{
  "markdown": "<p align=\"center\">\n  <img src=\"assets/logo.png\" alt=\"CC10x — The Loop Engine for Claude Code. Orange stylized X with starburst and CC10x wordmark.\" width=\"280\" />\n</p>\n\n<h1 align=\"center\">cc10x — The Loop Engine for Claude Code</h1>\n\n<p align=\"center\">\n  <em>Stop chasing better models. Engineer the loop.</em>\n</p>\n\n<p align=\"center\">\n  Same model. Same Claude Code. Different outcome.<br />\n  The agent forgets between turns. <strong>The harness remembers across runs.</strong><br />\n  cc10x writes every workflow to disk — intent, evidence, verdicts, failures — so resume, review, and verification read from the artifact, not from a context window that's already gone.\n</p>\n\n<p align=\"center\">\n  <strong>1 router</strong> &nbsp;·&nbsp; <strong>11 specialist agents</strong> &nbsp;·&nbsp; <strong>20 skills</strong> &nbsp;·&nbsp; <strong>4 workflows</strong>\n</p>\n\n<p align=\"center\">\n  Fail-closed gates &nbsp;·&nbsp; survives compaction &nbsp;·&nbsp; dispatch-by-reference &nbsp;·&nbsp; test honesty gates &nbsp;·&nbsp; anti-anchored review\n</p>\n\n<p align=\"center\">\n  <strong>Explore interactively:</strong> <a href=\"cc10x-explorer.html\">cc10x Explorer</a> &nbsp;·&nbsp; <a href=\"cc10x-architecture-explorer.html\">Architecture Explorer</a>\n</p>\n\n**Current version:** 12.8.0\n\n---\n\n## Install\n\n**Step 1 — Add the marketplace:**\n\n```bash\n/plugin marketplace add romiluz13/cc10x\n```\n\n**Step 2 — Install the plugin:**\n\n```bash\n/plugin install cc10x@cc10x\n```\n\nThen say **\"set up cc10x for me\"** in Claude Code and restart. Done.\n\n---\n\n## Quick Start Examples\n\n### Build Something\n\n```\n\"build a user authentication system\"\n\n→ Router detects BUILD intent\n→ Stops to resolve missing requirements first\n→ component-builder drives RED → GREEN → REFACTOR\n→ code-reviewer + failure-hunter run in parallel\n→ integration-verifier checks wiring, artifacts, and behavior\n→ Workflow state and memory are updated\n```\n\n### Fix a Bug\n\n```\n\"debug the payment processing error\"\n\n→ Router detects DEBUG intent\n→ Loads prior failures and project memory\n→ bug-investigator starts from logs and observed behavior\n→ code-reviewer checks the fix path\n→ integration-verifier confirms the bug is actually closed\n→ Useful findings go back into memory\n```\n\n### Review Code\n\n```\n\"review this PR for security issues\"\n\n→ Router detects REVIEW intent\n→ code-reviewer uses repo and git context\n→ Reports findings only when confidence clears the bar\n→ Every finding includes file:line evidence\n```\n\n---\n\n## Why cc10x\n\nAsk Claude for something complex. It works for a while. Then it declares **\"Done!\"** — tests still red, refactor half-finished, and by message 40 it's contradicting itself because the context is gone.\n\n**cc10x fixes the loop, not the prompt.** A better model running free loses to the same model, constrained and looped correctly. That's the whole bet.\n\n| The pain you know | How cc10x handles it |\n| --- | --- |\n| \"Done!\" on red tests | `integration-verifier` is independent of the builder. Phase-exit gates block advancement on partial evidence. |\n| Silent failures nobody asked about | `failure-hunter` runs in parallel with review — greps for swallowed errors and empty catches. |\n| Context falls apart after compaction | Workflow state on disk with stable UUIDs. Memory files the router auto-heals. |\n| Planning is just a chat | Three planning modes chosen by intent, with a fresh anti-anchored review by a reviewer who never saw the planner's rationale. |\n| 12 slash commands to remember | One router. Every request hits `cc10x-router` first. |\n| `.claude/` prompt spam on every fanout | State lives at `.cc10x/` — outside `.claude/`, so the harness's sensitive-file gate never fires. |\n| Green tests that prove nothing | Test Honesty Gates grep for `getByTestId('…-mock')`, `as any`, `.find()` bypass, `setTimeout()` waits. A hit can't count as PASS on that test's strength alone. |\n| Reviewer findings that sound right but aren't | Every finding at confidence ≥80 needs a verbatim `file:line` quote or it's auto-demoted. The verifier independently re-reads the line and drops hallucinated findings before they can gate. |\n| Orchestrator context rotting with pasted history | Dispatch by reference, not by blob — the diff is written to disk, the prompt passes a path, never a body. (One real dispatch hit 42k chars, 99% pasted history. The scar became law.) |\n\n---\n\n## How It Works\n\n**You describe the work. cc10x routes it, brings in the right specialists, and keeps the bar for \"done\" higher than a convincing paragraph.