Language: English | 한국어
Speak the name of cloud-gathering Zeus, and the entire pantheon descends — Zephermine drafts the spec, Poseidon raises the fleet, Argos counts every plank, Minos judges every test, and Clio carves the whole story into bronze.
A production agent harness for Claude Code, Codex CLI, and Gemini CLI — a harness-engineering and loop-engineering stack, named after the Twelve Olympians, forged across 3 months of daily real-product builds.
/zeus "Build a shopping mall. React + Spring Boot + PostgreSQL"One line. Twelve gods. Design → Implement → Inspect → Test → Ship. No questions asked. No blueprints needed. No human in the loop.
Built for the loop-engineering era. Agents don't run on a single prompt — they run on loops: act → observe → verify, repeated until an objectively verifiable completion criterion is met. Olympus was built that way from the start — Chronos (verification gate that runs real tests), Minos (fix-until-pass), Argos (AC cross-check), Clio (GO/NO-GO). Since v4.7.0 these loops run on top of CLI-native features (
/goalstop gate,/code-review, Agent Teams) — the harness is the foundation, the loop is the operating model. And since v4.8.0 the loops know how to stay running: a loop survives on structure, not willpower — heartbeat in the machine, state in the audit log, blocked issues parked with decision-ready briefs, and false completion claims rejected at the hook level.
| 🏛️ The Pantheon | 12 Greek gods (skills), each forged for one craft. Call one, or call Zeus to summon all twelve at once |
| ⚡ One-command pipeline | /zeus "..." ships an entire SaaS with zero human interaction (design → build → inspect → test) |
| 🧠 Cross-CLI memory | Persistent 3-layer memory (mnemo) that survives across sessions AND across Claude/Codex/Gemini |
| 🔁 Tireless loop | /chronos autonomously runs FIND → FIX → VERIFY until the bug dies or the dawn breaks |
| 👁️ Hundred-eyed watchman | /argos cross-references spec ↔ code ↔ tests. Nothing slips past 100 eyes |
| ⚖️ Underworld judge | /minos weighs every Playwright test on golden scales. Fix-until-pass loop, no escape |
| 📜 The chronicler | /clio — closer + muse. GO/NO-GO judgment first, then carves PRD, flow diagrams, docs, and doc site onto bronze |
| 🏠 Keeper of the hearth | /hestia scans dead code, unused exports, orphan files — and sweeps them clean |
| 📋 Launch checklist | /launch — pre-launch quality gates, staged rollout plan, rollback playbook |
| 📐 Decision records | /adr — architecture decisions with alternatives, trade-offs, and superseded tracking |
100 skills · 42 agents · 9 hooks · 3 CLIs · 1 mythology
Zeus is the harness layer: one non-interactive request is kept alive, decomposed, implemented, inspected, deployed, tested, and reported without handing "next steps" back to the user. In this repo that means skills/zeus/SKILL.md bootstraps Chronos persistence, drives Zephermine planning, routes implementation through Poseidon (agent-team-codex on Codex), then requires Argos, Docker, Minos, and a final evidence report before SUCCESS is allowed.
Chronos is the loop layer under that harness. It prefers native /goal, falls back to the Codex notify chain (save-turn -> continue-loop -> codex exec --skip-git-repo-check resume --last -), and only stops on Chronos Complete or an exact matching <promise>. Exhaustion is reported as incomplete, blockers are parked with Owner Decision Briefs, and each cycle returns to READ -> FIND -> FIX -> VERIFY -> LOG.
# Clone
git clone https://github.com/Dannykkh/skill-olympus.git
cd skill-olympus
# Windows
.\install.bat
# macOS/Linux
chmod +x install.sh && ./install.shThat's it. 100 skills, 42 agents, 9 hooks installed across Claude Code + Codex CLI + Gemini CLI.
Codex/Gemini steps auto-skip if the respective CLI is not installed.
Beyond the wine-dark sea, where clouds part above the world, Mount Olympus rises. On its windswept summit dwell the Twelve, each crowned with their own dominion, each known by an ancient and many-named song. Speak the name of one, and that one alone shall descend the holy mountain. Speak the name of cloud-gathering Zeus, and the whole pantheon shall come down with him in golden procession.
This is no mere tool-chest. It is a small mythology of work, a council of immortals each shaped to a single craft. They work as the old singers tell us they have always worked — gentle Zephermine, breath of the west wind, whispers a blueprint into the ear of earth-shaking Poseidon; Poseidon stirs the deep and his fleet sails forth in waves; hundred-eyed Argos walks the shore at dusk to count every nail and every beam; stern Minos sits upon his marble throne to weigh each soul at the gate; and last of all, fair-tressed Clio takes up her stylus to carve the whole tale upon a tablet of bronze, that mortals yet unborn may read of it.
Below stand the immortals. Call upon one — or call upon all.
| Skill | Name | Epithet | Domain |
|---|---|---|---|
/zephermine |
Zephermine (젭마인) | Breath of the West Wind, Bringer of Spring | The Architect — 26-step deep interview, spec generation, 5-expert team review |
/zeus |
Zeus (제우스) | Cloud-Gatherer, Hurler of Thunderbolts, Father of Gods and Men | The Sovereign — Zero-interaction full pipeline; at his nod the council convenes |
/agent-team / /poseidon |
Poseidon (포세이돈) | Earth-Shaker, Lord of the Wine-Dark Sea, Trident-Bearer | The Sea Lord — Reads dependency graphs as a sailor reads tides; sends the fleet in waves |
/workpm |
Daedalus (다이달로스) | The Master Builder, Maker of the Labyrinth, Father of Wings | The Hands-On Builder — Where there is no plan, he becomes the plan |
/argos |
Argos (아르고스) | Argos Panoptes, the All-Seeing, of the Hundred Eyes | The Watchman — Of his hundred eyes, none ever sleep at the same hour |
/minos |
Minos (미노스) | Judge of the Dead, Keeper of the Golden Scales | The Judge — At his marble throne, every soul and every test is weighed |
/clio |
Clio (클리오) | Kleio, Proclaimer, Muse of History, Daughter of Memory | The Chronicler — Her stylus sets down what mortals have done, that none may forget |
/chronos |
Chronos (크로노스) | Father of Time, the Unwearying, He Who Devours the Hours | The Tireless — Time itself is his servant; he turns the wheel until the deed is done |
/hermes |
Hermes (헤르메스) | Wing-Footed Messenger, Guide of Souls, Patron of Merchants | The Wayfinder — Reads the trade-winds and the marketplaces of distant lands |
/athena |
Athena (아테나) | Gray-Eyed Daughter of Zeus, Defender of Cities, Born from the Skull | The Strategist — Wisdom that cuts as cleanly as her father's bronze spear |
/aphrodite |
Aphrodite (아프로디테) | Foam-Born, Golden, Cytherean, Lover of Laughter | The Beauty — From her hand come forms that mortals cannot help but love |
mnemo |
Mnemo (므네모) | Mnemosyne, Titaness of Memory, Mother of all the Muses | The Keeper — She forgets nothing; her daughters are born of her remembering |
Hear now the voices of the Twelve, as the old singers heard them.
🜲 Zeus, Cloud-Gatherer Upon the topmost peak he sits, and his nod is the law of mountains. When his voice rolls out across Olympus, the council rises as one and descends — designer, builder, watchman, judge, and chronicler — all at his single word. "Speak my name once, mortal, and the whole council shall walk beside you to the end."
🜂 Zephermine, Bringer of the West Wind She is the soft breath that wakes the seed beneath the soil. Six and twenty are her questions, and the breath of each one is gentle — yet none may pass her by, for the spec is sacred and half-told tales bear no fruit. "I ask, and ask, and ask again — until what was unspoken becomes a thing of stone."
🜄 Poseidon, Earth-Shaker He stands knee-deep in the wine-dark sea, his trident raised, and the waters listen. The fleet of teammates lies in the harbor, and at his bidding the wave gathers them all and bears them out together, each prow pointed where the dependency graph commands. "The sea does not yield to the swimmer. The swimmer who knows the tide — she yields to him."
🜔 Daedalus, Master Builder Before him there was no labyrinth in all of Crete. He took stone from the mountain and shaped it with his own hands, and the work was good. Where the blueprint is missing, where the architect has not spoken, send for him — he will research, he will draft, and if no other hand will rise, his alone shall raise the walls. "Give me the stone. The plan I shall make as I go."
👁 Argos Panoptes, the Hundred-Eyed He paces the half-built city by night, and at no hour are all his eyes closed at once. The plank a mortal builder forgot to nail — he has already seen it. The line of code that fails to match the spec — he has already named it. "While fifty of my eyes rest, fifty more keep watch. Nothing passes Argos in the dark."
