typeclaw

Use this skill BEFORE every `channel_reply` or `channel_send` call whose adapter is `github`, AND before composing replies to GitHub-originated inbounds, AND before opening new issues with model-driven `gh` or preparing a PR for host-stage operator creation, AND ALWAYS when you are asked to review a PR — whether the inbound says "requested your review on PR

The ONLY supported way to render Markdown into a polished, professional PDF (and optionally attach it to a channel). Load this whenever you need to deliver a document as a PDF rather than raw markdown — reports, summaries, briefs, meeting notes, docs, render report, export document, anything a human would want to download, print, or forward, including a researcher's report file shipped as a Slack/Discord attachment. Triggers: 'make a PDF', 'export to PDF', 'markdown to PDF', 'PDF report', 'render report', 'export document', 'the report', 'attach the report', 'send me a PDF', 'as a PDF', 'turn this into a document', 'make it look good', 'beautiful PDF', 'nicer PDF', a researcher/subagent result you want to ship as a file, 'PDF로', 'PDF로 만들어', 'PDF로 변환', 'PDF 첨부', '리포트', '보고서', '예쁘게'. Provided by the bundled `doc-render` plugin: a small Typst toolchain is installed on first use via `bun add` (no PDF library is baked into the image), a bundled themed report library does the styling, and a bundled render script do

Use this skill whenever you decide to delegate substantial coding or code-analysis work to Claude Code (Anthropic's official coding-agent CLI). Triggers include "use Claude Code", "ask Claude Code", "delegate to claude", "claude cli", "have claude do it", any task where you want a more capable agent than yourself, and any time you're about to run `claude` from a shell. Read it before you spawn the CLI — Claude Code is a TTY-only TUI in interactive mode (you must drive it through tmux, not pipes), it operates inside a dedicated `git worktree` checkout under `/tmp/` so its commits never pollute the agent folder, and you detect "turn done" through a `Stop` hook that writes a sentinel file. Skipping this skill means you'll either fall back to `claude -p` (which strips plan mode and sub-agents), let claude mutate the live agent checkout (which loses you the rollback safety), or try to parse the TUI buffer with capture-pane heuristics (fragile, version-locked).

Use this skill whenever you decide to delegate substantial coding or code-analysis work to Codex CLI (OpenAI's official coding-agent CLI, `@openai/codex`). Triggers include "use Codex", "ask Codex", "delegate to codex", "codex cli", "have codex do it", "openai codex", any task where you want a second-opinion agent on top of (or instead of) Claude Code, and any time you're about to run `codex` from a shell. Read it before you spawn the CLI — Codex is a TTY-only TUI in interactive mode (you must drive it through tmux, not pipes), it operates inside a dedicated `git worktree` checkout under `/tmp/` so its commits never pollute the agent folder, and you detect "turn done" through a `Stop` hook that writes a sentinel file. Skipping this skill means you'll either fall back to `codex exec` (which is single-shot and strips multi-turn agency), let codex mutate the live agent checkout (which loses you the rollback safety), or try to parse the TUI buffer with capture-pane heuristics (fragile, version-locked).

Read or edit typeclaw.json — the host-stage runtime config: model, port, mounts, plugins, alias, channels, portForward, docker.file (container package toggles + append), git.ignore.append, plus provider credentials (secrets.json/.env) and the allowed-models registry. Strict schema with a mix of live-reloadable and restart-required fields — load before touching the file or you risk corrupting it or promising a behavior the runtime won't deliver. Also the authority on what a field defaults to and whether a behavior is already on out of the box (port forwarding, container packages, model choice) — load before saying you don't know what X defaults to, or before proposing to add a field whose default the user is asking about; most fields already default to the expected behavior, so the answer is usually 'no edit needed'. Owns the GitHub channel config — which repos it watches (channels.github.repos), the code-review trigger (channels.github.review.on/approve), webhook auto-registration via a tunnel — load on 'gith

