takeokunn
- 26 skills
- 0 followers
- 6 hours ago last updated
- ▌ Cold Read · takeokunnA context-free reviewer reads durable prose as its real audience; the writer applies surviving cuts. Use at task completion whenever docs, README, comment blocks, commit messages, or PR/issue bodies were written or revised.
- ▌ Effect TS · takeokunnUse when writing or reviewing Effect (Effect-TS) code (Effect.Service definitions, Layer composition, Effect.scoped resource handling, converting try/catch/async to the Effect error channel, Schema-as-SSOT type derivation, or testing with @effect/vitest and TestClock).
- ▌ Fact Check · takeokunnUse for claim verification, source checks, and validation of external references with Context7 or WebSearch.
- ▌ Sbcl Usage · takeokunnUse for SBCL execution and debugging (--script usage, REPL workflows, backtraces, ASDF loading, save-lisp-and-die, profiling, SLY development, terminating an unresponsive SBCL, or sb-thread/sb-cover hazards). Complements common-lisp-ecosystem's runtime operations.
- ▌ Define Core · takeokunnShared workflow phases and patterns for the /define command. Use this skill when implementing /define to ensure consistent workflow structure, agent delegation, and requirements documentation patterns.
- ▌ Serena Usage · takeokunnUse for Serena MCP work - semantic symbol search, find references, code navigation, memory read/write, and organizing memory as a linked reference graph rather than a flat list. Also covers recovering a subagent's report from its session transcript when a completion notification is lost.
- ▌ Core Patterns · takeokunnUse when authoring an agent or command that needs the shared decision-criteria or escalation template (copy them in, since a bare reference to this skill resolves to nothing at runtime). Also covers modelling absence without an in-range sentinel, safe alternatives to destructive Git commands, and when to escalate a review into an independent refutation pass.
- ▌ Nix Ecosystem · takeokunnUse when writing Nix expressions, flake.nix, home-manager config, programs.*/services.* modules, nixpkgs packaging, or nix flake check (including vacuous flake checks missing a target platform, overlays vs system packages, and activation-script hazards).
- ▌ SQL Ecosystem · takeokunnUse when working with SQL databases (SELECT/INSERT/UPDATE/DELETE, CREATE TABLE, JOIN, INDEX, EXPLAIN, transactions, or migrations) across PostgreSQL, MySQL, and SQLite.
- ▌ Context7 Usage · takeokunnUse for Context7 MCP documentation retrieval (current library docs, API signatures, version-specific behavior, or migration notes) to avoid stale API assumptions.
- ▌ Rust Ecosystem · takeokunnUse when working with Rust projects (Cargo.toml, rustc, cargo build/test/clippy/rustfmt, borrow-checker errors, lock-registry design, checked/saturating arithmetic on untrusted sizes, or Rust language patterns).
- ▌ Test Integrity · takeokunnUse when a test suite is green but its value is in doubt (false greens, zero tests collected, vacuous assertions, dead guards, or mocks that make the code under test unnecessary). Contrast testing-patterns (designing tests); this skill asks whether a passing suite proves anything.
- ▌ Emacs Ecosystem · takeokunnUse for Emacs Lisp, init.el, use-package, and Emacs runtime hazards such as hook ordering, condition-case versus quit, overlays versus text properties, buffer-local state, keymap precedence, and subprocess handling. Also covers macro hygiene and MELPA recipe and release gates.
- ▌ AI Slop Detector · takeokunnUse when auditing already-written prose or code for the tells output_discipline (ai-prompts/CLAUDE.md) bans, rather than applying the norm while drafting. Covers grep patterns per tell, the quoted-example false-positive trap, and code-artifact slop (dead branches, needless abstraction, restated docstrings, scaffolding).
- ▌ Testing Patterns · takeokunnUse when writing, structuring, or reviewing tests - test strategy, coverage, unit/integration/e2e split, mocks/fixtures/fakes, flaky-test isolation, async settlement. For whether a green suite actually proves anything, see test-integrity instead.
- ▌ Trust Boundaries · takeokunnUse when untrusted input crosses a privilege boundary, such as parsing client or network requests, decoding payloads or deserialization, expanding archives, restoring persisted files, evaluating input via eval, validating paths, URLs, or schemes, or guarding against TOCTOU. Covers fail-closed validation, decode budgets, format-template injection, and log injection.
- ▌ Technical Writing · takeokunnUse when writing a blog post, technical article, tutorial, report, PR/issue body, or doc/comment prose, in English or Japanese (includes general prose mechanics, a Japanese prose-quality ruleset, and a long-form structure ruleset for books and serials).
- ▌ Workflow Patterns · takeokunnPatterns for output formats, reflection checkpoints, agent references, and self-evaluation shared across agents and commands. Use when authoring an agent or command, not when executing one.
- ▌ Execution Workflow · takeokunnLoad at the start of implementing or delegating a task, and when judging whether work is done. Covers orchestration phases, verification gates, worktree and branch isolation, and code review standards. Not for authoring agents or commands, see workflow-patterns for that.
- ▌ Terraform Ecosystem · takeokunnUse for Terraform or OpenTofu HCL configuration and provider development (Go, terraform-plugin-framework). Covers state management, plan and apply, failed-apply recovery, moved blocks, import, and provider schema design.
- ▌ Common Lisp Ecosystem · takeokunnUse for Common Lisp, SBCL, or Coalton, covering CLOS, ASDF, defpackage and defsystem. Also covers package hygiene, condition design, format-string injection, Unicode predicates, reader macros, macro-argument evaluation hygiene, atomic file publishing, and hash-table key safety.
- ▌ Investigation Patterns · takeokunnUse when tracing a symptom to its cause (debugging, bisecting, or working out how an unfamiliar implementation behaves). Covers evidence-based analysis, bisecting a symptom whose boundary moves between runs, a probe that measures its own gate rather than the phenomenon, and checking a completion claim against the artifact instead of the summary.
- ▌
- ▌ Technical Documentation · takeokunnUse when writing or auditing technical documentation - README, API docs, design docs, specifications, user guides, in English or Japanese. Also covers auditing docs against code for both over-claims and stale future-work claims, and retired-name registries.
- ▌ Parallelization Patterns · takeokunnPatterns for parallel execution (deciding what is genuinely independent, settling disagreement between concurrent agents, retry policy, and scheduling data-parallel work across skewed inputs).
- ▌ Performance Benchmarking · takeokunnUse when producing, gating, or reviewing performance numbers - before/after benchmarks, speedup/regression claims, CI benchmark gates, noise floor and confidence interval, wall-clock vs allocation-count metrics, and wording a performance claim.