How do I install the Pre Compact skill?

Run npx skillmds@latest add tenequm/pre-compact in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Pre Compact safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Pre Compact?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Pre Compact free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Pre Compact?

tenequm (@tenequm) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Pre Compact by tenequm — Agent Skill

Name: Pre Compact
Rating: 4.5 (4 reviews)
Author: tenequm

Pre Compact

Prepare the session for context compaction - write a handoff file a fresh session can continue from, propose updates to the project's durable docs, apply them on approval or with `auto`. Use before compacting or clearing context, or on "pre-compact".

tenequm 138bd89 2 files · 11.3 KB Updated Sep 10, 2026

File contents

I will compact this context soon. The summary can be wrong. Files are the source of truth. Move everything a fresh session needs from this conversation into files. Mode: $ARGUMENTS. auto or a means: apply all proposed updates without asking.

Write .agents/compact-handoff/YYMM-DD-HHMM-<short-slug>.md. Start with the line # Handoff <date> <time> <repo>@<branch>. Then write these sections:

Position: what we are doing, why, and where we stopped.
Next steps: in order, with the paths, commands and identifiers needed to run them.
Decisions: each decision made here, with its reason.
Findings: what we learned works or fails, verbatim.
Open questions: numbered, each with your recommended answer.
Undone instructions: what I asked for that is not done.
References: files, links, related docs.

Copy numbers, commands and paths exactly. Do not paraphrase them. Never write secrets, tokens, passwords or API keys into the handoff. Write where they live instead, for example the env var name or the vault item. Record only decisions that were made. Keep open questions open.

Then check the docs this project uses to track work: plans, journals, status sections, task lists. For each item in the handoff, ask: is it recorded there, and is it current? Do not ask me what the repo can answer.

Also check my global agent instructions (~/.claude/CLAUDE.md, AGENTS.md, or the equivalent for this harness) for the marker . If the marker is absent, add this item: insert the line below verbatim.

After compaction, read the handoff file named in the summary (.agents/compact-handoff/) before you trust the summary. The summary can be wrong. The handoff file is the source of truth.

List what is missing or stale as numbered items. Give the file and the change, one line each. Then ask: y = apply all, n = apply none, numbers = apply those, or tell me what to change. Wait for my answer, unless in auto mode. Never commit. Never push.

Finish with one line: READY, the handoff path, and the files you changed.

---name: pre-compactdescription: Prepare the session for context compaction - write a handoff file a fresh session can continue from, propose updates to the project's durable docs, apply them on approval or with `auto`. Use before compacting or clearing context, or on "pre-compact".---I will compact this context soon. The summary can be wrong. Files are thesource of truth. Move everything a fresh session needs from this conversationinto files. Mode: $ARGUMENTS. `auto` or `a` means: apply all proposed updateswithout asking.Write `.agents/compact-handoff/YYMM-DD-HHMM-<short-slug>.md`. Start with theline `# Handoff <date> <time> <repo>@<branch>`. Then write these sections:- Position: what we are doing, why, and where we stopped.- Next steps: in order, with the paths, commands and identifiers needed to run them.- Decisions: each decision made here, with its reason.- Findings: what we learned works or fails, verbatim.- Open questions: numbered, each with your recommended answer.- Undone instructions: what I asked for that is not done.- References: files, links, related docs.Copy numbers, commands and paths exactly. Do not paraphrase them. Never writesecrets, tokens, passwords or API keys into the handoff. Write where theylive instead, for example the env var name or the vault item. Record onlydecisions that were made. Keep open questions open.Then check the docs this project uses to track work: plans, journals, statussections, task lists. For each item in the handoff, ask: is it recorded there,and is it current? Do not ask me what the repo can answer.Also check my global agent instructions (`~/.claude/CLAUDE.md`, `AGENTS.md`,or the equivalent for this harness) for the marker ``. Ifthe marker is absent, add this item: insert the line below verbatim.- After compaction, read the handoff file named in the summary (`.agents/compact-handoff/`) before you trust the summary. The summary can be wrong. The handoff file is the source of truth. List what is missing or stale as numbered items. Give the file and the change,one line each. Then ask: y = apply all, n = apply none, numbers = apply those,or tell me what to change. Wait for my answer, unless in auto mode. Nevercommit. Never push.Finish with one line: READY, the handoff path, and the files you changed.

Frequently asked questions