Use this skill BEFORE you open a new issue or pull request on GitHub with `gh issue create` / `gh pr create` (or before composing the `--body`/`--title` for one, or editing an issue/PR body after the fact). Triggers include any `gh issue create`, `gh pr create`, `gh api … /issues`, `gh api … /pulls`, any time you are about to file a bug report, feature request, or PR against a repo — yours or someone else's — and any phrasing like "open an issue", "file a bug", "raise a PR", "submit a pull request". Read it first, because every repo has its own contribution rules — an issue/PR template, a CONTRIBUTING.md, a title convention — and a maintainer's first impression of your contribution is whether you bothered to follow them. Ignoring a template that the repo author wrote on purpose reads as careless; following it reads as someone who belongs. This is platform etiquette, independent of how the work reached you (it applies whether the request came through the github channel or you decided to file on your own).

Use this skill whenever the user asks who you talk to, why you went silent in a channel, why a tool call was blocked with `blocked:` / "denied by permissions", how to grant access, what a role can or can't do, or whenever you are about to edit the `roles` block in `typeclaw.json`. Triggers include "who can talk to you", "why aren't you replying in

Use this skill whenever the user asks what you remember, what you forgot, what you dreamed, why a fact is or isn't in your memory, when memory consolidation happens, or whenever you are about to read or write `MEMORY.md`, anything under `memory/`, or `memory/skills/`. Triggers include "what do you remember", "do you remember X", "forget that", "what did you dream", "when do you dream next", "why did you forget X", "edit MEMORY.md", "add to memory", "your daily streams", "memory-logger", "dreaming", "muscle memory", or any mention of `memory.idleMs` / `memory.bufferBytes` / `memory.dreaming.schedule` in `typeclaw.json`. Read it before you touch any memory file — `MEMORY.md` and `memory/yyyy-MM-dd.jsonl` are runtime-owned, hand-edits are easy to do wrong, and the user almost always means something more specific than "edit memory" when they say it.

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction. Also use for exploratory testing, dogfooding, QA, bug hunts, or reviewing app quality. Also use for automating Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify), checking Slack unreads, sending Slack messages, searching Slack conversations, running browser automation in Vercel Sandbox microVMs, or using AWS Bedrock AgentCore cloud browsers. ALSO use whenever a browser step needs a human in the loop — login walls, 2FA, CAPTCHA, payment confirmation, "is this the right button?" ambiguity, or the user asking to

Use this skill when you are stuck in a fix-it loop — you've made roughly three attempts at the same failure and you're still cycling shell commands (kill the process, re-run, `sleep`, `capture-pane`, inspect, retry) without converging. Triggers include a hung or runaway process that won't die, a `C-c` that didn't stop the program, `<defunct>`/zombie processes piling up in `ps`, an interactive program that blocks `bash` waiting for input, a script that "ran" but produced no output and no file, repeated "not found"/timeout/same-error-again loops, and any moment you catch yourself thinking "let me wait a bit more and check again" for the third time. Read it before you spawn `operator` to take over the debugging — it covers the operator hand-off prompt, the tmux session pattern, killing stuck/zombie processes properly, and the edge-triggered capture-pane polling loop that the inline retry-and-sleep approach gets wrong.

TypeClaw plugin authoring and operation guide. Use when writing, editing, configuring, debugging, or installing a TypeClaw plugin — including any work with definePlugin, defineTool, defineSubagent, plugin hooks (session.start/end/idle/prompt, tool.before/after), plugin cron jobs, plugin commands (host/container/either CLI subcommands callable as `typeclaw <name>`), plugin skills, the typeclaw/plugin import path, or per-plugin config blocks in typeclaw.json. Also use when you need to bridge a cron `exec` job to LLM-driven work — the canonical pattern is a `surface: 'container'` plugin command whose `run` calls `ctx.prompt(...)`, invoked as `typeclaw <command>` from cron's `command` array. Triggers on mentions of 'TypeClaw plugin', 'definePlugin', 'plugin hook', 'plugin cron', 'plugin command', 'PluginCommand', 'ContainerCommand', 'HostCommand', 'EitherCommand', 'ctx.prompt', 'ctx.subagent', 'ctx.exec', 'plugins[]', 'typeclaw-plugin-', or any file under src/plugin/ or plugins/.

