claude

agent-conduct

Domain-free skills governing how an agent conducts itself while working, independent of orchestration, memory, or any specific engineering domain.

4 on GitHubby ryzhakarv1.10.0MITupdated 23 days ago
/

Plugin agent-conduct from marketplace my-claude-skills in ryzhakar/claude-skills. Commands read from the repo's manifest — 23 days ago.

This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.

What it installs

Skills (6)

  • check-back

    Before ending a turn that leaves a run pending, set the agent's own next wake, and on waking look at what was pending. "check back", "check back in 10 minutes", "wake up when it's done", "keep an eye on the build", "wait for the agents", "don't lose track of background work", "pick up after a resume

  • decide-while-away

    Settle the owner's decisions while the owner is away: decide each reversible choice, write it beside its single closest alternative, act on it, and leave each choice that cannot be undone unsolved in a queue. Use when a step needs a decision that is the owner's to make and the owner has not given it

  • git-etiquette

    Commit, push, and integrate with git: revertible commits, ten-word subjects, zero attribution, a linear main, one worktree per sibling session. "commit this", "push it", "integrate to main", "rebase on main", "squash the records", "make a worktree", "open a PR", or any turn about to run a git comman

  • office-hours

    Hold an office-hours round: one event in which the owner answers the questions the agent queued. "office hours", "questioning round", or an agent about to ask whether the owner is present for a round; never a one-off question mid-task.

  • work-silently

    Keep working while writing nothing to the conversation, ending every turn with an empty message, until the user says to stop. Use when the user says "/work-silently", "go silent", "back to silence", "work in silence", "don't write anything until I write", "nothing but tool calls", "stay quiet and ke

  • write-well

    Write all prose by Strunk's rules of grammar, composition, and usage, and reread each draft against them, most serious fault first. Use when the user says "/write-well", "write-well", "write well", or asks for prose held to Strunk's rules.

Hooks (4)

  • PreToolUse

    python3 ${CLAUDE_PLUGIN_ROOT}/hooks/office-hours.py

  • SessionStart

    bash ${CLAUDE_PLUGIN_ROOT}/hooks/session-start.sh

  • SessionStart

    python3 ${CLAUDE_PLUGIN_ROOT}/hooks/check-back.py resume

  • Stop

    python3 ${CLAUDE_PLUGIN_ROOT}/hooks/check-back.py stop

View source on GitHub

More from ryzhakar/claude-skills

All 11 plugins →

instruction-eval

claude

Measure whether an LLM instruction changes model behavior.

11 skills

4

dev-cycle

claude

Development cycle orchestration: specify, plan, implement, review, integrate. A spec-capturer that interviews the owner, worktree-isolated implementers, spec and code-quality reviewers, and a review chain enforced by hooks.

1 skill · 4 agents · hooks

4

dev-cycle-lite

claude

Cost-tiered development cycle over dev-cycle: an opus prescriber writes prescriptions that leave no decision, a sonnet suborchestrator runs the loop and both reviews, disposable haiku executors follow the prescription literally. No hooks.

2 skills · 3 agents

4

dev-discipline

claude

Software engineering discipline skills, agent-agnostic: outermost-test-first signature-first TDD, systematic debugging, bug triage, architecture improvement, and receiving code review. Used ad hoc or by dev-cycle.

5 skills

4

qa-automation

claude

Playwright test lifecycle orchestrator. One skill drives the full loop: plan from live browser exploration, generate accessible .spec.ts files, execute with failure classification, and self-heal broken locators via deterministic ten-tier recovery with confidence-based PR routing.

1 skill · 4 agents

4

orchestration

claude

Agent delegation framework and multi-agent research orchestration. Decompose work across model tiers, manage parallel swarms, and govern quality.

2 skills

4

Similar plugins

hookify

claude

Easily create custom hooks to prevent unwanted behaviors by analyzing conversation patterns or from explicit instructions. Define rules via simple markdown files.

1 skill · 4 commands · 1 agent · hooks

anthropics/claude-code 155,476

nextjs

claude

Official Next.js skills: adopt and optimize Cache Components, adopt Partial Prefetching, and verify runtime behavior against a running dev server.

6 skills

vercel/next.js 143,495

fullstory

claude

Connect Claude to Fullstory to query behavioral analytics, session replays, and customer experience insights.

5 skills · 1 agent · 1 MCP server

anthropics/claude-plugins-official 47,076

skill-forge-essentials

claudecodex

Behavioral skills for AI-specific failure patterns: hidden code debt from agent generation and context compaction amnesia via behavioral self-enforcement

3 skills

wshobson/agents 42,461

Zocks Advisor Intelligence

claude

Seven skills for financial advisors built on Zocks meeting intelligence: client behavioral profiles, attrition-risk ranking, held-away asset radar, client journey dossiers, household review packs, next-best-action analysis and a tax opportunity scan. Requires the Zocks connector and a Zocks account.

7 skills · 1 MCP server

anthropics/knowledge-work-plugins 35,601

unreal

claude

Unreal Engine 5.8 skills (Blueprints, C++ gameplay, Enhanced Input, Behavior Trees with EQS and StateTree, Niagara, packaging).

80 skills

gamedev-skills/awesome-gamedev-agent-skills 8,642

Maintain agent-conduct?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex plugins11 plugins