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 mocasus/paleo --skill paleo-trim-contextgit clone --depth 1 https://github.com/mocasus/paleoWrote 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/mocasus/paleo/paleo-trim-context)<a href="https://agentmods.dev/skills/mocasus/paleo/paleo-trim-context"><img src="https://agentmods.dev/badge/skills/mocasus/paleo/paleo-trim-context/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/mocasus/paleo/paleo-trim-context"><img src="https://agentmods.dev/badge/skills/mocasus/paleo/paleo-trim-context.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 | $0.00040 | $0.00440 |
| Opus 5 | $0.00020 | $0.00220 |
| Sonnet 5 | $0.00008 | $0.00088 |
| Haiku 4.5 | $0.00004 | $0.00044 |
Grade A, and why
paleo-trim-context 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 11d 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
paleo-trim-context
Keep context small. Drop stale, keep task state.
Rules
- Summarize old tool output into 1-2 lines; keep only the result that matters.
- Drop: resolved errors, obsolete file reads, duplicate logs.
- Keep: current plan, open TODO, key IDs/paths, last decision.
- Use session_search / notes for history instead of re-pasting.
- Prefer code/script over long prose explanations in context.
- Pre-thinking compression (biggest savings): BEFORE the model reasons, compress retrieved docs / RAG context / long file dumps into key facts + a source pointer. Drop full-doc dumps.
- Effort pinning: trivial task -> signal low effort, no over-think. Reuse a prior conclusion instead of re-reasoning the same thing.
Scope boundary (don't collide with siblings)
- Context hygiene + effort pinning + pre-thinking compression = THIS skill (runs automatically during the session).
paleo-converse= explicit conversation-turn compression + dedup (on-demand, when the user points at history or context is heavy).paleo-summary= one bulky artifact (log / diff / dump / stdout) -> tight intisari.- Hand off to those when the target is a specific turn-block or a single artifact; don't re-implement them here.
When
- Long session (>20 turns), big file dumps, repeated similar output.
- Before calling an expensive model — shrink first.
Gotchas
- Don't trim the user's latest instruction.
- Keep ground truth (file paths, IDs) — summarize derivations only.
- If unsure what's stale, ask once. Don't guess-drop critical state.
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.
- 11d ago First seen · 38 lines · 40 tokens per session scan A fb6ba5d9389b
paleo-trim-context is a skill published in the GitHub repository mocasus/paleo (20 stars, last pushed 1mo ago), licensed MIT. It adds 40 tokens to every session and 440 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 2026-08-30.
Other skills, from other repositories
hermes-context-optimization
Use this when the user asks about Hermes prompt/context size, “hello” startup cost, compression behavior, memory/profile bloat, tool-schema overhead, skill loading, session-store/search-index storage, or multimodal/visual-context approaches such as Snapcompact.
hermes-mnemosyne
Mnemosyne is Hermes' primary local-first memory engine — SQLite with vector + FTS5 hybrid search, 19+ tools, auto-consolidation, and a standalone CLI. It's a pip-installed plugin (not a built-in toolset) discovered via $HERMESHOME/plugins/mnemosyne/.
hermes-session-maintenance
Use this skill when the user asks about pruning Hermes sessions, session retention, session DB size, hermes sessions prune, sessions.autoprune, sessions.retentiondays, session export/backup, or whether old conversations should be capped.
memory-init
Scaffold a project-local memory directory and AGENTS.md guidance for reusable agent knowledge. Use when a user wants to initialize .agents/memories/ or another project memory path so future agents can read and maintain durable project context.
mnemosyne-maintenance
Use when: upgrading Mnemosyne, diagnosing slow/hung consolidation (mnemosynesleep), fixing missing embeddings, or troubleshooting import/version mismatches.
obsidian-memory-architecture
Use when designing, setting up, or maintaining an Obsidian vault as Hermes Agent's durable knowledge layer. Routes facts, conversation history, documents, procedures, and daily logs to the correct Hermes or vault system without duplicating everything into the prompt.