**\n\n```\n                        YOU\n                         │\n                         ▼\n            ┌────────────────────────┐\n            │      cc10x-router      │  ◄── only entry point\n            │   detects intent       │\n            └────────────┬───────────┘\n                         │\n          ┌──────────────┼──────────────┬─────────────┐\n          │              │              │             │\n          ▼              ▼              ▼             ▼\n        BUILD          DEBUG         REVIEW         PLAN\n          │              │              │             │\n          ▼              ▼              ▼             ▼\n    component-      bug-          code-          planner\n      builder    investigator   reviewer            │\n          │              │              │             ▼\n          ▼              ▼           (done)     plan-gap-\n  [code-reviewer    code-reviewer             reviewer\n   ∥ silent-            │\n   failure-            ▼\n    hunter]      integration-\n          │         verifier\n          ▼\n   integration-\n     verifier\n\n                 ┌──────────────────────────┐\n                 │  STATE (every workflow)  │\n                 │  activeContext.md        │\n                 │  patterns.md            │\n                 │  progress.md            │\n                 │  {wf}.json + .events    │\n                 └──────────────────────────┘\n```\n\n---\n\n## Claude Setup Instructions\n\nThis section is written for Claude Code to follow when the user says \"set up cc10x for me\".\n\n**When user says \"set up cc10x\", \"install cc10x\", or similar:**\n\n### Step 1: Check Existing Files\n\n- Read `~/.claude/CLAUDE.md` (may not exist)\n- Read `~/.claude/settings.json` (may not exist)\n\n### Step 2: Update `~/.claude/CLAUDE.md`\n\n**If file doesn't exist:** CREATE with the template below.\n**If file exists:** PREPEND the cc10x section below, keep user's existing content.\n\n> **Multi-project note:** The global `~/.claude/CLAUDE.md` activates cc10x in **every** project automatically — you do not need to reinstall or reconfigure per project. Only add the cc10x section to a project's `.claude/CLAUDE.md` if that project has its own conflicting CLAUDE.md already.\n\n```markdown\n# CC10x Orchestration (Always On)\n\nIMPORTANT: ALWAYS invoke cc10x-router on ANY development task. First action, no exceptions.\nIMPORTANT: Do only minimal orientation if needed, then invoke the router immediately.\nIMPORTANT: Prefer retrieval-led reasoning over pre-training-led reasoning for orchestration decisions.\nIMPORTANT: Never bypass the router. It is the system.\nIMPORTANT: NEVER use Edit, Write, or Bash (for code changes) without first invoking cc10x-router.\n\n**Skip CC10x ONLY when:**\n- User EXPLICITLY says \"don't use cc10x\", \"without cc10x\", or \"skip cc10x\"\n- No interpretation. No guessing. Only these exact opt-out phrases.\n\n[CC10x]|entry: cc10x:cc10x-router\n\n---\n\n## Complementary Skills (Work Together with CC10x)\n\n**Skills are additive, not exclusive.** CC10x provides orchestration. Domain skills provide expertise. Both work together.\n\n**GATE:** Before writing code, check if task matches a skill below. If match, invoke it via `Skill(skill=\"...\")`.\n\n| When task involves... | Invoke |\n|-----------------------|--------|\n| *(Add user's installed skills here)* | |\n```\n\n### Step 3: Update `~/.claude/settings.json`\n\n**If file doesn't exist:** CREATE with the template below.\n**If file exists:** MERGE these permissions into the existing `permissions.allow` array (don't overwrite!):\n\n```json\n\"Bash(mkdir -p .cc10x)\",\n\"Bash(mkdir -p docs/plans)\",\n\"Bash(mkdir -p docs/research)\",\n\"Bash(mkdir -p docs/solutions)\",\n\"Bash(git status)\",\n\"Bash(git diff:*)\",\n\"Bash(git log:*)\",\n\"Bash(git branch:*)\",\n\"Bash(git blame:*)\",\n\"Bash(git ls-files:*)\",\n\"Bash(git rev-parse:*)\",\n\"Bash(python3:*)\",\n\"Edit(.cc10x/*)\",\n\"Write(.cc10x/*)\"\n```\n\n> **Why the Edit/Write permissions?** The live cc10x memory namespace is `.cc10x/`. The permission examples use the `.cc10x/*` scope so the namespace works without re-prompting on every memory write.\n\n> **Why `git rev-parse` and `python3`?** Every BUILD phase records its base SHA (`git rev-parse HEAD`), and the router produces review diff packages and phase briefs by running plugin tools via `python3`. Without these two, every phase prompts mid-workflow.\n\n### Step 4: Set User Standards (Optional)\n\nAsk the user:\n> \"Do you have coding standards or principles you want cc10x agents to always follow? (e.g. 'always use TypeScript strict mode', 'follow SOLID principles', 'never use `any`', 'prefer functional patterns')\"\n\n**If user provides standards**, write them to the project's memory:\n\n```\nBash(command=\"mkdir -p .cc10x\")\n# Check if patterns.md already exists (Read returns error = doesn't exist)\nRead(file_path=\".cc10x/patterns.md\")\n\n# If it DOESN'T exist — create with standards already populated:\nWrite(file_path=\".cc10x/patterns.md\", content=\"# Project Patterns\\n<!-- CC10X MEMORY CONTRACT: Do not rename headings. Used as Edit anchors. -->\\n\\n## User Standards\\n- {standard 1}\\n- {standard 2}\\n\\n## Common Gotchas\\n\\n## Project SKILL_HINTS\\n\\n## Last Updated\\n{date}\")\n\n# If it DOES exist — append under User Standards:\nEdit(file_path=\".cc10x/patterns.md\",\n     old_string=\"## User Standards\",\n     new_string=\"## User Standards\\n- {standard 1}\\n- {standard 2}\")\n\nRead(file_path=\".cc10x/patterns.md\")  # Verify\n```\n\n**If user skips:** No action. The memory file will be created on first workflow run with an empty `## User Standards` section for them to fill in later.