスト埋め込みの コード品質を保つ規約とパターン集（Edition 2024）。

TypeScript / JavaScript ベストプラクティス。型安全性とコード品質を保つ規約。 Plecto では (1) Node ツーリング・統合テスト・wasm-bindgen の node_pkg 周り、 (2) jco / componentize-js で書く JS/TS 製 WASM フィルタ、の二用途に効く。

Runs an adversarial full review of a merge commit, PR, or branch range (AI-authored or not): Builder/Critic session separation, Spec+Diff-only critic input, parallel critic lanes (correctness, security, tests/CI), evidence-gated findings, and a structured merge verdict.

Find deepening opportunities in the codebase, informed by the domain language in CONTEXT.md, the project's design tenets, and the decisions in docs/ADR/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make the codebase more testable and AI-navigable — or says 「アーキテクチャ改善」「リファクタの種を探して」「設計を深く」.

Web research skill that prioritizes official documentation while covering broad literature. Searches, fetches, and synthesizes findings into a structured, source-cited, actionable report. This is the lightweight, in-repo researcher; for a deep, multi-agent, adversarially-verified report prefer the built-in `deep-research` skill.

Generate multiple radically different interface designs for a module using parallel sub-agents, then compare. Use when designing an API, a Rust trait, a host-API surface, or the `plecto:filter` WIT world; when exploring interface options or comparing module shapes; or when the user says "design it twice" / 「インターフェース設計」「WIT を設計」「二回設計して」.

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, validate a Fork-style decision (判断/根拠/再検討条件) before committing, or mentions "grill me" / 「詰めて」「壁打ち」.

Grilling session that challenges your plan against Plecto's documented design decisions and domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as decisions crystallise. Use when the user wants to stress-test a plan against the project's language and documented decisions, or says 「設計を詰めて」「用語を固めて」「ドキュメントと突き合わせて」.

Writes an Architecture Decision Record for the Plecto project in Japanese after a decision or completed implementation. Authoring only — no deploy / no container rebuild (Plecto has no deploy infra yet). Trigger when the user says "ADR書いて" / "ADRにまとめて" / "ADRに記録して" / "docs/ADR" 関連のまとめ依頼, or after a change that clearly warrants a decision record (a Fork-style 判断/根拠/再検討条件).

Writes a blameless incident postmortem for the Plecto project in Japanese, grounded in the real timeline, commits, and logs (never speculation). Use when the user says "ポストモーテム書いて" / "障害報告書いて" / "事後検証" / "incident report" / "RCA" / "根本原因", or after an incident, outage, sandbox/capability scare, or a committed bug fix worth a durable record. Authoring only — no deploy.

Interactive QA session where the user reports bugs or issues conversationally, and the agent files GitHub issues. Explores the codebase in the background for context and domain language. Use when the user wants to report bugs, do QA, file issues conversationally, or mentions "QA session" / 「QAして」「バグ報告」「issue 切って」.

Test-Driven Development workflow for Plecto (Rust-first, plus JS/TS filters). Use when implementing features, fixing bugs, or refactoring, or when the user says "TDDで". Enforces outside-in order E2E (request through the filter chain) → WIT-conformance (when a filter/host boundary is crossed) → Unit (RED-GREEN-REFACTOR) with concrete stubs, plus a local CI-parity sweep (fmt/clippy/type/test) before handoff.