- Skills
- 92
- 仓库
- 1
- 职业覆盖
- 6
- 场景覆盖
- 6
- Stars
- ★ 0
tzeusy-org 的 Skills 92
user-run
Use when a step needs privileges this session lacks (docker, sudo, another user's checkout or secrets, host access) and the owner must run it as tze. Writes a reviewable /home/orca/.tmp/<slug>/run.sh that impersonates tze (USER_TO_IMPERSONATE overrides), tees timestamped logs plus latest.log and collects artifacts there; Claude presents it with a risk and time brief, then continues from the logs. Triggers: "I'll run it for you", "permission denied on docker.sock", "needs sudo", "sudo: a password is required", "run as tze", "/user-run".
requesting-code-review
Use when completing tasks, implementing major features, or before merging to verify work meets requirements
receiving-code-review
Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation
finishing-a-development-branch
Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR, or cleanup
executing-plans
Use when you have a written implementation plan to execute in a separate session with review checkpoints
dispatching-parallel-agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
brainstorming
Use before creative work — creating features, building components, adding functionality, or modifying behavior — to explore user intent, requirements, and design before implementation. One question at a time; 2–3 approaches; design validated in sections.
th-workflow
Use for the day-to-day development workflow around a change, idea to merged branch: brainstorming intent and design before building or modifying a feature; executing a written implementation plan in checkpointed batches; dispatching parallel subagents for independent problems; requesting a code review before merge; verifying review feedback before acting on it; finishing a branch (merge, PR, keep, discard). Triggers: "brainstorm this feature", "design before building", "execute this plan", "these failures look independent", "request a review", "the reviewer says", "merge or PR?", "clean up this branch".
user-run-scripts
Use whenever a task needs something the agent's own sandbox cannot do — an interactive sudo/root password, another TTY prompt, an action blocked by sandbox or permission policy, or any manual out-of-band step (e.g. a 2FA code) — by handing the user a script to run themselves and reading its output back from disk instead of a copy/paste round-trip. Not for commands the agent can already run itself. Triggers: "I need sudo for this", "this needs your password", "run this yourself", "needs an interactive terminal", "sandbox won't let me do X".
transcript-retention
Use to manage the disk footprint of AI-tool session state — Claude Code transcripts (~/.claude/projects), Codex rollouts (~/.codex/sessions), shell snapshots, logs, and sqlite WAL files — by reporting usage and applying an age-based compress/delete policy. Triggers: "how big is my session history", "clean up old transcripts", "codex sessions are eating disk", "transcript retention".
refresh-snapshots
Use to re-run the harness's snapshot/mirror flows so installed state matches version-controlled source — skill symlinks into ~/.claude, ~/.codex, ~/.gemini tool homes, git submodule pointers, and shell plugin checkouts. Use after restructuring skills, when links look stale or broken, or on a periodic freshness pass. Triggers: "refresh snapshots", "relink skills", "update submodules", "my installed skills are stale".
memory-hygiene
Use to compact agent memory stores that grow append-only and rot — AGENTS.md "Notes to self" sections, bd memories, CLAUDE.md instruction drift — by deduplicating, merging superseded entries, and retiring notes whose subject no longer exists. Triggers: "compact my notes to self", "AGENTS.md is huge", "clean up bd memories", "stale agent memory".
dotfiles-review
Use to review the ~/.dotfiles tree — rc files, alias/function files under to_source/, home_rcs/, and .config tool dirs — for objective conflicts (duplicate or shadowing aliases, mutually exclusive settings, dead references) and best-practice recommendations, while never opening secret files. Triggers: "review my dotfiles", "clean up my zshrc", "are my aliases conflicting", "modernize my shell config".
config-surface-audit
Use to audit the non-skill recurring context surfaces of the AI tools — configured MCP servers (tool schemas load every session), permission allowlists, and hooks in settings.json / config.toml — against actual usage from transcripts, and recommend removing what never fires. Triggers: "audit my MCP servers", "are these permissions stale", "what hooks do I have", "config bloat".
audit-skill-hygiene
Use to produce a content-safe, reproducible measurement-only audit of cataloged skill usage from Claude Code and Codex. Emits an aggregate decision matrix; it never changes catalogs or recommends direct archival. Triggers: "audit skill usage", "which skills are actually used", "is our skill catalog bloated".
th-tooling
Use for hygiene of this machine's AI-tooling harness — the ~/.dotfiles repo, the ai-bootstrap skills tree, and the installed tool homes (~/.claude, ~/.codex, ~/.gemini): auditing real skill usage from transcripts, archiving unused skills, reviewing dotfiles for conflicting aliases/config, refreshing snapshot state (skill symlinks, submodules, mirrored assets), transcript disk retention, compacting agent memory (AGENTS.md notes, bd memories), and auditing MCP/permission/hook config against actual use. Route to exactly one subskill per task. Triggers: "audit my skills", "which skills are unused", "review my dotfiles", "clean up my zshrc", "refresh snapshots", "old transcripts eating disk", "compact my notes to self", "audit my MCP servers", "harness hygiene".
user-questionnaire
Use to accumulate genuine owner decisions into a local asynchronous review packet, publish that packet as a local web form the owner answers in a browser, walk the owner through it one question at a time, ingest the recorded answers in a later session, or route recorded signoffs through canonical owning workflows. Every problem scope and recommendation receives a separate adversarial verdict from an independent subagent against the target project's th-projects shape and th-engineering bar before it is review-ready. Triggers: "prepare decisions for me to review", "review when I wake up", "publish the questionnaire", "give me a link to review", "walk me through the blocked decisions", "ingest my answers", "collect owner signoffs", "decision packet", "questionnaire".
relentless-vision-pursuit
Use for a recurring, generative pursuit run that closes the gap between what a project is today and the ideal its doctrine describes: lane-sequenced audits of every surface against the applicable quality bar, vision-grounded ideation lenses, synthesis into a ranked move list with tier-board movement across runs, and a gated (never auto-released) work plan. Triggers: "run the vision pursuit", "what's the next best step toward the vision", "relentlessly improve this project", "audit the whole project against the ideal", "generate the next round of improvements".