\n\n### Step 5: Scan Installed Skills & Add to Table\n\n**Where to find installed skills:**\n\n1. `~/.claude/settings.json` → check `enabledPlugins` object (plugins with value `true`)\n2. `~/.claude/plugins/installed_plugins.json` → detailed plugin info\n3. `~/.claude/skills/` → personal skills (all projects)\n4. `.claude/skills/` → project-specific skills\n\n**Skill naming in table:**\n\n- **Plugin skills:** `plugin-name:skill-name` (e.g., `mongodb-agent-skills:mongodb-schema-design`)\n- **Personal/project skills:** just the skill name (e.g., `react-best-practices`)\n\n**Example:** If user has these:\n\n```\n# In enabledPlugins:\n\"mongodb-agent-skills@mongodb-agent-skills\": true\n\n# In ~/.claude/skills/:\nreact-best-practices/SKILL.md\n```\n\n**Add to the Complementary Skills table:**\n\n```markdown\n| When task involves... | Invoke |\n|-----------------------|--------|\n| MongoDB, schema, queries | `mongodb-agent-skills:mongodb-schema-design` |\n| React, Next.js, UI | `react-best-practices` |\n```\n\n### Step 6: Confirm\n>\n> \"cc10x is set up! Please restart Claude Code to activate.\"\n\n---\n\n## The 4 Workflows\n\n| Intent | Trigger Words | What Happens |\n| -------- | --------------- | -------------- |\n| **BUILD** | build, implement, create, make, write, add | Clarify scope → TDD implementation → adversarial review → integration verification |\n| **DEBUG** | debug, fix, error, bug, broken, troubleshoot | Reproduce from evidence → isolate cause → validate fix → prove no regression |\n| **REVIEW** | review, audit, check, analyze, assess | High-signal review with confidence thresholds and file:line citations |\n| **PLAN** | plan, design, architect, roadmap, strategy | Turn rough intent into an execution-ready plan with explicit decisions |\n\n---\n\n## Memory Persistence\n\ncc10x survives context compaction. This is critical for long sessions.\n\n```\n.cc10x/\n├── activeContext.md   # What you're working on NOW\n│   - Current task\n│   - Active decisions (and WHY)\n│   - Learnings this session\n│\n├── patterns.md        # Project conventions\n│   - Code patterns\n│   - Common gotchas (bugs → fixes)\n│   - Architectural decisions\n│\n└── progress.md        # What's done, what's left\n    - Completed items (with evidence)\n    - Remaining tasks\n    - Blockers\n```\n\nThe live namespace is `.cc10x/` (memory `.cc10x/*.md`, workflow state `.cc10x/workflows/*`). Two legacy residue locations are ignored by current router hydration if present: `.claude/cc10x/` (pre-10.1.20, before the workflow state moved out of `.claude/` to escape the harness sensitive-file gate) and the version-segmented `.cc10x/v10/` layout (v10.x, before the namespace was de-versioned in v11). cc10x does not migrate either; a fresh `.cc10x/` is created on first use.\n\n**Iron Law:** Every workflow loads memory at START and updates at END.\n\n---\n\n## Optional MCP Integrations\n\ncc10x works out of the box with no MCPs required. These are **optional** — they unlock specific features when installed in your own Claude Code MCP settings.\n\n| MCP | Feature Unlocked | How to Install |\n| ----- | ----------------- | ---------------- |\n| **[octocode](https://github.com/nicepkg/octocode)** | GitHub research: find packages, search code across repos, read PR history. Triggered automatically when planner or bug-investigator needs external research. | Install via Claude Code MCP settings using the server name `octocode` |\n| **[brightdata](https://github.com/nicepkg/mcp-brightdata)** | Web scraping for research tasks — used as fallback when web content is needed beyond GitHub. | Install via Claude Code MCP settings using the server name `brightdata` |\n\n**Important:** CC10X no longer ships MCP server config inside the plugin. This avoids startup warnings for users who do not have Bright Data or Octocode credentials configured.\n\n**Without these MCPs:** cc10x still works fully. The research agents degrade to built-in Claude Code tools and note the lower-confidence path in their outputs.\n\n**With octocode installed:** When the router detects new/unfamiliar tech, 3+ failed debug attempts, or explicit research requests, it automatically calls octocode tools to search GitHub before invoking the planner or bug-investigator.\n\n---\n\n## Troubleshooting\n\n### Claude Code keeps asking for permission to edit memory files\n\nAdd these two lines to `~/.claude/settings.json` under `permissions.allow`:\n\n```json\n\"Edit(.cc10x/*)\",\n\"Write(.cc10x/*)\"\n```\n\nThese permission examples cover the live `.cc10x/` namespace.\n\nOr run **\"Set up cc10x for me\"** again — the setup wizard adds them automatically.\n\n---\n\n### cc10x asks for permission before writing plan/research docs\n\nWorking as intended. `.cc10x/` orchestration state is pre-permitted, but outward-facing project artifacts (`docs/plans/`, `docs/research/`, `docs/solutions/`) prompt for approval by design — trust-first friction, not missing configuration. Approve the write to continue.\n\n---\n\n### cc10x not activating in a specific project\n\nThe global `~/.claude/CLAUDE.md` activates cc10x in every project — you only need one install. If it's not activating in a specific project:\n\n1. **Check if that project has its own `.claude/CLAUDE.md`** — open it and verify the cc10x section is present. If the project-level file exists but doesn't have the cc10x entry, add it there.