DefaultPerson

@defaultperson source repo

17 skills
0 followers
9 hours ago last updated

▌
Svgl 2 · defaultperson bundle
Fetch SVG brand/tech logos from the svgl.app public API into a project — by name, by category, or just listing categories. Downloads the actual `.svg` file(s) locally (handling light/dark theme variants and optional wordmarks), or with `--json` returns metadata/URLs only. Tradeoff: depends on svgl.app's catalogue (~660 logos) and its undocumented rate limit; only public logos. For a single logo you already have a URL for, just curl it. Triggers: "svgl", "/svgl", "get logo", "fetch logo", "svg logo", "download logo", "логотип", "svg иконка", "достань логотип".

0
installs
▌
Cleanup 2 · defaultperson bundle
Use when you have a messy notes/plan/chat dump (50+ lines, mixed topics, no clear structure) that needs lossless reorganization into a clean sectioned markdown file. Slow and thorough — overkill for files under 30 lines or already-sectioned docs. Triggers: "cleanup", "/cleanup", "почисти", "реорганизуй", "clean up", "rewrite plan", "sort plan", "plan-rewrite", "/plan-rewrite".

0
installs
▌
Blueprint 2 · defaultperson bundle
Use when you have a clean spec/notes file and want a readable, implementation-ready PLAN: atomic vertical-slice tasks each with a `Done when:` shell proof, plain-language requirements ([must]/[nice]/[later]), edge cases, ranked assumptions + open questions, and risks. Reads like a plan, not an RFC. Suitable for human implementation or the Claude Code goal feature. Tradeoff: slow and thorough — overkill for tasks under 1 hour. Triggers: "blueprint", "/blueprint", "plan this spec", "составь план", "clarify", "/clarify", "уточни спеку", "enrich spec", "decompose spec".

0
installs
▌
Goal Prep 2 · defaultperson bundle
Codex CLI variant. Compile a charter for Codex's native /goal command. Runs an Intake Compiler (11 fields), a diagnostic ladder (6 questions in 3 batches), classifies the goal, seeds per-kind anti-patterns, and emits a copy-pasteable native /goal command pointing at the on-disk charter (anti-patterns + completion audit live in the charter file). Strictly non-execution: never performs the user's requested work, never reads implementation files, never browses reference repos. Triggers: "goal-prep", "/goal-prep", "prepare goal", "intake", "compile goal charter", "подготовка цели", "интейк", "прогрев цели".

0
installs
▌
Verify Done 2 · defaultperson bundle
Plan-aware final ACCEPTANCE gate: given any plan (a /blueprint plan, a goal.md charter, OR a plain plan-mode / inline plan) + the built code, adversarially verify the result actually works against the ORIGINAL intent — re-run every `Done when:` proof (Tier 1), generate + run independent scenarios the plan may have missed (Tier 2), then an advisory maintainability pass (Tier 3). Read-only GATE, not a fixer — emits DONE / NOT-DONE + a gap list. Triggers: "/verify-done", "verify done", "accept", "acceptance gate", "is this actually done", "приёмка".

0
installs
▌
Autoresearch 2 · defaultperson bundle
Codex CLI variant. Autonomous keep-or-revert experiment loop. Iteratively optimizes any metric through single atomic changes, committing before verification and reverting failures. Adapted from agent-workflow autoresearch. Codex has no scheduling primitives (no /loop, no cron), so this variant runs an in-session sequential loop — Phases 1-7 execute back-to-back until max_iterations, convergence, or session close. Multi-day runs require manual /autoresearch --continue per fresh session; the scratchpad on disk survives. Triggers: "autoresearch", "/autoresearch", "auto research", "optimize metric", "keep or revert loop", "автоисследование", "оптимизируй метрику".

0
installs
▌
Extract Links 2 · defaultperson bundle
Use when a notes/plan file contains URLs (YouTube videos, Telegram posts, articles) and you want their gist available inside the file. By DEFAULT writes a one-line summary inline next to each URL (light, fast triage); pass --full to pull full content into a local extracted/ tree with pointers. YouTube subtitles via yt-dlp, public Telegram via embed-page scrape, HTML via pandoc/curl; private/auth resources are out of scope. For 1-2 URLs, copy-paste is faster. Triggers: "/extract-links", "extract links", "развернуть ссылки", "expand links", "fetch URLs", "извлеки контент".

0
installs
▌
Ship · defaultperson
Autonomous end-to-end orchestrator: hand it ONE freeform task and it picks the right skills, chains them, and drives the work to a verified finish — choosing the entry point itself (native plan mode / cleanup / extract-links / blueprint / goal-prep + /goal) and always ending with /verify-done. The autonomy level is chosen INTERACTIVELY at the start (guided / autopilot / checkpoint). Use ONLY when you want a well-scoped, RECOVERABLE task done start-to-finish with minimal hand-holding — not for exploratory or high-stakes work you want to review step by step (run the skills directly for that). Claude Code only. Triggers: "/ship", "ship it", "do the whole thing", "from start to finish", "handle it end to end", "autonomously", "сделай всё сам", "доведи до конца сам".

0
installs
▌
Svgl · defaultperson bundle
Fetch SVG brand/tech logos from the svgl.app public API into a project — by name, by category, or just listing categories. Downloads the actual `.svg` file(s) locally (handling light/dark theme variants and optional wordmarks), or with `--json` returns metadata/URLs only. Tradeoff: depends on svgl.app's catalogue (~660 logos) and its undocumented rate limit; only public logos. For a single logo you already have a URL for, just curl it. Triggers: "svgl", "/svgl", "get logo", "fetch logo", "svg logo", "download logo", "логотип", "svg иконка", "достань логотип".

