Terse

Compress output by removing filler, pleasantries, articles, and hedging while preserving code, terms, and errors for concise technical responses.

Security Scan

Terse is internally coherent: it provides prompt-prefix-based compression for sub-agents and includes only a small local helper script; main concerns are a metadata inconsistency and the runtime risk of accidentally applying compression to owner-facing or critical tasks.

Purpose & Capability

Name/description state a token-compression helper for agent responses and the package contains only a prompt-prefix generator and documentation that implement that feature. Required binaries/env/configs are none — proportionate to the described purpose.

Instruction Scope

SKILL.md focuses on generating compressed prompts and warns explicitly about 'hard exclusion' cases (planning, security, owner-facing output). That is appropriate, but the skill relies on the host/orchestrator to respect those exclusion rules; misuse (applying terse to critical outputs) would degrade quality. Also SKILL.md claims the skill 'auto-loads and applies to sub-agent responses' — ensure the platform enforces the stated exclusions.

Install Mechanism

No install spec and only a small, local Python helper script (caveman_prompt.py). No downloads from external URLs, no archive extraction, and no package manager pulls — low installation risk.

Credentials

No required environment variables, no credentials, and no config paths. The helper script is purely local string handling and does not access secrets or network resources.

Persistence & Privilege

Registry metadata shows always: false, but SKILL.md head includes metadata.openclaw: always: true (auto-classified). This is an inconsistency: if the platform honored the SKILL.md metadata and forced always-load, it would increase runtime presence and the blast radius of accidental misuse. Confirm which source the platform uses for the 'always' flag before enabling auto-application.

Assessment

This skill is basically a prompt-prefix/template for compressing sub-agent output and the included Python helper is harmless text-manipulation code. Before installing: 1) Confirm the platform will not apply 'terse' to owner-facing, planning, or security-sensitive outputs (the SKILL.md warns against that). 2) Ask the platform which 'always' source it honors — the SKILL.md embeds always: true while the registry shows always: false; ensure the skill will not be force-loaded. 3) Test on low-risk internal tasks (logs, CI steps, code implement/debug subtasks) to validate quality. 4) Review the linked repository yourself if you want extra assurance; there are no network calls or secret access in the included files. If you need help with specific checks (e.g., how your platform resolves SKILL.md metadata), share your platform details and I can guide next steps.

Terse Skill

🪨 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 — while keeping code, technical terms, and error messages verbatim.

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

---

⛔ HARD EXCLUSION RULES — NEVER USE terse FOR:

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

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

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

---

✅ APPROVED USE CASES:

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

• Code implementation — debug, refactor, fix bugs, write functions
• Lookups & queries — "what does this function do", "find the config for X"
• File operations — "list logs", "check disk space", "grep for X"
• Health checks & monitoring — cron job status, service checks, log parsing
• CI/CD steps — build, test, lint, deploy (non-decision parts)
• Data extraction — parse JSON, extract fields, transform data
• Internal agent handoffs — tool-to-tool communication where no human reads output
• Quick summaries — "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

# ✅ 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"
)

# ❌ BAD: planning task — DO NOT apply terse
sessions_spawn(
    task="Design the v0.7.0 architecture for EvoClaw. Consider the Phase 3 requirements...",
    model="CC-Opus46"
)

Via helper script

uv run python ~/.openclaw/workspace/skills/terse/scripts/caveman_prompt.py --level full "your task here"

---

Benchmarks (from caveman repo)

Task	Normal tokens	Terse tokens	Saved
React re-render bug	1180	159	87%
PostgreSQL pool setup	2347	380	84%
Git rebase conflict	891	374	58%
Average	—	—	~65–75%

March 2026 paper: brevity constraints improved accuracy by 26pp.

---

Model Pairing

Level	Best model	Why
Lite	Any	Minimal overhead
Full	Sonnet 4.6	Follows compression well, still accurate
Ultra	Haiku 4.5	Cheap + short = ultra-efficient

---

Integration

• orchestrator: Apply terse to Builder steps only (NOT Planner or Reviewer)
• clawmemory: Already terse by design; no change needed
• knowledge-base: Search → terse summary → save context tokens

---

Files

• SKILL.md — this file
• scripts/caveman_prompt.py — helper to generate prefixed prompts

---

Repo

https://github.com/AlexChen31337/terse