\n2. **Verify the entry format** — use `[CC10x]|entry: cc10x:cc10x-router` (plugin reference). A relative path like `./plugins/cc10x/...` only works in the cc10x repo itself, not in your projects.\n3. **Restart Claude Code** — the plugin system requires a restart after any CLAUDE.md change.\n\n---\n\n### Ubuntu / Linux install error: EXDEV cross-device link\n\nIf you see:\n\n```\nError: Failed to install: EXDEV: cross-device link not permitted\n```\n\nThis is a Linux filesystem issue — `/tmp` and your home directory are on different filesystems, so the installer can't `rename()` across them. Fix:\n\n```bash\n# Set TMPDIR to a directory on the same filesystem as your home:\nmkdir -p ~/.claude/tmp\nTMPDIR=~/.claude/tmp claude\n# Then install normally: /plugin install cc10x@cc10x\n```\n\nIf that doesn't work, install manually:\n\n```bash\n# Clone directly into the plugins directory\ngit clone https://github.com/romiluz13/cc10x.git ~/.claude/plugins/cc10x\n```\n\nThen follow Step 2 in the setup guide above to add the cc10x entry to `~/.claude/CLAUDE.md`.\n\n---\n\n### \"Unknown skill cc10x:cc10x-router\"\n\nThe cc10x plugin is disabled. Run:\n\n```\n/plugins enable cc10x\n```\n\nThen retry your command.\n\n---\n\n## Architecture Deep Dive\n\nEverything below the fold: how the loop actually works. For contributors and the curious — you do not need any of this to use cc10x.\n\n<details>\n<summary><strong>Runtime model, router kernel, agents, skills, hooks, diagrams, file layout</strong></summary>\n\n### Runtime Model\n\n#### 1. Router owns orchestration\n\n`cc10x-router` is the only orchestration authority.\n\nThe router now uses a **kernel + mandatory reference** shape:\n\n- universal orchestration law stays inline in `cc10x-router/SKILL.md`\n- workflow-specific playbooks and appendix-heavy artifact/remediation law live in `cc10x-router/references/*.md`\n- the kernel explicitly tells Claude which reference must be read before BUILD / DEBUG / REVIEW / PLAN branch logic continues\n\nThat keeps orchestration salient without turning the router into a context dump.\n\nIt decides:\n\n- which workflow to run\n- which subagent to invoke next\n- when to pause for clarification or scope decisions\n- when remediation is required\n- when a workflow is allowed to advance\n- when memory and workflow artifacts are finalized\n\nAgents do not own workflow state. They return structured results. The router interprets them.\n\n#### 2. Agents are narrow specialists\n\nThe shipped subagents are intentionally specialized:\n\n- `planner`\n- `plan-gap-reviewer`\n- `component-builder`\n- `bug-investigator`\n- `code-reviewer`\n- `failure-hunter`\n- `integration-verifier`\n- `researcher`\n- `doc-syncer`\n\nEach agent is optimized for one role. This keeps prompts sharper and makes workflow behavior easier to reason about.\n\n#### 3. Skills are reusable local instructions\n\nSkills are the reusable instruction layer that agents and the router depend on.\n\nThey provide:\n\n- planning patterns\n- TDD rules\n- debugging patterns\n- review rules\n- research synthesis\n- memory handling\n- verification-before-completion discipline\n\n#### 4. Workflow artifacts are the durable truth\n\ncc10x writes proof-of-work workflow state under:\n\n```text\n.cc10x/workflows/{wf}.json\n.cc10x/workflows/{wf}.events.jsonl\n```\n\nThese artifacts track:\n\n- workflow type and task ids\n- intent/spec context\n- agent results\n- evidence and quality state\n- remediation history\n- lifecycle events\n\nThis is what makes resume, review, and debugging more reliable than relying on chat context alone.\n\n#### 5. Hooks are guardrails, not a second orchestrator\n\ncc10x ships a minimal Claude Code-native hook set:\n\n- `PreToolUse`\n- `SessionStart`\n- `PostToolUse`\n- `TaskCompleted`\n\nHooks do not replace the router. They provide lightweight enforcement and diagnostics:\n\n- protected file and workflow write checks\n- resume context hydration\n- workflow artifact integrity audit\n- task metadata validation\n\nThis follows the official Claude Code pattern: hooks are small guardrails around tool use, not a parallel control plane.\n\n#### 6. MCP is optional acceleration only\n\ncc10x does **not** ship MCP server config inside the plugin.\n\nIf the user already has Claude Code MCP servers named:\n\n- `octocode`\n- `brightdata`\n\nthen research gets better automatically.\n\nIf not, the plugin still works. Research falls back to built-in Claude Code tools and records degraded confidence where appropriate.