القائمة
Boot a Konjo session for lopi. Produces a Session Brief, runs Discovery, identifies the next sprint. Use at the start of any work session or when invoked with /konjo.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Konjo sprint completion checklist and session handoff template. Use when closing out a sprint or ending a work session, in any consuming repo.
Documents the lopi CLI — a Rust multi-agent orchestrator that runs Claude Code agents concurrently in git-isolated branches. Use when the user asks to run a coding task through lopi, check on running/queued lopi agents, review lopi task history or logs, or interpret lopi's task status/output.
Retrofit the Konjo Quality Framework onto an existing repo that predates it. Use when asked to add konjo quality gates, improve code quality, audit an existing codebase, or run a quality sprint on any repo. Provides the step-by-step migration plan and triage protocol.
Full lopi project context — complete phase plan, current health metrics, standing rules for agents, crate details. Auto-load when working on lopi sprint planning, architecture decisions, or Phase 5+ work.
| name | konjo-boot |
| description | Boot a Konjo session for lopi. Produces a Session Brief, runs Discovery, identifies the next sprint. Use at the start of any work session or when invoked with /konjo. |
| user-invocable | true |
Read in order: CLAUDE.md, README.md, CHANGELOG.md, PLAN.md, MASTER_PLAN.md, docs/ (if it exists). Do not skip. Do not assume contents.
REPO lopi — high-performance Rust agent orchestrator for Claude Code
LAST SHIPPED [most recent meaningful change from CHANGELOG.md]
OPEN WORK [stated next steps from PLAN.md]
BLOCKERS [failing tests, broken modules, open issues]
HEALTH [Green / Yellow / Red — one line]
Unknown is stated as unknown. Fabricated state is a lie to the next session.
Before executing any sprint, ask:
Search: arXiv (agent scheduling, LLM orchestration, self-improving systems), GitHub (new Claude Code tooling, agent frameworks), HuggingFace (relevant models/benchmarks).
If PLAN.md exists: load it, validate against codebase, flag drift. If drift found:
PLAN DRIFT
✗ [item] appears completed — not marked done
✗ [item] references [module] that no longer exists
CORRECTED NEXT STEP: [what actually needs to happen]
If no plan: run Discovery Protocol → propose a sprint:
PROPOSED SPRINT [N — Name]
MOTIVATION [real problem, real user, real value]
RESEARCH [papers, repos, techniques informing this sprint]
DELIVERABLES [concrete, shippable, verifiable things]
SUCCESS CRITERIA [tests, benchmarks, features working end-to-end]
SCOPE / RISKS [Small / Medium / Large — what could block this]
Small/Medium: propose and proceed. Large or irreversible: propose and confirm.
This skill activates on any of:
konjokonjo lopilopi konjoread KONJO_PROMPT.md and begin