Claude Skills by markphelps

A `CLAUDE.md` or `AGENTS.md` file is loaded into an agent's context on every task. Different agents and repos use different filenames, and some repos keep one as a symlink or compatibility alias to the other. Treat them as the same class of file: high-leverage repo instructions for coding agents. A good one silently prevents whole classes of mistakes; a bloated or vague one wastes the context budget and gets ignored. This skill audits the target context file against a known-good structure, de...

Reduce repeated thinking across the active vault. This skill is for semantic overlap, not just exact duplicate text: compact passages when they do the same job in the knowledge system. Portent reference: `../references/portent-knowledge-base-spec.md`. Prefer compaction when files represent the same Portent object or repeat the same claim for the same object. Do not merge merely because objects share a Topic. Preserve `belongs_to` and `related_to` metadata when absorbing files.

Maintain canonical concept pages and surface recurring themes

Categorize captures in `raw/sources/`, route owned/user-authored material to the right vault directories, and synthesize tagged external sources into curated notes while preserving complete source evidence. Portent reference: `../references/portent-knowledge-base-spec.md`.

Hygiene pass over active notes, ideas, and projects

Run a comprehensive weekly maintenance sweep across the active vault surface. This is the unified entry point for periodic vault care, coordinating ingestion, hygiene, semantic compaction, project lifecycle management, and concept promotion into a single bounded workflow. Portent reference: `../references/portent-knowledge-base-spec.md`.

Research a target note, idea, project, or topic. Preserve external sources as traceable source records by default; write summaries only when the user asks for synthesis, assessment, or a concise research brief. Any source used in a summary keeps a complete immutable copy in `raw/processed/YYYY-MM-DD/`. Notes tagged `external` are treated as clipped/imported external sources and are eligible for synthesis while preserving the complete source record. Portent reference: `../references/portent-kn...

Unified PORT object lifecycle management: track Projects, Operations, Responsibilities, and externally tracked Task references; detect mismatches between filesystem location, Portent metadata, and active navigation; propose transitions; and coordinate archival. Manage lifecycle and navigation for PORT objects: `Project`, `Operation`, `Responsibility`, and externally tracked `Task` references. Portent reference: `../references/portent-knowledge-base-spec.md`.

Use this skill in one of two modes: - **Capture mode:** Use the bundled TypeScript helper to review a bounded slice of the authenticated user's X bookmarks and capture selected items into `raw/sources/`. - **Prune mode:** Triage existing X bookmark source records in `raw/sources/` using LLM judgment, deleting only records that are clearly irrelevant or too thin to be useful. This is source-first: the helper writes `external` source records, records captured bookmark IDs in `raw/state/x-bookma...

Design, build, or review command-line interfaces using the Command Line Interface Guidelines from clig.dev. Use this skill whenever the user is creating a new CLI, adding commands/subcommands/flags, designing help text, choosing stdout/stderr/exit-code behavior, adding JSON or plain output modes, improving error messages, making destructive operations safer, or asking whether a CLI feels good, scriptable, discoverable, or Unix-friendly.

Fix review feedback and CI failures on an existing GitHub pull request. Work on the PR's source branch. Do not open a new PR.

A README is a landing page, not documentation. Most visitors give it about ten seconds before deciding to scroll past, star it, or actually try it. The job of this pass is to make sure that in those ten seconds, a developer who has never heard of the project understands three things — **what is this, why does it exist, how do I try it** — and can act on the third one immediately. Everything else is secondary. This draws on patterns from repos that consistently convert well: Vite, shadcn/ui, t...

Getting a repo "open source ready" isn't about the license — that's a five-minute checkbox. It's about DevEx: whether a developer who has never talked to the maintainer can understand the project, run it, and land a contribution without having to ask a question first. Every piece of friction you remove here is a contributor you don't lose. This skill audits a repo against that bar, using patterns pulled from repos that are genuinely good at this (Vite, shadcn/ui, tRPC, Sentry, Supabase, Rust,...

Audit a project you build with AI agents by mining your own local agent session logs (Claude Code, Pi, Codex) for papercuts, bugs, and design flaws — the friction visible in what you actually typed rather than what the README claims. Use this when the user wants to find rough edges in their own tool or project, asks "what's annoying about X," wants to know why they keep working around their own software, is preparing a UX or DX pass on something they use daily, or wants to turn a week of agen...

When the user asks about their notes, mentions brainstorming, or wants to work on a topic, first gather context about their existing vault.