⚖ Minos, Judge of the Dead He sits upon a throne of cold marble at the gate where the souls of the dead must come. He raises his golden scales, and the work is weighed against itself. His verdict is two-fold and no other: it shall pass, or it shall return to the fire. "Stand before the scales, child of mortals. We shall see if your tests are honest."
📜 Clio, Muse of the Long Memory She comes last of all the gods, after the labor is laid down. Her stylus is bronze and her tablet is the years to come. What the heroes have done, she sets down — diagram, decree, manual, and song — that the children of the children of those mortals may know the deeds were real. "The work has ended. Now begins the telling, and the telling endures."
⏳ Chronos, the Unwearying He is older than memory, older than the gods themselves. He turns the great wheel of the hours and does not tire when mortals sleep. The bug shall die or the dawn shall come — and Chronos shall outlast them both. "Mortals close their eyes. I do not. The work shall be finished, by sunrise or by the next."
🪶 Hermes, Wing-Footed He moves between the worlds — the high palace and the low marketplace, both are his road. He reads the trade-winds of distant lands and the price of grain in city-gates yet unseen. Before a single coin is risked, before a single line of code is written, he speaks first. "Every market is a road, traveler. Every road has its toll. Bring silver, or bring nothing."
🦉 Athena, Gray-Eyed She was born full-grown from the skull of her father, helmeted, spear in hand. Her wisdom does not flatter; her counsel is the cold edge of the bronze. She will ask the question the mortal fears most — should this thing be made at all? "Wisdom, child, is to know which work must never be begun. I shall ask. You shall answer."
🌹 Aphrodite, Foam-Born She rose from the white foam of the sea, and the world has not been plain since. A hundred and sixty-one palettes lie at her hand, four and eighty fonts, four and eighty styles. What leaves her workshop is not merely useful — it is loved, and that is the difference. "Beauty is not the ornament of the work. Beauty is what makes the work survive its maker."
📚 Mnemo, Mother of the Muses Long before the nine sisters sang, Mnemosyne kept the long memory of the world. The conversation a mortal had three moons ago is the answer she carries to him today. Three layers she keeps — the index of names, the meaning of things, the tale itself — and her remembering crosses every session, every CLI, every dawn. "Forget nothing, child. The word you spoke long ago is the gift you needed now."
v4.21.0 — Aphrodite Experience-Led Redesign · Stitch as Execution Adapter · Zeus Scope Gate (August 2026)
Aphrodite no longer starts from color tokens. The pipeline now runs source-mode routing → site-benchmark dissection (evidence-backed Adopt/Adapt/Avoid verdicts on header, message, section order, CTA, trust, and mobile transformations) → three actually-rendered directions → an Experience Contract capturing user tasks, messaging, CTAs, trust, and responsive behavior → implementation → rendered UX/accessibility/performance gates → a learning handoff, with a validation script enforcing the contract shape. Stitch is repositioned as an execution adapter: it compiles Aphrodite's decisions (DESIGN.md, design-refs) into Stitch MCP operations — work-contract, state-management, and file-transfer patterns selectively adopted from google-labs-code/stitch-skills — and never invents a visual direction of its own. Zeus gained a scope gate: a zero-interaction run has no user to approve scope creep, so any task, feature, or dependency not in the fixed plan must answer three questions (does it serve the one-line goal? what observed evidence demands it now? would something smaller do?) and record a pass/reduce/hold verdict in the Decision Ledger — held items surface as Deferred for post-hoc promotion (principle absorbed from hosioobo/track).
(spans v4.20.1–v4.20.2) The Korean policy path is now anchored to the PIPC's official Privacy Policy Writing Guideline (Apr 2025): the guideline edition is pinned in the skill, and at generation time Themis fetches the official notice board to detect newer revisions — adopting them and telling the user when one exists (environments without web tools fall back to the pinned edition with an explicit "check latest" note). The ko template gained the 2025.4 specifics: a grievance-handling contact separate from the privacy officer, behavioral-advertising collection/refusal clauses, and a children's-data clause (Art. 22-2). Statute citations are verified against law.go.kr by direct fetch; the briefly-added optional law-MCP integration was removed as an unnecessary dependency — the "guidelines keep changing" problem is solved by runtime checks, not by owning scraper infrastructure.
Themis, Titaness of law and order, joins as skill #100. She audits a codebase for every point where personal data is received, stored, sent out, or deleted (5 audit questions + grep hints + a trap checklist covering soft deletes, append-only logs, uninstall scope, masking coverage, silent outbound requests, and git ls-files-verified ignore rules), interviews the operator for the 14 things code cannot tell you (privacy officer, retention, children, server region, marketing use…), then generates per-country policy drafts (Korean PIPA Art. 30 / US CCPA-CPRA / EU GDPR Art. 13) from audit facts only — unverified items stay explicit fill-in blanks, legal judgments are flagged for counsel, and every draft carries a not-legal-advice notice. Dogfooded on this very repo: the three resulting drafts ship as docs/privacy-policy-draft-*. That audit also exposed our own gap — recording hooks with no off switch — so mnemo gained MNEMO_DISABLE (all 15 hook entry points across 4 CLIs exit immediately; existing data untouched) and the session-start version check gained OLYMPUS_UPDATE_CHECK_DISABLE, both verified by on/off comparison runs.
v4.19.0 — Native-First Realignment: 4-CLI Native Multi-Agent · impeccable Detection Hook (August 2026)
The Aphrodite interference experiment (real defects: native-alone 1 ≪ custom 20 ≈ mixed 25) proved the native-first principle — "engines upstream, extensions local" — and this release applies it across the skill roster. Design axis: the frontend-design fork now yields auto-invocation to the official plugin (with the v2.9.1 core grafted in), and prohibition rules moved from prompts to a linter — impeccable's 59 deterministic rules wired as a PostToolUse hook. Shipping that surfaced a real defect: PostToolUse exit-0 plain stdout never reaches the model, so the hook was rewritten to emit hookSpecificOutput.additionalContext JSON (without this fix it would have run silently forever). Multi-agent axis: with all four CLIs now shipping native multi-agent, Poseidon and Daedalus delegate execution to each CLI's native primitives (Claude Agent Teams / Codex spawn_agent / Gemini subagents / Grok spawn_subagent), repositioning the orchestrator MCP as a policy layer for hard file locks and external task boards plus a legacy fallback. Zeus gained an ultracode-session-only Workflow verification fan-out branch, Chronos got its Gemini AfterAgent wiring corrected (plus removal of stale deprecated /loop help text), and Argos now delegates its generic code-quality layer to the native review engines while keeping spec-vs-implementation audit as its own domain. Humanizer re-absorbed both upstreams (blader v2.9.1 + im-not-ai v2.3). The Gemini paths are designed from official docs but not yet field-tested — the verification checklist is on record.
Grok Build (xAI's CLI) joins the roster with near-zero integration cost: grok inspect --json confirmed that Grok reads ~/.claude/ directly via its [compat.claude] defaults — 99 skills, 41 agents, MCP servers, and the global CLAUDE.md rules all load without any sync script. The only real parity gap was mnemo auto-save: Grok's hook envelope is camelCase (no transcript_path), so Claude's Stop hooks silently no-op'd and Grok conversations were never recorded. grok-mnemo closes that gap — one script dispatching on two events (UserPromptSubmit strips Grok's <user_query> wrapper; Stop saves lastAssistantMessage directly, no transcript parsing, filtering out the session-end observe re-fire by reason == "end_turn"), writing to conversations/YYYY-MM-DD-grok.md. Since Grok also loads ~/.claude/settings.json hooks, the five Claude mnemo hook pairs gained a GROK_HOOK_EVENT guard to prevent double/mislabeled saves (Grok prompts were landing in -claude.md). Shipping this surfaced two environment gotchas now on record: Korean-commented .ps1 files must be UTF-8 with BOM (PowerShell 5.1 reads BOM-less files as CP949, and a misdecoded comment can swallow the newline and absorb the next code line — the <user_query> strip silently never ran), and the installer test had been failing on NoDefaultCurrentDirectoryInExePath=1 machines because it invoked install.bat by relative name. Verified end to end with a real grok -p headless session.
