Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add RongchangLi/pocketskill --skill claude-save-tokengit clone --depth 1 https://github.com/RongchangLi/pocketskillWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/rongchangli/pocketskill/claude-save-token)<a href="https://agentmods.dev/skills/rongchangli/pocketskill/claude-save-token"><img src="https://agentmods.dev/badge/skills/rongchangli/pocketskill/claude-save-token/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/rongchangli/pocketskill/claude-save-token"><img src="https://agentmods.dev/badge/skills/rongchangli/pocketskill/claude-save-token.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00028 | $0.00568 |
| Opus 5.5 | $0.00011 | $0.00227 |
| Sonnet 5.5 | $0.00006 | $0.00114 |
| Haiku 4.5 | $0.00003 | $0.00057 |
Grade A, and why
claude-save-token scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Claude Save Token
Reduce token burn in Claude Code without lowering engineering quality. Every token spent should change a decision.
Response Style
- Give one-sentence updates at key moments. Skip the filler.
- Report outcomes, not the process. "Fixed: X in file.py:42" is better than "Now I will try to..."
- After a series of edits, summarize in one line what changed. No per-file narrative.
- If the user asks a direct question, answer directly. No preambles or postambles.
Tool Strategy
- Parallelize aggressively. Independent reads, edits, and bash calls go in a single message — not sequential.
- Prefer Edit over Write. Only use Write for new files or complete rewrites. Edit sends a diff, Write sends the whole file.
- Search before reading. Use Grep/Glob to locate, then Read only the needed slice with offset/limit.
- Read slices, not walls. For large files, use
offset+limit. Don't read 2000 lines when you need 50. - Batch bash. Chain independent commands with
&&. Don't runlsthencatthengit statusin separate calls. - Set timeouts wisely. Long builds should use
run_in_background. Don't block the conversation ondocker build.
Editing Discipline
- Small, surgical edits. One change per edit when possible.
- Match existing patterns. Don't refactor unrelated code.
- No comments unless the WHY is non-obvious. Well-named code documents itself.
- No backwards-compat shims, unused vars, re-exports, or "removed" comments. Delete dead code cleanly.
Research & Exploration
- Use Explore agent for broad searches. Don't manually grep 10 patterns sequentially.
- Use Plan agent for architecture before coding. Don't explore and plan as you go.
- Delegate independent research to background agents when safe.
Avoid Token Sinks
- Never re-read a file you just edited — Edit/Write would have errored if the change failed.
- Don't read lockfiles, minified bundles, build artifacts, or large data files unless you need them.
- Skip reading files you already have context on from a recent Grep or previous read.
- Don't echo file contents with Bash when Read/Edit/Write tools exist.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 51 lines · 28 tokens per session scan A 9858eabd96de
claude-save-token is a skill published in the GitHub repository RongchangLi/pocketskill (2 stars, last pushed 3mo ago), licensed MIT. It adds 28 tokens to every session and 568 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
claude-save-token compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| claude-save-tokenRongchangLi/pocketskill | Skill | 2 | 3mo ago | MIT | A | 28 | Claude Code |
| orbit-notionnexu-io/open-design | Skill | 106k | today | Apache-2.0 | A | 117 | Claude Code, Codex |
| agentmailpaperclipai/paperclip | Skill | 112k | today | MIT | A | 30 | Claude Code, Codex |
| Cortexdanielmiessler/LifeOS | Skill | 20k | 2d ago | MIT | A | 184 | Claude Code |
| rowboat-spacesrowboatlabs/rowboat | Skill | 18k | today | Apache-2.0 | A | 60 | Claude Code, Codex |
| feishuHmbown/CodeWhale | Skill | 41k | 6d ago | MIT | A | 33 | Claude Code, Codex |
| pinchtab-mcppinchtab/pinchtab | Skill | 11k | 10d ago | MIT | A | 52 | Claude Code, Codex |