Use this skill BEFORE every `channel_reply` or `channel_send` call whose adapter is `instagram`. Instagram DMs render messages as plain text — no markdown, no surfaced @-mention syntax, no threads, and no outbound attachments or stickers. Read this skill before composing anything on Instagram.

TypeClaw channel behavior: how the agent decides to engage vs. stay silent on external messenger inbound (Discord, Slack, Telegram, LINE, KakaoTalk). Covers the `channels.<adapter>.engagement` triggers (mention/reply/dm), reply stickiness, the non-configurable solo-human fallback, history-prefetch windows, and the `alias` system — plain-text names the agent answers to, substring match semantics, peer-name suppressors, and engagement priority. Load when the user asks why the agent did or did not respond in a channel, wants to change when it auto-replies, asks to 'be quieter'/'stop auto-replying', wants it to answer to a nickname, or mentions engagement, stickiness, aliases, mentions, trigger words, suppressors, or '응답', '호출', '채널', '별칭', '왜 답을 안 해'. Access control (who is admitted at all) lives in `roles` — see typeclaw-permissions. The `channels`/`alias` schema, defaults, and safe-edit workflow live in typeclaw-config.

Use this skill whenever the user asks you to schedule recurring work OR a one-off future task/reminder, run something on a cron, do something every day/hour/week, do something once at a future time, set up a periodic task, list or inspect scheduled jobs, or read or edit your cron schedule — AND whenever the user wants you to STOP, disable, pause, or remove a recurring/scheduled job, or complains that you "keep" posting, sending, or doing the same thing on a schedule (a repeating message you send every N minutes is almost always a cron job). Triggers include "every morning", "every Monday", "schedule a", "remind me every", "set up a cron", "run X periodically", "remind me in 3 days", "remind me tomorrow", "remind me at 9am", "remind me next Monday", "in N hours/days do X", "do X once at hh:mm", "stop after N times", "until <date>", "list cron jobs", "list scheduled jobs", "show me the cron", "what cron jobs do you have", "what's on my cron", "when does X run", as well as STOP/DISABLE triggers in any language l

Use this skill whenever the user asks you to install, find, list, update, or remove an agent skill, whenever you yourself want to add a new capability via a skill, or whenever you are about to edit any file under `.agents/skills/`. Triggers include "install skill", "add a skill", "find a skill", "update skills", "remove skill", "skill from <repo>", any mention of the `skills` CLI / `bunx skills`, any reference to `SKILL.md`, `skills-lock.json`, `.skill-lock.json`, and any time you read or write under `.agents/skills/<name>/`. Read it before you touch a skill — TypeClaw has three skill layers with different ownership rules, and editing a skill that the `skills` CLI manages will silently get overwritten on the next `bunx skills update`.

Use when the user mentions tunnel, ngrok, webhook URL, cloudflared, expose to internet, show my friend, public URL, GitHub webhook, port forward to public, reverse proxy, trycloudflare, or making a container-local service reachable from the internet. Read it before suggesting tunnel add/remove/status/logs or editing typeclaw.json tunnels[]. Also read it the moment a tunnel "doesn't work": a Cloudflare tunnel with no public URL usually means `cloudflared` was never baked into the image — it is opt-in (`docker.file.cloudflared`, default false), so a hand-added tunnel needs it set explicitly. Diagnose root cause by reading typeclaw.json + checking `command -v cloudflared` rather than trusting a single error line; tell the user to set `docker.file.cloudflared: true` and `typeclaw restart`; never curl/vendor cloudflared yourself or report a cryptic error as if the tunnel were down.