0
installs
▌
Herdr · defaultperson
Control herdr — the terminal-native agent multiplexer — from inside a herdr-managed pane. Manage workspaces and tabs, split panes, spawn sibling agents (including more Claude Code instances), read another pane's output, run servers/tests in siblings, and block until output appears or another agent reaches a status — all via the `herdr` CLI talking to the running herdr instance over a local unix socket. Use ONLY when running inside herdr (environment variable HERDR_ENV=1); otherwise do not inspect or control the focused pane. Triggers: "herdr", "/herdr", "spawn an agent", "split a pane", "run in another pane", "wait for the server to be ready", "coordinate with another agent", "what is the other agent doing", "запусти в соседней панели", "спавни агента", "координируй агентов".

0
installs
▌
Babysit · defaultperson
Autonomous watch → fix → deploy → recheck LOOP for a running service: every N minutes it reads only NEW logs and, on trouble, delegates a bounded fix to a sub-agent, ships it via your `deploy_cmd` (or a deploy MCP), and verifies — repeating until you stop it. Escalates with a repeating sound + desktop notification only when stuck or a change is unsafe. Use ONLY when the user explicitly wants this ongoing UNATTENDED loop — not a one-off log read, health check, or single fix. Claude Code only (needs the native /loop). Triggers: "/babysit", "babysit the service", "watch and auto-fix/deploy", "keep the service healthy unattended", "присматривай за сервисом".

0
installs
▌
Cleanup · defaultperson bundle
Use when you have a messy notes/plan/chat dump (50+ lines, mixed topics, no clear structure) that needs lossless reorganization into a clean sectioned markdown file. Slow and thorough — overkill for files under 30 lines or already-sectioned docs. Triggers: "cleanup", "/cleanup", "почисти", "реорганизуй", "clean up", "rewrite plan", "sort plan", "plan-rewrite", "/plan-rewrite".

0
installs
▌
Blueprint · defaultperson bundle
Use when you have a clean spec/notes file and want a readable, implementation-ready PLAN: atomic vertical-slice tasks each with a `Done when:` shell proof, plain-language requirements ([must]/[nice]/[later]), edge cases, ranked assumptions + open questions, and risks. Reads like a plan, not an RFC. Suitable for human implementation or the Claude Code goal feature. Tradeoff: slow and thorough — overkill for tasks under 1 hour. Triggers: "blueprint", "/blueprint", "plan this spec", "составь план", "clarify", "/clarify", "уточни спеку", "enrich spec", "decompose spec".

0
installs
▌
Goal Prep · defaultperson bundle
Compile a charter for Claude Code's native /goal command. Runs an Intake Compiler (11 fields), a diagnostic ladder (6 questions, in 3 batches), classifies the goal, seeds per-kind anti-patterns, and emits a copy-pasteable native /goal command pointing at the on-disk charter (anti-patterns + completion audit live in the charter file). Strictly non-execution: never performs the user's requested work, never reads implementation files, never browses reference repos. Triggers: "goal-prep", "/goal-prep", "prepare goal", "intake", "compile goal charter", "подготовка цели", "интейк", "прогрев цели".

0
installs
▌
Verify Done · defaultperson bundle
Plan-aware final ACCEPTANCE gate: given any plan (a /blueprint plan, a goal.md charter, OR a plain native plan-mode / inline plan) + the built code, adversarially verify the result actually works against the ORIGINAL intent — re-run every `Done when:` proof (Tier 1), generate + run independent user-scenarios the plan may have missed (Tier 2), then an advisory maintainability pass (Tier 3). Read-only GATE, not a fixer — emits DONE / NOT-DONE + a gap list. Heavy verification runs in a Workflow's sub-agents, so the main session gets only the verdict. Triggers: "/verify-done", "verify done", "accept", "acceptance gate", "is this actually done", "verify the whole plan", "приёмка", "проверь что готово".

0
installs
▌
Autoresearch · defaultperson bundle
Autonomous keep-or-revert experiment loop. Iteratively optimizes any metric through single atomic changes, committing before verification and reverting failures. Adapted from agent-workflow autoresearch with the Ralph Stop-hook dependency removed — now ScheduleWakeup-paced (short iterations) or CronCreate-paced (multi-hour iterations). Each wake-up reads state from disk, so context-window growth never blocks the loop. As of v0.2.0, per-iteration MODIFY+COMMIT is delegated to a bounded subagent (iron-skills:autoresearch-worker) that returns a structured final message; the main loop's context stays constant across iterations. Triggers: "autoresearch", "/autoresearch", "auto research", "optimize metric", "keep or revert loop", "автоисследование", "оптимизируй метрику".

0
installs
▌
Extract Links · defaultperson bundle
Use when a notes/plan file contains URLs (YouTube videos, Telegram posts, articles) and you want their gist available inside the file. By DEFAULT writes a one-line summary inline next to each URL (light, fast triage); pass --full to pull full content into a local extracted/ tree with pointers. YouTube subtitles via yt-dlp, public Telegram via embed-page scrape, HTML via pandoc/curl; private/auth resources are out of scope. For 1-2 URLs, copy-paste is faster. Triggers: "/extract-links", "extract links", "развернуть ссылки", "expand links", "fetch URLs", "извлеки контент".

0
installs

17 published skills