ClawKit Logo
ClawKitReliability Toolkit
Back to Registry
Official Verified

Terse

Skill by bowen31337

skill-install β€” Terminal

Install via CLI (Recommended)

clawhub install openclaw/skills/skills/bowen31337/terse
Or

Terse Skill

πŸͺ¨ why use many token when few token do trick

Compressed output mode for sub-agents. Cuts ~65–75% of output tokens by stripping filler words, pleasantries, articles, and hedging β€” while keeping code, technical terms, and error messages verbatim.

Based on: https://github.com/JuliusBrussee/caveman


β›” HARD EXCLUSION RULES β€” NEVER USE terse FOR:

These task types require full expressive output. Applying terse here degrades quality:

  • Planning β€” strategic plans, roadmaps, sprint planning, milestone design
  • Critical thinking β€” risk analysis, security audits, trade-off evaluations, incident response
  • Solution architecture β€” system design, API contracts, data models, infrastructure decisions
  • Article/writing β€” blog posts, MbD content, Payhip books, tweets, emails, any user-facing prose
  • Owner-facing communication β€” main session replies, status updates to the human operator
  • Code review β€” PR reviews, architecture reviews, design doc feedback
  • Prompt engineering β€” system prompts, skill instructions, agent personas

If in doubt: DON'T compress. Full output is always safer than compressed output.


βœ… APPROVED USE CASES:

Terse is safe and beneficial for these internal, non-critical sub-agent tasks:

  • Code implementation β€” debug, refactor, fix bugs, write functions
  • Lookups & queries β€” "what does this function do", "find the config for X"
  • File operations β€” "list logs", "check disk space", "grep for X"
  • Health checks & monitoring β€” cron job status, service checks, log parsing
  • CI/CD steps β€” build, test, lint, deploy (non-decision parts)
  • Data extraction β€” parse JSON, extract fields, transform data
  • Internal agent handoffs β€” tool-to-tool communication where no human reads output
  • Quick summaries β€” "summarize this URL/file" for internal context (NOT for owner-facing output)

Compression Levels

Lite

Drop filler phrases, hedging. Keep full sentences.

Prefix: Be concise. Skip filler phrases, pleasantries, and unnecessary hedging. Keep technical terms and code verbatim.

Full (default)

Omit articles, use fragments, bare imperatives. Code/errors verbatim.

Prefix: CAVEMAN MODE: Omit articles, filler, pleasantries. Use fragments. Steps as bare imperatives. Keep code/errors verbatim. No apologies. No "I". Just signal.

Ultra

Max compress. Labels only. No sentences. Code verbatim.

Prefix: ULTRA CAVEMAN: Max compress. Drop ALL non-essential words. Labels only. No sentences. Keep code verbatim.


How to Apply in sessions_spawn

# βœ… GOOD: internal code task
sessions_spawn(
    task="CAVEMAN MODE: Omit articles, filler, pleasantries. Use fragments. Steps as bare imperatives. Keep code/errors verbatim. No apologies. No \"I\". Just signal.\n\nFix the goroutine leak in internal/server/pool.go",
    model="CC-Sonnet46"
)

Metadata

Stars4190
Views1
Updated2026-04-18
View Author Profile
AI Skill Finder

Not sure this is the right skill?

Describe what you want to build β€” we'll match you to the best skill from 16,000+ options.

Find the right skill
Add to Configuration

Paste this into your clawhub.json to enable this plugin.

{
  "plugins": {
    "official-bowen31337-terse": {
      "enabled": true,
      "auto_update": true
    }
  }
}
Safety NoteClawKit audits metadata but not runtime behavior. Use with caution.

Related Skills

Identity Resolver

Skill by bowen31337

bowen31337 4190

whalecli

Agent-native whale wallet tracker for ETH and BTC chains. Track large crypto wallet movements, score whale activity, detect accumulation/distribution patterns, and stream real-time alerts. Integrates with FearHarvester and Simmer prediction markets for closed-loop signal→bet workflows. Use when: user asks about whale activity, on-chain signals, large wallet movements, smart money flows, or when pre-validating crypto trades/bets with on-chain data.

bowen31337 4190

agent-self-governance

Self-governance protocol for autonomous agents: WAL (Write-Ahead Log), VBR (Verify Before Reporting), ADL (Anti-Divergence Limit), VFM (Value-For-Money), and IKL (Infrastructure Knowledge Logging). Use when: (1) receiving a user correction β€” log it before responding, (2) making an important decision or analysis β€” log it before continuing, (3) pre-compaction memory flush β€” flush the working buffer to WAL, (4) session start β€” replay unapplied WAL entries to restore lost context, (5) any time you want to ensure something survives compaction, (6) before claiming a task is done β€” verify it, (7) periodic self-check β€” am I drifting from my persona? (8) cost tracking β€” was that expensive operation worth it? (9) discovering infrastructure β€” log hardware/service specs immediately.

bowen31337 4190

solidity-lsp

Solidity language server providing smart contract development support including compilation, linting, security analysis, and code intelligence for .sol files. Use when working with Ethereum smart contracts, Substrate pallets, or any Solidity code that needs compilation, security checks, gas optimization, or code navigation. Essential for ClawChain pallet development.

bowen31337 4190

pyright-lsp

Python language server (Pyright) providing static type checking, code intelligence, and LSP diagnostics for .py and .pyi files. Use when working with Python code that needs type checking, autocomplete suggestions, error detection, or code navigation.

bowen31337 4190