Use this skill whenever you are about to write code that another part of your agent folder might want to reuse, or you need to decide where new code goes. Triggers include "create a package", "extract this into a library", "make this reusable", "where should this script live", "add a workspace", "edit root scripts", "bun workspaces", "packages/", "monorepo", or any tool/utility/library you intend to call from more than one place. Read it before scaffolding anything under `packages/` — the layout has conventions for plugins, custom scripts, and root-level wiring that you must not improvise around.

Use this skill BEFORE every `channel_reply` or `channel_send` call whose adapter is `line`. LINE renders messages as plain text — `**bold**`, `## headings`, `| tables |`, fenced code blocks, and other markdown all appear literally. There is no `@mention` syntax, no message threads, no replies-with-quote, and no outbound attachments or stickers. Inbound non-text content (images, stickers, files) is not fetchable through this adapter. Read this skill before composing anything on LINE.

Load this skill when your `SOUL.md` (or the current conversation) calls for a warm, cute, adorable, playful, or affectionate tone — or when the user explicitly mentions kaomojis, emoticons, or asks you to "feel more like a person and less like a chatbot." TypeClaw's name puns on "Type" — typed emoticons fit. Triggers include any-language requests for a "cute", "adorable", "warm", "playful", "soft", or "cozy" tone, or the word for kaomoji/emoticon in the user's language (e.g. English "kaomoji"/"emoticon", Korean "카오모지"/"친근하게"/"귀엽게"/"다정하게", Japanese "顔文字"/"かわいく", Chinese "颜文字"/"可爱", Spanish "emoticono"/"tierno"), or any signal that the user is tired of generic AI emoji slop (🚀✨🎉) and wants real texture in your voice. This skill gives you a curated palette and the rule for using it: prefer kaomojis over generic emojis, but mix freely — kaomojis lead, emojis still allowed, neither is mandatory.

Use this skill whenever you edit any file in the agent folder — config, cron, memory, identity, scaffolded markdown, scripts, anything tracked by git. Triggers include any `write`/`edit` tool call against a non-gitignored file, any time you finish a logical change, or any mention of "commit", "git", "version control", "history", or "what did I change". Read it before you edit, not after — the rule is *commit immediately after every edit, with decision context in the message*. Skipping the commit pollutes the next one; skipping the decision context makes the history useless to future-you.

Use this skill BEFORE every `channel_reply` or `channel_send` call whose adapter is `kakaotalk`, AND before fetching/viewing KakaoTalk inbound attachments. KakaoTalk renders messages as plain text — `**bold**`, `## headings`, `| tables |`, fenced code blocks, and other markdown all appear literally. There is no `@mention` syntax, no message threads, no replies-with-quote, and no outbound stickers. Outbound file attachments (photos, videos, audio, generic files, multi-photo galleries) ARE supported — pass them via `attachments[]` on `channel_send` / `channel_reply` and the adapter routes by MIME. Inbound attachments appear as `[KakaoTalk attachment

How replies sent through the `telegram-bot` channel adapter actually render in Telegram. Read this BEFORE composing a reply that contains formatting markers — `**bold**`, `*italic*`, `_italic_`, `` `code` ``, fenced code blocks (```), `[label](url)`, `~~strike~~`, `||spoiler||`, `# heading`, `- list`, `| table |`, raw `.` `!` `(` `)` `_` `*` punctuation in URLs/IDs, snake_case identifiers — and your draft is going to a Telegram chat or supergroup. Also load if a Telegram user reports your message arrived as raw markdown (literal `**asterisks**`), as garbled text with stray backslashes, or with a Telegram error like `Bad Request: can't parse entities`. Covers: which markers actually render, which ones are stripped or escaped, what Telegram has no equivalent for (headings, bulleted/numbered lists, tables — they fall through as escaped literals), how the adapter's MarkdownV2 escaping behaves, and how to keep your message readable when the rendering and the source diverge.