A full round-trip: build a real page with Aphrodite, notice it's still generic, fix the pipeline, rebuild, repeat — validated end-to-end on this project's own landing page (dannykkh.github.io/skill-olympus). Layout block anatomy: bag-ui's (MIT) block wireframe catalog translated into a structure grammar — 30 block anatomy contracts (Marketing/App/Ecommerce) with shared rules (three-level ink hierarchy, exactly-one emphasis per block, CTA syntax, sustained-asymmetry ratio reuse across sections) wired in as design-plan Phase 2.5, so structure is locked before style. Prompt consumption gate: DESIGN.md's prose contracts and docs/design-refs/ superprompts were being produced but never mandatorily read at implementation time — Phase 3 now requires reading them first, and direction cards must be saved to file instead of living only in chat. Atmosphere/texture recipe: SKILL.md told implementers to add grain/mesh/noise "atmosphere" with no verified CSS values to reach for, so it evaporated in practice — a new technique-recipes.md §11 supplies a real grain overlay, a restrained mesh gradient, an emphasis-break rule for repeated blocks, and a "boring test." The rebuild also surfaced two ground-truth bugs only visible by actually shipping: a font-loading false negative (document.fonts reports "unloaded" for subsetted Korean webfonts even after the network fetch succeeds — verified by inspecting the actual woff2 requests) and a mobile CSS-selector bug (.pipeline-grid > div deleted the whole timeline, not just the empty gutter) caught by screenshot, not static review.
Vendored Anthropic's official data-visualization skill (from anthropics/knowledge-work-plugins, Apache-2.0, body unmodified) — chart-selection guidance by data relationship (trend/comparison/ranking/distribution/correlation/flow), chart anti-patterns (pie charts over 6 slices, 3D, dual axes), Python code patterns (matplotlib/seaborn/plotly), design principles, and an accessibility checklist.
Aphrodite absorbed the best of MengTo/Skills (MIT) after a full read-through. 12 style recipes: each aesthetic capsuled as "identity boundary + hex tokens + Korean/Latin font stacks + tuning knobs + avoid list" — the original's value-less prose was fixed by binding to our CSV DBs, and every color pair was verified by WCAG computation (3 sub-4.5:1 pairs fixed before shipping). 9 technique recipes: copy-paste-ready values for shadows, progressive blur, border gradients, text reveals, and the GSAP+Lenis motion system. Reference capture (Phase 2 rework): screenshots/URLs/videos/HTML become section-anatomy superprompts (docs/design-refs/) — prompts as versioned assets. Chronos heartbeat: skipping /goal setup used to leave the loop engineless and stall mid-run; the new 1.5-tier engine re-enters via native /loop intervals (--heartbeat), and score-based completion (e.g., "90+ points") is now formalized as a 3-part --completion-promise (threshold, measurement, printed evidence). The pipeline was validated end-to-end with an editorial-tech sample page (Korean fallback rendering, motion, responsive, contrast), surfacing and fixing gotcha 045 (document.fonts.check() false-negatives on unicode-range subset fonts).
Zephermine now plans around what is unknown instead of stopping for broad preference interviews. Unknowns-first discovery: Step 4 auto-selects codebase/web/GitHub/academic/competitor research from the spec and risk profile, writes research-decision.md, then Step 5A writes unknowns.md with known knowns, known unknowns, unknown knowns, unknown unknowns, and architecture-changing questions. Lean interview: Step 6 asks only critical blockers that could change architecture, data model, security boundary, UX flow, rollout, or compliance; otherwise it writes inferred assumptions and continues. No-stop domain flow: domain-dictionary now auto-seeds clear global terms, auto-merges low-risk ADD/REFINE/MERGE updates, and asks only on blocking dictionary conflicts. Implementation learning: workpm and agent-team prompts now keep implementation-notes.md deviations; Clio can generate a CHANGE-QUIZ.md; frontend-design can use divergent static prototypes to surface "you know it when you see it" preferences before implementation.
v4.10.0 — DESIGN.md as Canonical Source · Korean Font Real-Load Guardrails · Hook Timeout & Schema Fixes (July 2026)
Driven by running /aphrodite end-to-end on a real Korean-language UI. DESIGN.md canonical source: Google's @google/design.md format (YAML tokens + two-layer prose) is adopted as Aphrodite's design source of truth, with the agent wiring, single-source, and verification Google omits — a new design-md-guide.md (schema + lint/export + legacy migration), CSV palettes pinned into DESIGN.md to stop re-invocation drift, and a "read DESIGN.md first" rule injected into the 3-CLI always-on guardrails. Korean font real-load guardrail: picking a Latin-only pairing let Korean silently fall back to a system font (Space Grotesk/DM Sans have no Hangul glyphs) — fonts must now be actually loaded (@import/<link> + document.fonts.check), and Korean UIs prefer a Korean-only pairing (Hahmlet/Noto Serif KR + Pretendard/Noto Sans KR) that covers both scripts, structurally avoiding the fallback trap (gotcha 041); the Latin+Pretendard stack remains the fallback path. Dark/light theme rule: the three commonly-missed spots (container div backgrounds, text-color inversion, <select> option color-scheme) are now checked by the mnemo guardrail and ui-ux-auditor. Fixes: hook 30s-timeout prevention (60s timeout + powershell -NoProfile across all four hook-command generators, closing "hook timed out after 30s — output discarded"), and AskUserQuestion schema violations (≤4 options, ≤12-char headers) in the Aphrodite presets and zephermine templates.
Three always-on capabilities, plus a compatibility audit. Design guardrail injection: frontend-design's anti-slop guidance only fired when the skill was explicitly invoked (the auto_apply field is read by no hook — a no-op), so casual design requests defaulted to "internet-average" output. The condensed guardrail is now injected into the 3-CLI always-on context (CLAUDE.md/AGENTS.md), updated to the 2026 web platform — native scroll-driven animation-timeline first (GSAP·Lenis only for pin/snap/WebGL), View Transitions API, container queries, :has(), OKLCH — plus a Korean/Latin font-pairing principle (Pretendard + noonnu, matched by weight/contrast DNA). Condensed ~24% and browser-verified on three distinct directions (dopamine SaaS / editorial / brutalist). Consult-before-implement guard: "implement X" requests now check existing work first (codemap → README/handoffs → grep), classify new/improve/duplicate, and compare against adjacent pipelines (zeus·zephermine·agent-team) — closing the "re-implementing what already exists" loop. Handoff feature map: every handoff records a Feature/Flow/Decision Snapshot + a Menu/Screen Map (screen-level features with done/partial/planned status), with diagrams required only for feature-bearing sessions. agent-team·chronos gain a static boundary-coherence cross-check ("build passes ≠ correct" — TS generics hide API↔hook contract mismatches) and chronos grounds its FIND in tool signals before model intuition. Compatibility audit: Pydantic v2, Next.js 15 async params, MySQL 8.4, docker compose v2, MUI v7, OpenAPI 3.1, Tailwind v4, and stale tool/model labels across the utility skills.
Loop Library 028/034 patterns propagated across the loop stack. A completion contract (each requirement mapped to reproducible evidence, scored proved/weak/missing/contradicted, and exhaustion never reported as success) now governs termination in Chronos, Zeus, minos, autoresearch, argos, agent-team, and workpm. Independent cross-model verification is added where a single actor would otherwise grade its own work — autoresearch re-scores the final champion with a different model family, and Zeus's argos gate runs cross-model only on risk triggers (deterministic build/test/static gate first, the other model in the background) so the implementation path stays fast. At the hook layer, loop-stop and continue-loop now surface EXHAUSTED on max-iterations/stale termination and inject a final-allowed-turn honest-report warning (dry-run verified, ps1 5/5; sh syntax-checked). Hermes gains a grounding guard: TAM/SAM/SOM figures require a cited source or an explicit [확인 필요].
Auto-save hooks used to key off the latest cwd, so a cd into a subfolder (e.g. reference/1week) in a non-git project scattered conversations/·memory/ into that subfolder. Root resolution is now a 2-pass candidate evaluation — Pass 1 takes the first candidate with a git root (a git repo normalizes from any subfolder), Pass 2 falls back to the session-launch cwd for non-git projects, so a persisted cd can't move the anchor. Also guards an environment where HOME itself is a git repo (HOME excluded from candidates; a git root equal to HOME is treated as a dotfiles repo and skipped). Applied across all 8 hooks (save-response/save-conversation/save-tool-use/reconcile-conversations × ps1·sh) with installed copies synced; PS 7/7 · SH 7/7 scenario tests pass.
A loop is sustained by structure, not by the model's will to continue. Five failure modes, five structural cures: heartbeat in the machine, state outside the context window, blockers parked instead of blocking, idle work defined, and the human taken off the critical path with decision-ready briefs.