\n\n---\n\n### BUILD flow, up close\n\n```\n┌──────────────────────────────────────────────────────────────────────────────┐\n│                                                                              │\n│   YOU: \"build a user auth system\"                                            │\n│                                     ┌────────────────────────────────────┐   │\n│                              ┌─────►│  component-builder                 │   │\n│                              │      │  + TDD enforcement                 │   │\n│   ┌────────────────────┐     │      │  + code-generation skill           │   │\n│   │                    │     │      └──────────────┬─────────────────────┘   │\n│   │   cc10x-router     │─────┤                     │                         │\n│   │   (auto-detects    │     │      ┌──────────────▼─────────────────────┐   │\n│   │    BUILD intent)   │     │      │  code-reviewer ∥ silent-failure    │   │\n│   │                    │     │      │  (parallel execution)              │   │\n│   └────────────────────┘     │      └──────────────┬─────────────────────┘   │\n│                              │                     │                         │\n│                              │      ┌──────────────▼─────────────────────┐   │\n│                              └─────►│  integration-verifier              │   │\n│                                     │  + E2E validation                  │   │\n│                                     └────────────────────────────────────┘   │\n│                                                                              │\n└──────────────────────────────────────────────────────────────────────────────┘\n```\n\n---\n\n### Architecture\n\n```\nUSER REQUEST\n     │\n     ▼\n┌─────────────────────────────────────────────────────────────────┐\n│                    cc10x-router (ONLY ENTRY POINT)              │\n│              Detects intent → Routes to workflow                │\n└─────────────────────────────────────────────────────────────────┘\n     │\n     ├── BUILD ──► component-builder ──► [code-reviewer ∥ failure-hunter] ──► integration-verifier\n     │\n     ├── DEBUG ──► bug-investigator ──► code-reviewer ──► integration-verifier\n     │\n     ├── REVIEW ─► code-reviewer\n     │\n     └── PLAN ───► planner\n\nMEMORY (.cc10x/)\n├── activeContext.md  ◄── Current focus, decisions, learnings\n├── patterns.md       ◄── Project conventions, common gotchas\n└── progress.md       ◄── Completed work, remaining tasks\n\nWORKFLOW STATE (.cc10x/workflows/)\n├── {wf}.json         ◄── Durable workflow artifact\n└── {wf}.events.jsonl ◄── Append-only workflow event log\n```\n\n---\n\n### The 11 Agents\n\n| Agent | Purpose | Key Behavior |\n| ------- | --------- | -------------- |\n| **component-builder** | Builds features | TDD: RED → GREEN → REFACTOR (no exceptions) |\n| **bug-investigator** | Fixes bugs | LOG FIRST: Evidence before any fix |\n| **code-reviewer** | Reviews code | Confidence ≥80%; 6-pass adversarial review (security, performance, quality, friction, plan validity, spec compliance) |\n| **failure-hunter** | Hunts silent failures | Runs in parallel with code-reviewer; zero-tolerance for empty catches, log-only handlers, discarded errors |\n| **integration-verifier** | E2E validation | Exit codes: PASS/FAIL with evidence |\n| **doc-syncer** | Diff-driven documentation sync | Classifies doc impact; SKIPPED/PARTIAL/COMPLETE/FAIL contract gating Memory Update; honors `DIFF_DRIVEN_DOCS: skip` |\n| **planner** | Creates plans | Saves to `docs/plans/` + updates memory |\n| **plan-gap-reviewer** | Fresh plan challenge pass | Read-only anti-anchoring review before final plan handoff |\n| **researcher** | Web + GitHub research (Bright Data / Octocode MCP accelerators, built-in fallbacks) | Saves findings to file |\n| **triage-agent** | Triages incoming issues/PRs | Read-only; categorizes, verifies, checks redundancy + prior rejection, writes agent-ready briefs |\n| **architecture-scanner** | Codebase health audit | Read-only; scans for shallow modules + deepening candidates, produces HTML report with before/after diagrams |\n\n---\n\n## The 20 Skills\n\nSkills are **loaded automatically by agents**. You never invoke them directly.\n\n| Skill | Used By | Purpose |\n| ------- | --------- | --------- |\n| **agent-common** | ALL agents | Shared preamble: memory protocol, CONTRACT envelope, output rules |\n| **memory-and-handoff** | main session (router-gated) | Persist context across compaction; portable handoff package |\n| **verification** | builder, verifier, investigator | Evidence before claims: gate function, validation levels, evidence array |\n| **building** | component-builder | TDD RED-GREEN-REFACTOR, false-RED guard, integration & live proof |\n| **debugging** | bug-investigator | Root cause analysis, feedback loop first, blast radius after fix |\n| **code-review** | main session | Verify human/external review feedback before agreeing or implementing |\n| **planning** | planner | Comprehensive plans + plan completeness gate |\n| **plan-review-gate** | planner | Final fail-closed plan sanity gate before handoff |\n| **architecture** | planner, builder (router-gated) | System & API design for multi-component work |\n| **frontend** | UI work (router-gated) | Authoring + critique modes: UX, accessibility, DESIGN.md, AI-slop detection |\n| **exploration** | PLAN workflow (router-gated) | Design brainstorm + throwaway spike modes with machine-readable handoff |\n| **diff-driven-docs** | doc-syncer | Doc impact classification + audit-doc format |\n| **research** | planner, bug-investigator (via researcher agent) | Synthesis-only: how to interpret research results |\n| **codebase-hygiene** | code-reviewer, planner (router-gated) | Semantic-duplicate audit + shallow-module deepening |\n| **codebase-design** | planner, builder, investigator, reviewer (router-gated) | Canonical deep-module vocabulary: module, interface, depth, seam, adapter, leverage, locality |\n| **domain-modeling** | planner, doc-syncer (active); builder, investigator (read-only) | Active glossary discipline: challenge terms, sharpen language, write CONTEXT.md + ADRs |\n| **mcp-cli** | researcher | On-demand MCP server use without permanent context pollution |\n| **update** | maintainers | Maintenance meta-skill for updating cc10x itself |\n| **cc10x-guide** | users asking about cc10x (model-invoked) | Answers questions about cc10x itself: install, setup, workflows, memory, troubleshooting; never executes work |\n| **resolving-merge-conflicts** | any agent hitting a git conflict (model-invoked) | Resolve merge/rebase conflicts hunk by hunk by intent; never --abort |\n\n> `cc10x-router` is the entry-point skill that routes every workflow; it ships alongside these 20.\n\n---\n\n### Task-Based Orchestration\n\ncc10x uses Claude Code's Tasks system for workflow coordination:\n\n```\n┌─────────────────────────────────────────────────────────────────┐\n│  BUILD: User Authentication                                     │\n│  ├── component-builder (pending)                                │\n│  ├── code-reviewer (blocked by: builder)                        │\n│  ├── failure-hunter (blocked by: builder)                │\n│  └── integration-verifier (blocked by: reviewer, hunter)        │\n└─────────────────────────────────────────────────────────────────┘\n```\n\n- **Dependency chains**: Agents wait for blockers to complete\n- **Parallel execution**: reviewer + hunter run simultaneously\n- **Resume capability**: TaskList() checks for active workflows\n- **Automatic handoff**: Each agent updates status when done\n- **Router-owned advancement**: only the router decides whether a workflow can continue\n\n---\n\n### Hooks\n\nThe plugin currently ships these Claude Code-native hooks:\n\n| Hook | Purpose |\n| ------ | --------- |\n| `PreToolUse` | Guard protected files and workflow-owned writes |\n| `SessionStart` | Rehydrate workflow context after restart or compaction |\n| `PostToolUse` | Audit workflow artifact integrity after writes |\n| `TaskCompleted` | Validate CC10X task metadata before task completion |\n| `PostCompact` | Capture compaction events for workflow telemetry |\n| `SubagentStop` | Audit subagent outputs for required contract markers |\n| `PreCompact` | Snapshot workflow state before compaction |\n| `Stop` | Snapshot workflow state on session stop |\n| `StopFailure` | Log API failure telemetry asynchronously |\n| `InstructionsLoaded` | Audit instruction-file loading asynchronously |\n\nThese hooks are intentionally minimal. They improve reliability without turning the plugin into a second runtime.\n\n---\n\n### Plan → Build Automation\n\nPlans don't get lost. They flow automatically to the build phase.\n\n```\nDay 1: \"plan a task tracker\"\n  → Planner creates docs/plans/2025-01-29-task-tracker-plan.md\n  → Memory updated with plan reference\n\nDay 2: \"build it\"\n  → Router detects BUILD\n  → component-builder finds plan in memory\n  → Follows plan phases with TDD\n```\n\n---\n\n### Expected Behavior\n\n#### When You Say \"Build a Task Tracker\"\n\n**cc10x Response:**\n\n```\nDetected BUILD intent. Executing BUILD workflow.\n\nLoading memory...\n✓ Memory loaded\n\nClarifying requirements (REQUIRED):\n1. What framework/language? (Node.js, Python, etc.)\n2. Should tasks have priorities, due dates, categories?\n3. CLI app or web app?\n4. Any features beyond basic CRUD?\n\nWaiting for your answers before proceeding.\n```\n\n**Without cc10x:**\n\n```\nI'll help you build a task tracker! Let me start...\n[Writes code without asking]\n[Skips tests]\n[Claims \"it should work\"]\n```\n\n---\n\n### Files Structure\n\n```\nplugins/cc10x/\n├── .claude-plugin/\n│   └── plugin.json\n├── hooks/\n│   └── hooks.json\n├── config/\n│   └── hook-mode.json\n├── scripts/\n│   ├── cc10x_event_logger.py\n│   ├── cc10x_git_guard.py\n│   ├── cc10x_hooklib.py\n│   ├── cc10x_posttooluse_artifact_guard.py\n│   ├── cc10x_pretooluse_guard.py\n│   ├── cc10x_sessionstart_context.py\n│   ├── cc10x_state_persist.py\n│   └── cc10x_task_completed_guard.py\n├── tools/\n│   ├── review_package.py\n│   ├── phase_brief.py\n│   └── live_harness_runner.py\n├── tests/\n│   └── fixtures/\n├── agents/\n│   ├── component-builder.md\n│   ├── bug-investigator.md\n│   ├── code-reviewer.md\n│   ├── integration-verifier.md\n│   ├── doc-syncer.md\n│   ├── planner.md\n│   ├── plan-gap-reviewer.md\n│   ├── researcher.md\n│   ├── triage-agent.md\n│   ├── architecture-scanner.md\n│   └── references/silent-failure-red-flags.md\n│\n└── skills/\n    ├── cc10x-router/\n    │   ├── SKILL.md\n    │   └── references/\n    │       ├── workflow-artifact-and-hook-policy.md\n    │       ├── workflow-artifact.skeleton.json\n    │       ├── build-workflow.md\n    │       ├── debug-workflow.md\n    │       ├── review-workflow.md\n    │       ├── plan-workflow.md\n    │       ├── triage-workflow.md\n    │       ├── codebase-health-workflow.md\n    │       └── remediation-and-research.md\n    ├── agent-common/SKILL.md\n    ├── memory-and-handoff/SKILL.md\n    ├── building/SKILL.md\n    ├── debugging/SKILL.md\n    ├── code-review/SKILL.md\n    ├── planning/SKILL.md\n    ├── plan-review-gate/SKILL.md\n    ├── architecture/SKILL.md\n    ├── frontend/SKILL.md\n    │   └── references/{ui-state-and-feedback,accessibility-and-forms,performance-and-layout,design-md-authoring,design-md-inspiration-index}.md\n    ├── exploration/SKILL.md\n    ├── diff-driven-docs/SKILL.md\n    ├── research/SKILL.md\n    ├── verification/SKILL.md\n    ├── codebase-hygiene/SKILL.md\n    ├── codebase-design/SKILL.md\n    │   └── references/{DEEPENING,DESIGN-IT-TWICE}.md\n    ├── domain-modeling/SKILL.md\n    │   └── references/{CONTEXT-FORMAT,ADR-FORMAT}.md\n    ├── mcp-cli/SKILL.md\n    ├── resolving-merge-conflicts/SKILL.md\n    ├── cc10x-guide/SKILL.md\n    └── update/SKILL.md\n```\n\nAdditional developer docs live under:\n\n```text\ndocs/cc10x-orchestration-bible.md\ndocs/cc10x-orchestration-logic-analysis.md\ndocs/cc10x-orchestration-safety.md\ndocs/router-invariants.md\n```\n\nIf you need to understand or evolve the harness, start there after reading `cc10x-router`.\n\n</details>\n\n---\n\n## Version History\n\n<details>\n<summary><strong>Release history (v5.3 → v12.8.0)</strong></summary>\n\n| Version | Highlights |\n| --------- | ------------ |\n| **v10.1.20** | Escape the Claude Code sensitive-file gate: workflow state root relocated from `.claude/cc10x/v10/` to `.cc10x/v10/` across every prompt surface, runtime hook, and fixture. Every router fanout, event-log append, and memory refresh is now silent in default-permission setups. Audit now catches runtime/prompt path drift so this class of regression fails self-tests before release. |\n| **v10.1.19** | Harmony hardening release: contradiction cleanup across router-facing instructions, router-owned self-contained handoffs, phase-local BUILD context, early memory capture before validation, review/hunt fan-in at the router, and full docs/release metadata alignment. |\n| **v10.1.15** | Hook expansion: 4 audit-only hooks (PreCompact, Stop, StopFailure, InstructionsLoaded) for workflow state persistence and telemetry. 6→10 hook events. Zero blocking, zero context injection, router remains sole authority. |\n| **v10.1.14** | Multi-repo harmony integration: 29 certified patterns from 11 reference repos via 3-phase harmony pipeline. Test tampering detection, claim extraction, environment escape hatches, analysis paralysis guards, near-miss negative testing, de-sloppify scans, plans-are-prompts principle, professional objectivity hard rules. |\n| **v10.1.13** | Ruflo harmony integration: 29 prompt engineering edits across 16 files — research quality heuristics, multi-language silent-failure detection, friction-scan thresholds, rollback decision trees, plan completeness gates, behavioral TDD focus, partial-phase review scoping, abstraction thresholds, split-brain contradiction handling, evidence-before-reporting hard rules |\n| **v10.1.12** | Prompt engineering uplift: 15 techniques from mattpocock/skills integrated across 13 files — durable decisions, tracer bullets, vertical-slice TDD, dependency taxonomy, HITL/AFK phases, opinionated review, friction scan, scope assessment, domain context injection |\n| **v10.1.11** | DAG-visible PLAN review loop: the full bounded planning review chain is now pre-created in the task graph, with explicit branch pruning and `plan-gap-reviewer` restored to `gpt-5.4-mini` |\n| **v10.1.10** | Always-on fresh planning review: every saved plan artifact now queues the bounded `plan-gap-reviewer` task before final plan handoff, with replay coverage locking it in |\n| **v10.1.4** | Fresh planning review cleanup: raw user request passed to `plan-gap-reviewer`, lighter read-only reviewer contract, bounded pass counting fixed, docs/version surfaces refreshed |\n| **v10.1.3** | Planning recovery: code-grounded plans, explicit plan-vs-code gap surfacing, stronger repo-aware plan review, and planning-specific replay coverage |\n| **v10.1.2** | Trust-preserving latency instrumentation: verifier workload telemetry, phase-exit vs extended-audit classification, no proof-gating change |\n| **v10.1.1** | Prompt-only hardening: sharper anti-false-completion wording, better trigger/description hygiene, reduced prompt dilution, no orchestration/runtime changes |\n| **v10.1.0** | Competition-grade release: decision-grade planning, adversarial plan gates, proof-oriented BUILD, harsher VERIFY, and benchmark-backed prompt/harness hardening |\n| **v10.0.0** | Trust-first recovery: agreement-first planning, phase-gated BUILD, stable workflow UUIDs, versioned v10 state, advisory internal skills |\n| **v9.1.1** | Removed shipped MCP config to avoid startup warnings; MCP research remains optional via user-configured Claude Code MCP servers named `brightdata` and `octocode` |\n| **v9.1.0** | Publication polish: intent-first planning, BDD-style scenario evidence, DDD-style domain language preservation, proof-of-work workflow artifacts, built-in harness drift audit |\n| **v9.0.0** | Plugin-native packaging: bundled Claude Code hooks, optional plugin MCP acceleration, router-owned research quality model, workflow artifacts as durable truth |\n| **v8.5.0** | Fix 1: READ-ONLY task completion as explicit mandatory gate (3-GATE). Fix 2b: CRITICAL+HIGH scope question via text (Rule 1a-SCOPE + Scope Decision Resume + scope-aware re-hunt) |\n| **v8.0.0** | Radical simplification — remove Router Contract YAML from read-only agents; text-based verdict extraction; JUST_GO session mode; ~280 lines removed |\n| **v7.9.0** | OBS-2/3/4/6/7/8/10/11/12/13/14 batch fix — self-healing verifier, explicit DEBUG-RESET marker, conditional frontend-patterns load |\n| **v7.8.0** | OBS-1/9/15/16/DEBUG-RESET — 5-issue fix, 13/13 smoke test pass |\n| **v6.0.21** | User standards support; multi-project docs; Linux install troubleshooting |\n| **v6.0.20** | Agent self-report task completion; MCP docs; permissions fix for memory files |\n| **v6.0.19** | Babysitter-inspired: Multi-signal HARD/SOFT scoring, evidence arrays, decision checkpoints, completion guard |\n| **v6.0.0** | Orchestration hardening: Tasks contract correctness + Task-enforced gates + re-review loop |\n| **v5.25.1** | GSD-inspired enhancements (wiring verification, hypothesis criteria) |\n| **v5.25.0** | Critical orchestration fixes + README redesign |\n| **v5.24.0** | Research persistence with THREE-PHASE pattern |\n| **v5.23.0** | Plan-task linkage (legacy: metadata.planFile; now deprecated) |\n| **v5.22.0** | Stub detection patterns |\n| **v5.21.0** | Task-based orchestration with TaskCreate/TaskUpdate |\n| **v5.20.0** | Goal-backward verification lens |\n| **v5.13.0** | Parallel agent execution (~30-50% faster) |\n| **v5.10.0** | Anthropic Claude 4.x best practices |\n| **v5.9.1** | Plan→Build automatic connection |\n\n<details>\n<summary>Full version history</summary>\n\n- **v8.0.0** - Radical Simplification: Removed Router Contract YAML from code-reviewer, silent-failure-hunter, integration-verifier. Replaced ~200-line YAML validation block in router with 30-line text extraction (reads heading from first 5 lines). Added JUST_GO session mode (AUTO_PROCEED flag). Simplified Empty Answer Guard — only ⚠️ REVERT gates block; all others auto-default. Removed REM-EVIDENCE retry loop (root cause of 6/6 stress test failures). Net: ~280 lines removed, 0 new complexity.\n- **v7.9.0** - OBS-2/3/4/6/7/8/10/11/12/13/14 batch fix: self-healing integration-verifier (creates REM-FIX + blocks own task), explicit DEBUG-RESET marker written by router, conditional frontend-patterns load (.tsx/.jsx/.vue/.css/.scss/.html only)\n- **v7.8.0** - OBS-1/9/15/16/DEBUG-RESET 5-issue fix, 13/13 smoke test pass\n- **v6.0.19** - Babysitter-inspired enhancements: Multi-signal HARD/SOFT scoring (per-dimension review), evidence array protocol (structured proof), decision checkpoints (mandatory pause points), completion guard (final gate before Router Contract)\n- **v6.0.0** - Orchestration hardening:\n  - Tasks contract correctness (no undocumented TaskCreate fields; canonical TaskUpdate object form)\n  - CC10X task namespacing + safer resume rules\n  - Task-enforced gates + re-review loop after remediation (prevents unreviewed changes)\n- **v5.25.1** - GSD-inspired enhancements: Wiring verification patterns, hypothesis quality criteria, cognitive biases table\n- **v5.25.0** - Critical orchestration fixes: Plan propagation, results collection, skill hierarchy, validation + README redesign\n- **v5.24.0** - Research Documentation Persistence: THREE-PHASE research pattern\n- **v5.23.0** - Plan-Task Linkage: metadata.planFile for context recovery (legacy; deprecated in v6.0.0)\n- **v5.22.0** - Stub Detection Patterns: GSD-inspired stub detection\n- **v5.21.0** - Task-Based Orchestration: TaskCreate, TaskUpdate, TaskList integration\n- **v5.20.0** - Goal-Backward Lens: Verification enhancements\n- **v5.19.0** - OWASP Reference + Minimal Diffs + ADR patterns\n- **v5.18.0** - Two-Phase github-research\n- **v5.13.1** - Bulletproof chain enforcement\n- **v5.13.0** - Parallel agent execution\n- **v5.12.1** - Fixed orphan skills\n- **v5.12.0** - Pre-publish audit\n- **v5.11.0** - Workflow chain enforcement\n- **v5.10.6** - Foolproof Router with decision tree\n- **v5.10.5** - Complete Permission-Free Audit\n- **v5.10.4** - True Permission-Free Memory\n- **v5.10.3** - Fixed invalid agent color\n- **v5.10.2** - Permission-Free Memory\n- **v5.10.1** - Router Supremacy\n- **v5.10.0** - Anthropic Claude 4.x alignment\n- **v5.9.1** - Plan→Build connection\n- **v5.9.0** - Two-step save pattern\n- **v5.8.1** - Router bypass prevention\n- **v5.7.0** - Fixed agent keyword conflicts\n- **v5.6.0** - Fixed agent tool misconfigurations\n- **v5.5.0** - Fixed skill keyword conflicts\n- **v5.4.0** - Router AUTO-EXECUTE\n- **v5.3.0** - Confidence scoring, silent-failure-hunter\n\n</details>\n\n</details>\n\n---\n\n## Contributing\n\n- Star the repository\n- Report issues\n- Suggest improvements\n\n---\n\n## License\n\nMIT License\n\n---\n\n<p align=\"center\">\n  <strong>cc10x v12.8.0</strong><br>\n  <em>The Intelligent Orchestrator for Claude Code</em>\n</p>\n",
  "bytes": 40555,
  "sha": "c35218cf80ad39312e04aa4e99be17ca6601dab0ba5edd11fdbc3fdc7c95c7e8",
  "repo_slug": "romiluz13/cc10x",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_romiluz13_cc10x_cc10x_router_owned_harne_0c505a9b/readme"
}