conserve

Audit codebase for AI-generated code quality issues (vibe coding, Tab bloat, slop)

Progressive bloat detection for dead code and duplication

Analyze and optimize context window usage using MECW principles

Safe bloat remediation with user approval at each step

Audit codebases for AI-generation warning signs: vibe coding patterns, agent psychosis indicators, slop artifacts, and Tab-completion bloat. Specialized complement to bloat-auditor.

Execute progressive bloat detection scans (Tier 1-3), generate prioritized reports, and recommend cleanup actions.

Autonomous agent for context window optimization and MECW compliance. Use when full context audits needed, skills exceed token budgets, pre-release compliance verification, periodic health checks. Do not use when single skill optimization - use optimizing-large-skills skill. quick token counts - use skills-eval instead. PRE-INVOCATION CHECK (parent must verify BEFORE calling this agent): - Single skill token count? Parent runs `wc -w skill.md` or estimates - Quick size check? Parent reads file header - One-off query? Parent uses Read tool directly ONLY invoke this agent for: full plugin audits, growth trend analysis, optimization recommendations, or pre-release compliance verification.

Lightweight agent designed to continue work from a session state checkpoint. Spawned when the parent agent exceeds context thresholds. This agent: 1. Reads the session state file 2. Re-establishes necessary context 3. Continues the task without interruption 4. Can spawn another continuation agent if needed

Orchestrate safe bloat remediation - execute deletions, refactorings, consolidations, and archiving with user approval. Creates backups, runs tests, provides rollback.

Track per-agent token usage and flag waste patterns in parallel dispatch workflows. Consult after running parallel agents to evaluate whether expenditure was proportional to value. Cross-references the plan-before-large-dispatch rule.

Detect codebase bloat through progressive analysis: dead code, duplication, complexity, documentation bloat. Use when context usage high, quarterly maintenance, pre-release cleanup, before refactoring. Do not use when active feature development, time-sensitive bugs, codebase < 1000 lines.

Automatic context management with graceful handoff to continuation subagent. Use when Context usage approaches 80% during long-running tasks. This skill enables automatic continuation without manual /clear. The key insight: Subagents have fresh context windows. By delegating remaining work to a continuation subagent, we achieve effective "auto-clear" without stopping the workflow.

Provides KISS, YAGNI, and SOLID code quality principles for clean code,

Analyze current context and recommend compression strategies. Use when context feels bloated, before major task transitions, or when quota is burning fast. location: user token_budget: 150. Do not use when context-optimization already invoked or simple queries.

Use this skill BEFORE starting complex tasks. Check context levels proactively. Use when context usage approaches 50% of window, tasks need decomposition, complex multi-step operations planned, context pressure is high. Do not use when simple single-step tasks with low context usage. DO NOT use when: already using mcp-code-execution for tool chains.

Use this skill BEFORE resource-intensive operations. Establish baselines proactively. location: plugin token_budget: 400. Use when session starts (auto-load with token-conservation), planning builds or training that could pin CPUs/GPUs for >1 minute, retrying failed resource-heavy commands. Do not use when simple operations with no resource impact. DO NOT use when: quick single-file operations.

Guidance on when to ask clarifying questions vs proceed with standard

Use this skill BEFORE building complex tool chains. Optimize proactively. location: plugin token_budget: 200. Use when >3 tools chained sequentially, large datasets (>10k rows), large files (>50KB), context usage >25%. Do not use when simple tool calls that do not chain. DO NOT use when: context pressure is low and tools are fast.

hype, and unnecessary framing. Includes termination and directness guidelines.

balancing accuracy with token efficiency.

Use this skill at the START of every session. This is MANDATORY for quota management. location: plugin token_budget: 300. Use when session starts (mandatory), prompt sizes spike, tool calls increase, before long-running analyses or massive context loads. Do not use when context-optimization already handles the scenario. DO NOT use when: simple queries with minimal context.

5 hooks across 5 events