The loop state lives in the audit log, not the context window; each cycle re-enters through READ, verifies with objective checks, parks valid blockers with decision-ready briefs, and lets hooks reject false completion claims.
- Chronos PARK rule — one blocked issue can no longer stall the whole loop. Four valid park reasons (verification 3-fail / permission boundary / missing external access / product decision); declaring "blocked" without naming a reason counts as evasion; the worker must go as far as it can (reproduce, root-cause, in-permission fixes) before parking
- Escalation ladder before PARK — a verification failure is not parked to the owner until the model raises its own capability once (higher reasoning effort, a stronger model, or a focused review pass when neither is available). Same-approach retries alone cannot trigger a park; escalation runs at most once per issue and its outcome is recorded in the Owner Decision Brief evidence
- Owner Decision Brief — parked issues are reported decision-ready, never as raw questions: what / why now / evidence / trade-offs / recommendation (mandatory — never offload analysis to the owner) / exact choices. The owner's job becomes a 4-way choice: approve as recommended, reject, grant exactly one access, or pick a documented alternative
- Re-entry protocol (READ step) — every cycle re-reads
docs/chronos/chronos-log.mdbefore FIND; when the model's memory and the log disagree, the log wins. A new session resumes the loop from the audit log alone — loop state lives in files, not in the context window - Deadlock guard — goal statements now include the park clause: when only parked issues remain, the loop ends with a Brief-carrying
Chronos Completereport. Outputting an untrue<promise>is forbidden — and the hooks stop rewarding it: the "any<promise>tag = done" branch is removed from loop-stop/continue-loop (a mismatched promise re-injects instead of terminating), multiline promises now match, and re-injection nudges teach the park rule. loop-stop.ps1 functionally tested (4/4 cases: mismatch-rejected / exact-match / parked-only complete / no-marker re-inject); .sh mirrored + syntax-verified - Zeus Decision Ledger —
[ZEUS-AUTO:taste]decisions now record rationale + rejected alternatives + how to reverse; zero-interaction stays intact — approval moves post-hoc instead of pre-hoc, with reversible defaults preferred - codex-mnemo installer fix — notify-wrapper check order corrected: a wrapper that chains save-turn is preserved (refreshed) before the IDE-notification removal heuristic runs. External tools sharing Codex's single notify slot are no longer silently disconnected by reinstall
- ui-ux-auditor visual verification — after the Grep static scan (first-pass signal), it now starts the dev server, captures screenshots (desktop 1440×900 / mobile 390×844 × light/dark), and scores by directly observing the rendered screen. When observation and code inference conflict, observation wins. Falls back to static-only with a
*grade marker when no server can start. Proven end-to-end with a 4-planted-defect smoke test (dark-mode contrast collapse, purple gradient, symmetric 3-col grid, fixed-width overflow) — 4/4 caught by observation - clio v2.1.1 — GO/NO-GO verdict hardening — minos results now enter the verdict (PASS/CONDITIONAL/FAIL), vacuous GO blocked (zero tests caps at CONDITIONAL GO), gate bypass via
--force/--docs-onlymust be flagged on every artifact - Aphrodite scope boundary — Phase 3 implementation limited to "appearance": tokens, markup, styles, visual interactions are hers; state, API wiring, business logic belong to Poseidon/Daedalus (no logic changes in pipeline mode)
Groundwork for loop engineering — aligning the loop's parts (stop gate, review engine, team tools) with CLI-native features.
- code-reviewer v4 — engine delegation + policy layer — generic review is delegated to native engines (Claude
/code-review, Codexcodex review --base); the skill keeps only what natives can't do: Scope Drift detection, domain checklists (LLM output trust boundary, enum completeness), Fix-First triage, unified report. Gemini falls back to the full path (2-Pass + specialists)./code-review ultra(billed) is never invoked nor suggested - Chronos legacy
/loopalias retired — name collision with native/loop(interval re-runner) hijacked the alias on Claude. Removed across all CLIs + goal (stop gate) / loop (re-runner) / chronos (loop discipline) comparison table - Audit follow-ups (5 parallel Explore agents) — zeus × /goal relationship (hook auto-resume stays default for zero-interaction; bootstrap skipped when a goal is pre-set to avoid double stop gates), agent-team experimental env var demoted to legacy, project-root memory vs native auto-memory boundary in 4 files, orchestrator native-vs-MCP selection criteria, chronos
--flow-verifyreceiver definition, zephermine vs native plan-mode distinction - clio v2.1.0 — humanizer Korean copyediting hookup — translationese/AI-style constraints injected at generation + post-generation S1 pass (USER-MANUAL > PRD > TECHNICAL priority)
- Codex native
/reviewdocumented — TUI/review,codex review --base/--uncommitted,codex exec reviewindocs/resources/codex-cli.md
- Korean translationese module (67 patterns, 10 categories) — comma-after-connective (4.84x vs human writing, strongest single signal), ~성/~적/~화 nominalization, progressive overuse, literal pronoun translation. Quantitative first-pass scan + genre guardrails (essay/paper/blog/script/formal)
- Severity tiers + over-editing guard — S1 (always remove) / S2 (clusters only) / S3 (overlaps only) across 24 English + 67 Korean patterns; 30% change-rate warning / 50% hard stop against meaning damage
- Copyediting procedure — do-not masking (proper nouns, numbers, quotes), risk-ordered rewriting, live change-rate tracking + rollback (based on im-not-ai v2.0 taxonomy)
- Reframed as a /goal wrapper — Native
/goal("Set a goal Claude checks before stopping") landed in Claude Code and Codex. Chronos now layers its discipline on top of/goal(the persistence engine) instead of owning the loop: verification gate, priority cycle, audit log - Goal-statement model (no auto-invocation) — Chronos cannot call
/goalprogrammatically (no slash-command tool), so it generates a goal statement with its rules baked in and the user sets/goalonce. The earlier "auto-delegation" framing was an impossible fiction and was removed - 3-tier persistence fallback —
/goal(tier 1) → Stop hook / notify (tier 2) → direct loop (tier 3). Hooks are preserved so Chronos keeps working on Gemini / older builds (parity) - Hard guard against hook collision —
setup-loop --goal-moderemoves any existingloop-state.mdacross.claude/.codex/.chronos, so the Stop hook has nothing to re-inject. Collision is impossible at the code level, not by convention. Verified with isolated.ps1/.shtests - Codex vs Claude /goal differences documented — same entry syntax, different completion judging (Codex runs commands directly; Claude's evaluator only sees chat output). Goal statements now require "print the verification result to chat" + a turn cap for cross-CLI compatibility
- Done-pattern false positive fixed — Stop hook killed loops mid-cycle when the agent narrated progress ("모든 작업 완료, 다음 진행"). Loose narrative regexes (
모든.*작업.*완료,더 이상.*고칠.*없, etc.) removed; only the explicitChronos Completemarker and<promise>tag terminate the loop, matching the documented contract - tail-500 guard removed — The 500-char window prevented stop signals when the marker sat above long explanations. Detection now scans the full assistant output; the guard's original anti-false-positive purpose disappeared once loose patterns were dropped
- Gemini state-path bug fixed —
loop-stop.ps1/.shhard-coded.claude/loop-state.mdand silently passed through Gemini's.chronos/loop-state.md, breaking Chronos on Gemini entirely. Hooks now probe.claude/,.codex/,.chronos/in order - Notification fanout removed — Desktop/IDE notification chaining stripped from Mnemo installers and save-turn hooks. save-turn, Chronos, and hook-bridge flows preserved
- Codex compatibility audit refreshed — Verified
notify → ide-response-notify-wrapper → save-turn → continue-loop → codex exec resume --lastchain end-to-end - Memory distill — gotchas/learned entries refined via memory-distill regular runs
- Stress-tested — 5-iteration counter loop confirmed re-injection mechanism end-to-end on Claude (Stop hook block + reason re-inject)
- mnemo-status notify hook (no LLM cost) — Stop/save-turn hooks now check raw jsonl total and last-handoff age. When
