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 JesusGarcia9009/lethe-engineering --skill claude-codegit clone --depth 1 https://github.com/JesusGarcia9009/lethe-engineeringWrote 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/jesusgarcia9009/lethe-engineering/claude-code)<a href="https://agentmods.dev/skills/jesusgarcia9009/lethe-engineering/claude-code"><img src="https://agentmods.dev/badge/skills/jesusgarcia9009/lethe-engineering/claude-code.svg" alt="Measured on agentmods" 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 | $0.00032 | $0.00367 |
| Opus 5 | $0.00016 | $0.00183 |
| Sonnet 5 | $0.00006 | $0.00073 |
| Haiku 4.5 | $0.00003 | $0.00037 |
Grade A, and why
lethe-offload 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 7d 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.
What it actually says
LETHE Offload
You have a LETHE MCP memory available. Use it to keep your context window small during long tasks, which saves tokens and preserves your attention for what matters.
The rule
When a tool result, file read, log, or API response is large (roughly > 1500 tokens) and you will not need all of it in the next step or two, archive it instead of keeping it:
- Call
lethe_archive(content=<the big text>, label=<short name>). - Keep only the returned handle (e.g.
handle=a3f9) in your reasoning. Drop the full text. - When you need it again, call
lethe_recall("a3f9")(by handle) orlethe_recall("<keywords>")(by content) to get it back.
When to archive
- ✅ Verbose file dumps you only needed to scan once.
- ✅ Long logs, stack traces, large JSON/API responses.
- ✅ Earlier step outputs that are "done" but might be referenced later.
When NOT to archive
- ❌ The current task instructions or active goal.
- ❌ Small outputs (under ~1500 tokens) — archiving them saves nothing.
- ❌ Anything you are about to use in the very next step.
Check your savings
Call lethe_status() to see how many items you have archived and how many tokens you have
offloaded so far. LETHE never deletes anything — archived content is always recoverable.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 7d ago First seen · 37 lines · 32 tokens per session scan A 682eb613ccc3
lethe-offload is a skill published in the GitHub repository JesusGarcia9009/lethe-engineering (0 stars, last pushed 1mo ago), licensed Unlicense. It adds 32 tokens to every session and 367 once invoked, about $0.0002 per session on Opus 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.
Other skills, from other repositories
cco-patterns
Share learned file patterns across a team — export an anonymized digest of what's usually waste/useful/co-edited, and import a teammate's so a fresh clone benefits day one.
pf-handoff
A Russian-language checkpoint and session-closing procedure that records the verified state of ongoing work in a HANDOFF file. A HANDOFF file is a written task record containing what is done, what remains, blockers, statuses, and journal entries.
pf-resume
A session-resume skill that restores work from a HANDOFF file, task packet, and recent journal entries after a chat ends or is cleared.
memorybox
Diagnoses bloated MEMORY.md files, splits oversized sections into domain-specific files, archives stale daily logs, and deduplicates content using a 3-tier hierarchy (MEMORY.md → domains/ → archive/). Use when MEMORY.md exceeds 10KB, context pressure is high, or the user asks to clean up, organize, or maintain agent…
relay
Capture the full context of the CURRENT Claude chat into a complete, portable briefing the user can paste into a NEW chat to continue seamlessly. Use ONLY when the user explicitly asks — e.g. "relay", "relay this", "handover", "hand off", "handoff", "continue in a new chat", "this chat is getting long", "carry this…
context-optimizer
Reduce Claude AI token consumption by 5x-27x using prompt-native workflows and structural code manifests. Forces Claude to reason from CONTEXTMANIFEST.md first, fetch max 3 files per turn, and output in a strict compressed format. Use when user says optimize context, reduce tokens, context optimizer, manifest, blast…