notify_threshold_total(500) ornotify_threshold_handoff_days(14) is exceeded, writesmemory/.mnemo-status.md+ stderr one-liner. Pure text, zero LLM calls - Design ↔ docs alignment — v4.4.0 mentioned a "threshold 50 auto-analyzer" that had no implementation (the auto-analyzer is intentionally absent to avoid silent LLM cost). config.json / SKILL.md now reflect the actual design: distillation runs only via
/memory-distillor handoff - list_handoffs.py fix —
YYYY-MM-DD-{slug}.mdfilenames (no HHMMSS) were showing "Date Unknown"; now parsed correctly - check_staleness.py --all — bulk mode for
docs/handoffs/; previously required one file per invocation - Codex sync EXCLUDE —
gemini-mnemowas leaking into~/.codex/skills/; now correctly excluded
/memory-distillskill (new) — User-triggered distillation of rawobservations.jsonlinto refined.md. Modes:--scan,--apply,--rebuild. The--rebuildmode merges duplicates, resolves contradictions (SUPERSEDED pattern), and archives originals to.archive/— same logic Anthropic Dreaming runs in the cloud- gotcha-analyzer model upgrade —
cleanup-low(Haiku/mini/flash-lite) → main session model inheritance. Claude Opus 4.7 / GPT-5.5 / Gemini 3.1 Pro analysis quality, equivalent to Dreaming'smodel: claude-opus-4-7 - Threshold downgrade 20 → 50 — Auto analyzer becomes safety net; primary distillation moves to handoff and
/memory-distill - Multi-tier triggers — Stop hook (collect) → threshold 50 (safety net) →
/memory-distill(user-driven) → handoff (session boundary)
- Handoff path migration —
.claude/handoffs/→docs/handoffs/for cross-CLI sharing (gitignore was hiding handoffs from teammates) - Auto gotcha/learned extraction — handoff procedure now auto-extracts new jsonl observations into refined
.mdfiles (no review prompt, secret scrubbing) - Memory item hardening — 3 mnemo templates gain explicit guards:
source:single-word only,tags:≥3 keywords, no generic titles, ≤3 lines body - Memory hygiene — 48 missing
source:fields back-filled across 4memory/*.mdfiles; MEMORY.md cleanup (118→54 lines)
- pdf skill — Markdown → PDF generator (playwright + Chromium), Korean defaults (A4 + 25mm + Pretendard)
- Clio integration — Phase 3-4 auto-emits PRD/TECHNICAL/USER-MANUAL.md as PDF alongside Markdown
- Cover/TOC/watermark —
--cover --toc --title --author --org --watermark "초안" --confidential
- domain-dictionary (new skill) — DDD Ubiquitous Language for Korean SI environments. 3-tier storage: master (
docs/domain-dictionary.md) + delta (<planning_dir>/) + global (~/.claude/memory/domain-dictionaries/) - Full pipeline integration — 12 skills now share a single dictionary: zephermine, code-reviewer, argos, poseidon, daedalus, minos, clio, hermes, athena, hestia + 2 codex variants
- zephermine 6-Phase grouping — 26 steps reorganized into Discovery/Spec/Domain/Plan/Design/Validation. Dictionary v1→v2→v3 evolves as Step 8/10/11 byproducts (no extra steps)
- explain --zoom-out — new mode showing callers/siblings/upper map (absorbed from mattpocock/skills)
- code-reviewer module-depth — new category for "shallow vs deep module" refactoring opportunities (absorbed improve-codebase-architecture)
- argos Phase 8 — domain dictionary audit (4 checks: identifier compliance / forbidden terms / Korean UI labels / unregistered new identifiers)
- ceo (Athena) — CEO coaching skill: Go/No-Go gate, strategic challenge, scope decisions (Expand/Reduce/Pivot/Kill)
- Pipeline expansion — New phase:
/hermes→/athena→/zephermine(Analyze → Challenge → Design) - Hermes synergy — Athena auto-reads Hermes output for data-driven strategic challenge
- README overhaul — Star-optimized structure, Meet the Team with Greek myth naming
- project-gotchas — Auto mistake tracking + success pattern learning (analyzer inherits main session model — Opus/Sonnet quality, Dreaming-equivalent)
- 2-layer storage — Global (
memory/gotchas/) + project-specific (memory/learned/) - Cross-CLI observation — Claude save-tool-use + Codex/Gemini save-turn hooks integrated
- CHANGELOG.md — Version history v1.0.0 ~ v1.8.0
- orchestrator — state.json → SQLite WAL migration for crash recovery
- minos — Playwright MCP real-browser QA testing
- codemap — CodeMap index for codebase navigation
- design-plan (Aphrodite) — Design orchestrator with 161 palettes, 73 fonts, 84 styles
- estimate — Development cost estimation with Excel output
- biz-strategy (Hermes) — Business model canvas, TAM/SAM/SOM, GTM strategy
- Anthropic best practices — Applied across all skills
See the full changelog in CHANGELOG.md and Releases.
One command does everything:
/zeus "Build a shopping mall. React + Spring Boot"
→ Design (26-step interview) → Implement (parallel workers) → Inspect → Test
→ Zero interaction — never asks questions, all decisions automated
| Phase | Skill | What it does |
|---|---|---|
| Analyze | /hermes (헤르메스) |
Business model, TAM/SAM/SOM, GTM, metrics, cohort |
| Challenge | /athena (아테나) |
CEO coaching — Go/No-Go gate, scope decisions, kill test |
| Design | /zephermine (젭마인) |
26-step interview → SPEC.md → 5-agent team review |
| Implement | /agent-team |
Wave-grouped parallel execution with Agent Teams |
| Inspect | /argos (아르고스) |
Construction inspection: verify code matches design |
| Test | /minos (미노스) |
Playwright E2E tests + fix-until-pass loop |
| Deliver | /clio (클리오) |
Flow diagrams + PRD + technical docs + user manual |
| Full Auto | /zeus (제우스) |
All phases chained, zero interaction |
Each skill works standalone or as part of the pipeline.
The myth above is the flavor. Here is the plain version: what each god is for (which work it handles), how you call it, and what files you get back. Every god works standalone; chained in order, they form the pipeline.
/hermes — Business case (Wayfinder)
- When: validating whether an idea is worth building — demand, market, money.
- Use:
/hermes(aliases: 헤르메스, 사업성) - Process: demand validation → BMC → TAM/SAM/SOM → revenue & pricing → GTM → North Star → cohort. Generates 2-3 scored business-model candidates before committing to one.
- Output:
docs/hermes/{project}.md(analysis + Mermaid) + a domain-term seed Zephermine reuses. - Next:
/athenato pressure-test, or/zephermineto design.
/athena — CEO challenge (Strategist)
- When: deciding should this be built at all? before committing resources.
- Use:
/athena(aliases: 아테나, ceo) — auto-readsdocs/hermes/*.mdif present. - Process: demand confidence → moat → scope mode → ROI → verdict.
- Output:
docs/athena/{project}.md— GO / CONDITIONAL GO / NO-GO + scope recommendation (Reduce / Expand / Pivot / Kill). - Next:
/zephermineif GO.
/zephermine — Deep design (Architect)
- When: a feature/product needs a thorough spec before implementation.
- Use:
/zephermine [spec-path](aliases: 젭마인, 제퍼마인) - Process: research → 26-step interview → spec synthesis → 5-expert team review → strategy-candidate scoring (ToT) → plan → DB schema / API spec / flow diagrams → section split → operation & QA scenarios.
- Output:
docs/plan/<feature>/→spec.md,plan.md,db-schema.md,api-spec.md,flow-diagrams/,sections/,operation-scenarios.md,qa-scenarios.md. - Next:
/agent-team(build) or/argos(inspect).
/aphrodite — Design system (Beauty)
- When: a UI project needs an intentional experience, benchmark interpretation, and consistent implementation before the frontend build.
- Use:
/aphrodite(aliases: 아프로디테) - Process: source-mode routing → site benchmark evidence (header/message/section order/CTA/trust/mobile transformations) → Adopt/Adapt/Avoid → 3 rendered directions → Experience Contract → implementation → rendered UX/accessibility/performance gates → learning handoff.
- Output:
DESIGN.md(visual tokens), Experience Contract (hierarchy, behavior, responsive and quality decisions), layout blueprint, benchmark evidence, and the frontend build. - Boundary: Aphrodite owns experience structure, visual behavior, responsive transformations, states, and quality gates. API wiring, persistent state, and business logic remain with
/agent-teamor/workpm. - Next:
frontend-designor Stitch as a rendering adapter, then/agent-team//workpmfor application logic.
/agent-team (/poseidon) — Parallel build (Sea Lord)
- When: a spec with
sections/exists and you want parallel implementation. - Use:
/agent-team <planning_dir>(aliases: 포세이돈) - Process: dependency graph → wave grouping → teammates build in waves → mandatory post-merge integration gate (build + full test suite + one E2E).
- Output: implemented code + verification report; the integration gate is the sole completion authority (code-existence checks are pre-checks only).
- Next:
/argos,/minos.
/workpm (/daedalus) — Build without a spec (Master Builder)
- When: no design exists and you want a PM that goes straight to implementation.
- Use:
/workpm(aliases: 다이달로스) - Process: research → 3 proposals scored on fit/risk/effort → flow diagram → implement → verify (tests/lint, bounded retries); activity log externalized for resume.
- Output: working code + decision/activity log.
- Next:
/argos,/minos.
/argos — Construction inspection (Watchman)
- When: after build — confirm the code matches the design artifacts.
- Use:
/argos [planning_dir](aliases: 아르고스, 감리) - Process: static analysis → runtime checks → API-spec match → QA-scenario checklist → flow-diagram-vs-code → security (Phase 0–7).
- Output:
<planning_dir>/verify-report.md. - Next: fix findings, then
/minos.
/minos — E2E test loop (Judge)
- When: you need real browser/E2E tests that actually pass, not "looks done".
- Use:
/minos(aliases: 미노스) - Process: scenarios (from
qa-scenarios.mdor auto-generated) → Playwright code → run → fix-until-pass → browser-exploration QA. - Output:
tests/e2e/{feature}.spec.ts+tests/api/{feature}-api.spec.ts+ pass results. - Next:
/clio.
/clio — Closer + docs (Chronicler)
- When: the work is done and you need a GO/NO-GO call plus deliverable docs.
- Use:
/clio(aliases: 클리오; legacy/closer) - Process: pipeline GO/NO-GO (reads argos/minos + runs build/test) → source-based flow extraction → PRD / technical / manual generation → doc fact-check gate.
- Output:
docs/clio/latest/→CHECKLIST.md,flow-diagrams/,PRD.md,TECHNICAL.md,USER-MANUAL.md,FINAL-REPORT.md. - Next: ship.
/zeus — One command, the whole council (Sovereign)
- When: you want a whole SaaS from one sentence, zero questions.
- Use:
/zeus "Build X. React + Spring Boot"(aliases: 제우스) - Process: chains hermes/athena → zephermine → agent-team → argos → docker → minos → clio; all decisions automated and logged to a reversible Decision Ledger.
- Output: a running app +
docs/zeus/zeus-report.md(SUCCESS is bound to the minos pass-rate + a green build). - Next: review the Decision Ledger.
/chronos — Tireless fix loop (Time)
- When: "fix everything in X until tests pass" — autonomous and resumable.
- Use:
/chronos [scope] --completion-promise '...'(aliases: 크로노스) - Process: FIND → FIX → VERIFY (a real test run, not self-judgment) per cycle, priority-ordered, one issue per cycle; blocked issues are parked with an Owner Decision Brief, with a capability-escalation step before parking.
- Output: fixes +
docs/chronos/chronos-log.mdaudit log (the loop resumes from the log, not memory). - Next: —
mnemo — Cross-CLI memory (Keeper)
- When: always — and whenever you ask "what did we do before?"
- Use:
mnemo(aliases: 므네모); auto-saves every turn via hooks. Opt-out:MNEMO_DISABLE=1(version check:OLYMPUS_UPDATE_CHECK_DISABLE=1). - Process: 3-layer memory that survives across sessions and across Claude/Codex/Gemini; past-conversation search; auto handoff near the context limit.
- Output:
MEMORY.md(index) +memory/*.md(semantic) +conversations/*.md(episodic). - Next: —
Same skills, same memory, same experience across 3 CLIs.
| Feature | Claude Code | Codex CLI | Gemini CLI |
|---|---|---|---|
| Skills | ~/.claude/skills/ |
~/.codex/skills/ |
~/.gemini/skills/ |
| Agents | ~/.claude/agents/ |
~/.codex/agents/ |
~/.gemini/agents/ |
| Memory (Mnemo) | save-response hook | save-turn hook | save-turn hook |
| Gotchas/Learned | save-tool-use hook | save-turn hook | save-turn hook |
| Orchestrator | MCP server | MCP server | MCP server |
| Install | install.bat/sh |
Auto (steps 8-11) | Auto (step 12) |
Cross-CLI sync is handled by sync-codex-assets.js and sync-gemini-assets.js.
3-layer persistent memory that survives across sessions and CLIs.
Session A: work → #tags saved → /wrap-up → MEMORY.md updated
Session B: MEMORY.md auto-loaded → past search → context restored
| Layer | Storage | Loaded |
|---|---|---|
| Index | MEMORY.md |
Always (< 100 lines) |
| Semantic | memory/*.md |
On demand |
| Episodic | conversations/*.md |
On search |
Includes auto gotcha/learned tracking:
- Errors →
memory/gotchas/observations.jsonl→ Haiku analyzes patterns - Successes →
memory/learned/observations.jsonl→ Haiku detects workflows - Backlog diagnosis → observation logs are append-only and never truncated; backlog is judged by delta against the
.mnemo-distill-offsetmarker (hooks handle this via.mnemo-status.md), never by cumulative line count
| Category | Skills | Highlights |
|---|---|---|
| AI Tools | codex, gemini, orchestrator, workpm, agent-team + 5 more | Multi-AI orchestration, PM-Worker pattern |
| Pipeline | zephermine, zeus, argos, minos, closer, shipping-and-launch | Zero-interaction full dev pipeline, launch checklist |
| Frontend | react-dev, frontend-design, theme-factory, stitch, seo-audit, ui-ux-auditor, data-visualization + 5 more | 161 palettes, 84 fonts, 14 themes (4 Hangul), SEO+AEO+GEO audit, chart selection guide |
| Development | docker-deploy, database-schema-designer, deprecation-and-migration, documentation-and-adrs, social-login, code-reviewer + 7 more | Docker, DB design, ADR, migration, social login, code quality |
| Business | biz-strategy, ceo, estimate, okr, daily-meeting-update | CEO coaching, cost estimation, OKR, standup |
| Testing | minos, auto-continue-loop, flow-verifier, themis + 3 more | Chronos loop, Playwright QA, privacy-policy generator (Themis) |
| Memory | mnemo, memory-compact, project-gotchas, memory-distill | 3-layer memory, auto learning, raw distillation (rebuild) |
| Docs | mermaid-diagrams, marp-slide, docx, pdf, draw-io, domain-dictionary + 3 more | Diagrams, presentations, documents, domain dictionary (DDD UL) |
| Meta | autoresearch, skill-judge, manage-skills, plugin-forge, release-notes + 4 more | Skill auto-optimization (Hill Climbing), management, release |
| Git | commit-work, release-notes, deploymonitor | Conventional commits, CHANGELOG |
| Media | video-maker | Remotion-based React video |
| Research | reddit-researcher | Market research + lead scoring |
| Translation | ko-en-translator | Korean↔English bidirectional translation |
| Utilities | humanizer, jira, datadog-cli, excel2md + 3 more | AI pattern removal, integrations |
Specialized subagents for every development task:
| Area | Agents |
|---|---|
| Architecture | architect, spec-interviewer, fullstack-coding-standards |
| Frontend | frontend-react, react-best-practices, stitch-developer, ui-ux-designer |
| Backend | backend-spring, backend-dotnet, desktop-wpf, python-fastapi |
| Database | database-postgresql, database-mysql, database-schema-designer |
| Quality | code-reviewer, security-reviewer, qa-engineer, tdd-coach |
| Performance | performance-engineer, debugger |
| AI/ML | ai-ml (RAG, LLM APIs, latest SDKs) |
| Writing | writing-specialist, humanizer, writing-guidelines |
| Language | typescript-spec, python-spec |
| Hook | Event | Purpose |
|---|---|---|
| reconcile-conversations | SessionStart | Backfill missed Claude/Codex turns from JSONL transcripts |
| save-response | Stop | Auto-save assistant responses with #tags |
| save-tool-use | PostToolUse | Tool logging + gotchas/learned observation |
| save-conversation | UserPromptSubmit | Persist user input |
| check-new-file | PreToolUse | Reducing entropy check |
| protect-files | PreToolUse | Sensitive file protection |
| validate-api | PostToolUse | API file validation |
| loop-stop | Stop | Chronos auto-iteration |
| orchestrator-detector | UserPromptSubmit | PM/Worker mode detection |
PM distributes tasks, Workers execute in parallel across Claude + Codex + Gemini.
Terminal 1 (PM): /workpm → analyze → create 3 tasks
Terminal 2 (Claude): /pmworker → claim task-1 → execute → complete
Terminal 3 (Codex): /pmworker → claim task-2 → execute → complete
Terminal 4 (Gemini): /pmworker → claim task-3 → execute → complete
| Component | Description |
|---|---|
| Orchestrator MCP | SQLite WAL task queue, file locks, dependency resolution |
| workpm | Unified PM entrypoint (native multi-agent per CLI; MCP fallback) |
| pmworker | Unified Worker entrypoint (all CLIs) |
| Resource | Description | Install |
|---|---|---|
| everything-claude-code | Anthropic hackathon winner (28 agents, 116 skills) | /plugin marketplace add |
| Vercel Agent Skills | React/Next.js best practices (45+ rules) | npx add-skill vercel-labs/agent-skills |
| claude-code-dotnet | C#/WPF/MAUI/.NET skills | npx add-skill Aaronontheweb/claude-code-dotnet |
| MCP | Description | Install |
|---|---|---|
| Context7 | Latest library docs (Next.js 15, React 19) | claude mcp add context7 -- npx -y @upstash/context7-mcp |
| Playwright | Browser automation for QA | claude mcp add playwright -- npx -y @playwright/mcp@latest |
| Stitch | Google Stitch UI design | npx -p stitch-mcp-auto stitch-mcp-auto-setup |
| Resource | Description |
|---|---|
| skills.sh | 25K+ skills directory by Vercel |
| awesome-agent-skills | 200+ curated skills |
| awesome-claude-code | Claude Code resource curation |
| Version | Date | Highlights |
|---|---|---|
| v4.21.0 | 2026-08-10 | Aphrodite experience-led redesign + Stitch execution adapter + Zeus scope gate — design-plan pipeline: source routing → benchmark dissection (Adopt/Adapt/Avoid) → three rendered directions → Experience Contract (tasks, message, CTA, trust, mobile transformations; validator-enforced) → implementation → rendered UX/a11y/perf gates → learning handoff; Stitch compiles Aphrodite's decisions into Stitch MCP operations (contract/state/transfer patterns from google-labs-code/stitch-skills) instead of inventing direction; Zeus records pass/reduce/hold verdicts in the Decision Ledger before any off-plan task/feature/dependency, holds surfacing as Deferred (absorbed from hosioobo/track) |
| v4.20.2 | 2026-08-08 | Themis guideline pinning + runtime freshness check (spans v4.20.1–v4.20.2) — Korean drafts anchored to the PIPC Writing Guideline (2025.4) with a generation-time notice-board check that detects and adopts newer revisions (user notified; pinned-edition fallback without web tools); ko template aligned to 2025.4 (grievance contact, behavioral-ads clause, children Art. 22-2); statute citations verified via law.go.kr direct fetch — law-MCP dependency removed as unnecessary |
| v4.20.0 | 2026-08-08 | Themis privacy-policy generator + mnemo opt-out — skill #100 Themis audits a codebase for personal-data collection/storage/outbound/deletion (trap checklist: soft delete, append-only logs, masking coverage, silent outbound, git ls-files-verified ignores), interviews the operator for the 14 things code can't tell (officer, retention, children, server region…), then drafts per-country policies (PIPA Art. 30 / CCPA-CPRA / GDPR Art. 13) from file:line evidence only — blanks are never invented; dogfooded on this repo (3 drafts shipped). The audit exposed our own no-off-switch recording hooks → MNEMO_DISABLE (15 hook entry points, 4 CLIs) + OLYMPUS_UPDATE_CHECK_DISABLE, verified by on/off comparison runs |
| v4.19.0 | 2026-08-05 | Native-first realignment — the Aphrodite interference experiment (real defects: native-alone 1 ≪ custom 20 ≈ mixed 25) proved "engines upstream, extensions local," now applied roster-wide: frontend-design fork yields auto-invocation to the official plugin + impeccable's 59 deterministic rules as a PostToolUse hook (fixing a real defect: PostToolUse exit-0 stdout never reaches the model — rewritten as additionalContext JSON); Poseidon/Daedalus delegate to each CLI's native multi-agent (Claude Agent Teams / Codex spawn_agent / Gemini subagents / Grok spawn_subagent) with orchestrator MCP repositioned as hard-lock policy layer + legacy fallback; Zeus gains an ultracode-only Workflow fan-out branch; Chronos Gemini AfterAgent wiring corrected + deprecated /loop help remnants removed; Argos delegates its generic quality layer to native review engines (spec audit stays its own); humanizer re-synced to both upstreams (blader v2.9.1 + im-not-ai v2.3); Gemini paths designed from official docs, field test pending |
| v4.15.0 | 2026-07-25 | Grok Build support: grok-mnemo adapter — measured with grok inspect --json that Grok reads ~/.claude/ skills/agents/MCP/rules directly via [compat.claude] defaults (no sync script needed); the one parity gap, mnemo auto-save, closed by a UserPromptSubmit+Stop adapter (<user_query> wrapper strip, end_turn-only saves, lastAssistantMessage direct save, conversations/*-grok.md); GROK_HOOK_EVENT guard on 5 Claude mnemo hook pairs stops double/mislabeled saves; installer-test fix for NoDefaultCurrentDirectoryInExePath=1 machines (absolute path + windowsVerbatimArguments); new gotcha: Korean-commented .ps1 needs UTF-8 BOM or PS 5.1's CP949 misdecode silently merges code lines into comments; verified with a real grok -p headless session |
| v4.14.0 | 2026-07-22 | Layout block anatomy · atmosphere recipe · prompt consumption gate — bag-ui's (MIT) block wireframe catalog translated into 30 structure contracts wired as design-plan Phase 2.5; Phase 3 now requires reading DESIGN.md prose contracts and docs/design-refs/ superprompts before implementing (previously produced but never mandatorily consumed); technique-recipes.md §11 supplies verified grain/mesh/emphasis-break/sustained-variance values where SKILL.md previously only instructed "add atmosphere" with nothing to implement against; all three gaps found and fixed by shipping this project's own landing page end to end (including a Korean-webfont document.fonts false-negative and a mobile CSS-selector bug caught by screenshot) |
| v4.13.0 | 2026-07-10 | Data-visualization skill (vendored from Anthropic) — chart selection by data relationship (trend/comparison/ranking/distribution/correlation/flow), chart anti-patterns (pie >6 categories, 3D never, dual-axis caution), Python (matplotlib/seaborn/plotly) code patterns, design principles, accessibility checklist; Apache-2.0 attribution kept, invocable as /data-visualization, role split vs design-plan/mermaid-diagrams documented; 96 -> 97 skills across 3 CLIs |
| v4.12.2 | 2026-07-10 | Non-git project-root guard — when git yields no root, hooks walk ancestors for an existing mnemo marker (MEMORY.md/conversations/, HOME excluded) then cut before build-output segments (bin/obj/dist/...); stops conversations/·memory/ scattering into bin\Debug or app-data folders; 12 files (4 Claude hooks + codex/gemini save-turn, ps1+sh) |
| v4.12.1 | 2026-07-10 | Hook stdin watchdog fail-open — fixes "UserPromptSubmit hook timed out after 60s" (stdin starvation on mid-turn prompt submission): 15s bounded stdin read + quiet exit, PS 5.1 [Console]::In sync-blocking workaround (StreamReader over OpenStandardInput), exit-on-timeout rule (orphaned read + native spawn = deadlock, reproduced), chronos continue-loop payload order argv-first, HOME git-root guard on codex/gemini save-turn |
| v4.12.0 | 2026-07-09 | Style Recipes · Reference Capture · Chronos Heartbeat — 12 style recipes (hex + Korean/Latin font stacks, WCAG-computed) + 9 technique recipes (shadows/blur/border-gradient/reveals/GSAP+Lenis) + Phase 2 reference capture (section-anatomy superprompts); Chronos 1.5-tier heartbeat engine (--heartbeat, native /loop interval re-entry) + score-based completion contract; validated end-to-end with an editorial-tech sample + gotcha 045 (fonts.check subset false-negative) |
| v4.11.0 | 2026-07-05 | Unknowns-first planning · lean Zephermine flow · implementation learning loop — Zephermine auto-selects research and records research-decision.md, adds unknowns.md before interviews, asks only architecture-changing blockers, and writes inferred assumptions otherwise; domain-dictionary now auto-seeds/auto-merges low-risk terms and asks only on blocking conflicts; workpm/agent-team keep implementation-notes.md deviations, Clio can generate CHANGE-QUIZ.md, and frontend-design can prototype divergent directions to expose unknown knowns before implementation |
| v4.10.0 | 2026-07-01 | DESIGN.md canonical source · Korean font real-load guardrails · hook timeout & schema fixes — Google @google/design.md format adopted as Aphrodite's design source of truth (new design-md-guide.md, palettes pinned to stop drift, "read DESIGN.md first" in 3-CLI always-on); Korean font real-load (@import + document.fonts.check, Korean-only pairing preferred to avoid silent system fallback, gotcha 041); dark/light rule (container div bg · text inversion · <select> color-scheme); hook 30s-timeout prevention (60s + -NoProfile across 4 generators); AskUserQuestion schema fixes (≤4 options, ≤12-char headers) |
| v4.9.0 | 2026-06-25 | Always-on design guardrail · consult-before-implement · handoff feature map — anti-slop design guidance injected into the 3-CLI always-on context (frontend-design auto_apply was a no-op), updated to the 2026 web platform (native scroll-driven animation-timeline first, View Transitions, container queries, :has(), OKLCH) + Korean/Latin font pairing, condensed ~24%, browser-verified on 3 directions; consult-before-implement guard (codemap-first → handoffs → grep → classify, compare adjacent pipelines); handoff Feature/Flow/Decision Snapshot + Menu/Screen Map; harness-engineering boundary-coherence cross-check on agent-team·chronos (build-pass≠correct) + loop-engineering tool-grounded FIND on chronos; compat audit (Pydantic v2, Next.js 15, MySQL 8.4, docker compose v2, MUI v7, OpenAPI 3.1, Tailwind v4) |
| v4.8.7 | 2026-06-23 | Loop honesty (028/034) — completion contract (requirement→evidence, 4-state scoring, exhaustion≠success) across Chronos·Zeus·minos·autoresearch·argos·agent-team·workpm; independent cross-model verify (autoresearch champion re-score; Zeus argos risk-gated cross-model — deterministic gate first, other model in background); hook-layer EXHAUSTED surfacing + final-turn honest-report warning (loop-stop·continue-loop ps1/sh, dry-run ps1 5/5); Hermes TAM/SAM/SOM source-citation grounding guard |
| v4.8.1 | 2026-06-14 | mnemo root-resolution fix — auto-save hooks no longer misplace conversations/·memory/ into a subfolder when you cd deeper in a non-git project; 2-pass project-root (git root if any candidate resolves, else session-launch cwd) + HOME-as-git-repo guard; applied across 8 hooks (ps1·sh) + installed copies, PS 7/7 · SH 7/7 |
| v4.8.0 | 2026-06-13 | Loop programming: park, brief, re-entry — Chronos PARK rule (4 reasons; reasonless "blocked" = evasion) + Owner Decision Brief (decision-ready escalation, mandatory recommendation, 4-way owner choice) + re-entry protocol (log beats memory; new-session resume from audit log); deadlock guard for parked-only queues (no untrue <promise>; hooks stop rewarding mismatched tags — ps1 tested 4/4, multiline match added, park rule in re-injection nudges); Zeus Decision Ledger (rationale + alternatives + how-to-reverse, post-hoc approval); codex-mnemo notify order fix (save-turn-chaining wrappers preserved across reinstalls) |
| v4.7.1 | 2026-06-11 | Design visual verification + Clio gate hardening — ui-ux-auditor scores by observing rendered screenshots (observation beats grep; 4/4 planted-defect smoke test), clio v2.1.1 verdict hardening (minos in verdict, vacuous GO blocked, bypass flagged), Aphrodite scope limited to appearance |
| v4.7.0 | 2026-06-11 | Native harness integration — code-reviewer v4 (engine delegation: Claude /code-review · Codex codex review --base, policy layer P1~P5 with Scope Drift/Fix-First/domain checklists, Gemini full-path fallback); Chronos legacy /loop alias retired (native /loop name collision) + goal/loop/chronos comparison table; audit follow-ups (zeus /goal relationship + double stop-gate guard, agent-team env var demoted to legacy, project-root memory vs native auto-memory boundary in 4 files, orchestrator native-vs-MCP selection criteria, chronos --flow-verify receiver definition); clio v2.1.0 humanizer Korean copyediting hookup; zephermine vs native plan-mode distinction |
| v4.6.0 | 2026-06-10 | Humanizer Korean writing module — 67 translationese patterns in 10 categories (A~J), quantitative first-pass scan (comma-after-connective 4.84x signal), genre guardrails, S1/S2/S3 severity tiers, procedural over-editing guard (do-not masking, change-rate rollback); absorbed im-not-ai v2.0 taxonomy; deployed to 3 CLIs |
| v4.5.0 | 2026-06-05 | Chronos × native /goal integration — reframed as a /goal wrapper (goal=persistence, Chronos=verification gate/priority/log); goal-statement model (no auto-invocation — generates the statement, user sets /goal once); 3-tier fallback (goal → hook/notify → direct) for Gemini parity; hard guard setup-loop --goal-mode clears loop-state to make hook collision impossible at the code level (tested .ps1/.sh); Codex vs Claude judging differences documented |
| v4.4.2 | 2026-05-30 | Chronos hardening + cross-CLI parity — done-pattern false positive removed (Chronos Complete + <promise> only); tail-500 guard dropped (full-output marker scan); Gemini state-path bug fixed (3-path probe .claude/.codex/.chronos); notification fanout stripped from Mnemo installers/save-turn; stress-tested 5-iteration re-injection on Claude |
| v4.4.1 | 2026-05-11 | mnemo audit patches — Stop/save-turn notify_mnemo_status hook (zero-LLM-cost user awareness via memory/.mnemo-status.md + stderr); SKILL.md/config.json aligned with actual design (no auto-analyzer); list_handoffs.py parses YYYY-MM-DD-{slug} filenames; check_staleness.py --all bulk mode; Codex EXCLUDE adds gemini-mnemo |
| v4.4.0 | 2026-05-08 | /memory-distill + Dreaming-equivalent self-improvement — new user-triggered skill with --scan/--apply/--rebuild modes (duplicate merge, SUPERSEDED conflict resolution, archive backup); gotcha-analyzer upgraded from cleanup-low to main session model inheritance (Opus/GPT-5.5/3.1-Pro analysis quality); threshold 20→50 (safety-net role); multi-tier distillation triggers |
| v4.3.0 | 2026-05-05 | Mnemo Memory Integrity Pass — handoffs to docs/handoffs/ (cross-CLI shareable); auto gotcha/learned extraction in handoff (no review, secret scrubbing); item format hardening (source/tags/title/length); 48 source backfills; 3-CLI parity verified |
| v4.2.0 | 2026-05-04 | Markdown → publication-quality PDF — pdf skill adds generator (playwright + Pretendard), Korean defaults (A4 + 25mm), automatic Clio Phase 3-4 integration |
| v4.1.0 | 2026-04-28 | Domain Dictionary Pipeline — new domain-dictionary skill (DDD Ubiquitous Language) integrated across 12 skills with 3-tier storage (master/delta/global); zephermine 6-Phase grouping; explain --zoom-out; code-reviewer module-depth |
| v4.0.2 | 2026-04-27 | Skill description diet — shorter routing metadata for safer Codex/Claude prompt budgets; docs and installed copies synced |
| v4.0.1 | 2026-04-20 | Hook install cleanup, stale references removed, structure principles replace old line-count guidance |
| v4.0.0 | 2026-04-20 | The Great Cleanup: skill/agent consolidation, obsolete hooks archived, Skill Olympus quality pass |
| v3.0.0 | 2026-04-08 | 🏛️ Skill Olympus — The Pantheon Awakens (repo rename, 그리스 신화 통일, mnemo 데이터 유실 방지 종합 개편, 3-CLI parity, README 호메로스 톤) |
| v2.1.0 | 2026-04-06 | Pipeline Integrity Audit + gstack-Inspired Improvements (Zeus 7-Phase, hermes/athena 강화, AI Slop 탐지) |
| v2.0.0 | 2026-03-25 | Athena CEO Coaching + Pipeline Expansion |
| v1.9.0 | 2026-03-24 | Athena CEO Coaching + Pipeline Expansion |
| v1.8.0 | 2026-03-23 | Project Gotchas + Learned Patterns |
| v1.7.0 | 2026-03-21 | Orchestrator SQLite WAL + Minos |
| v1.6.0 | 2026-03-18 | Design + Business + Skill Best Practices |
| v1.5.0 | 2026-03-09 | Closer + SEO Audit + Pipeline Overhaul |
| v1.4.0 | 2026-03-02 | Chronos + Argos + Memory Compact |
| v1.3.0 | 2026-02-19 | Cross-CLI: Codex + Gemini |
| v1.2.0 | 2026-02-09 | Agent-Team + Zeus + QA Pipeline |
| v1.1.0 | 2026-02-01 | Zephermine + Mnemo + Install |
| v1.0.0 | 2026-01-29 | Initial Release |
MIT License
Last Updated: 2026-06-13