ch-compacting-conversations
Use when the user explicitly asks to compact, condense, summarize-and-replace, or archive old conversation history in the current worktree.
Skill by chauipp
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/chauipp-ch-compacting-conversations
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/ch-compacting-conversations
What's inside
- Skill name
ch-compacting-conversations- Description
- Use when the user explicitly asks to compact, condense, summarize-and-replace, or archive old conversation history in the current worktree.
- License
- MIT
- Instructions
- 44 lines in SKILL.md · view on GitHub
Files (2)
- SKILL.md
- openai.yaml · agents/
SKILL.md (first lines)
# Compacting Conversations ## Overview Compact preserves useful project memory while replacing an obsolete **continuous** run of raw chat files with one concise summary. A direct user instruction to compact authorizes this exception to verbatim conversation storage. ## Required procedure 1. Work only in `conversation/` of the current worktree. Do not create a nested conversation folder. 2. Select one or more **separate continuous ranges**. Never combine scattered files merely because they share a topic. 3. Read every file in each candidate range. Keep a range intact if any active requirement, decision, blocker, workflow fact, or uncertainty is not recorded in the GDD or a conversation outside the range. 4. Create one replacement file for each eligible range: `<first-timestamp>-compact.md`. 5. State the first and last replaced filenames; summarize the original objective, durable decisions, superseded directions, and the current source of truth. Do not invent missing context. 6. Only after the compact file exists, delete exactly the source files in that range. Preserve files immediately before and after it. 7. Verify the compact files exist, every selected source is absent, unselected files remain, the count is correct, and Markdown has no whitespace errors. ## Authorization and logging - Compact only after an explicit action request such as “compact conversation”, “tóm tắt và thay thế chat cũ”, or an unambiguous follow-up instruction. “Có compact được không?” is a question, not authorization. - Do not create normal conversation logs for the compact trigger, interim updates, or cleanup result. - A compact file is the authorized exception to the rule that a conversation file contains one verbatim user/agent pair. ## Non-negotiable boundaries - Do not compact by topic across non-contiguous time ranges. - Do not delete first and plan to summarize later. - Do not use GDD as an excuse to skip reading the source range. - If uncertain whether a file matters, retain it. - Do not modify the GDD while compacting unless the user separately asks. ## Common rationalizations | Rationalization | Required response | |---|---| | “Raw conversation must stay verbatim.” | An explicit compact request authorizes the replacement compact file. | | “Grouping all old time-related files is faster.” | Only continuous ranges preserve a readable timeline. | | “The GDD has it, so source review is unnecessary.” | Review every source; GDD may omit workflow facts or rationale. | | “The user asked whether it is possible.” | Explain the method and wait for an action request. | ## Final report Report the number of source files replaced, replacement files created, retained file count, and verification result. Do not recreate a conversation log for this report.
From chauipp/dotagents, MIT.
More from chauipp/dotagents
- 0.570score
ch-writing-prompts-map
chauipp
Use when the user explicitly invokes ch-writing-prompts-map with an optional level to create or improve a prompt for building a game map, level, or environment.
SkillAFreshness: AClaude Code +9 - 0.570score
ch-learn-english
chauipp
Use when the user explicitly invokes $ch-learn-english on, off, status or sync, or when a trusted per-chat hook confirms English learning is enabled in the…
SkillAFreshness: AClaude Code +9 - 0.565score
ch-build-map
chauipp
Use when building or editing an Unreal Engine room, sector, or level from a detailed spatial specification, especially with UE/Blender assets, shared .umap…
SkillAFreshness: AClaude Code +9 - 0.561score
ch-verifying-ui-with-playwright
chauipp
Dùng khi một task có đụng tới thứ người dùng nhìn thấy trên trình duyệt — component, trang, CSS, form, luồng bấm — TRƯỚC khi báo là xong.
SkillAFreshness: AClaude Code +9 - 0.561score
ch-research
chauipp
Research open-ended or time-sensitive questions with multiple GPT-6 Luna subagents, independent sources, cross-checking, adversarial review, and cited…
SkillAFreshness: AClaude Code +9 - 0.559score
ch-no-clarify
chauipp
Use when the user explicitly invokes $ch-no-clarify and wants the current request executed directly without explanatory commentary.
SkillAFreshness: AClaude Code +9 - 0.559score
ch-clear-conversation
chauipp
Use when the user explicitly invokes $ch-clear-conversation to remove obsolete conversation log entries from the current worktree.
SkillAFreshness: AClaude Code +9 - 0.559score
ch-compact-conversation
chauipp
Use when the user explicitly invokes $ch-compact-conversation to condense a contiguous range of conversation logs in the current worktree.
SkillAFreshness: AClaude Code +9 - 0.557score
ch-capturing-what-worked
chauipp
Dùng ngay sau khi một task đã làm xong VÀ đã kiểm chứng là đúng, để đề xuất lưu bài học cho lần sau và chờ người dùng xác nhận trước khi ghi.
SkillAFreshness: AClaude Code +9 - 0.544score
ch-updating-skill-models
chauipp
Use when the user explicitly asks to reassess model or reasoning assignments in a Codex skill, or to apply a previously approved model update proposal.
SkillAFreshness: AClaude Code +9 - 0.539score
ch-research
chauipp
Use when a user asks to investigate, compare, verify, or evaluate a question and needs evidence from multiple sources, cross-checking, or a…
SkillAFreshness: AClaude Code +9 - 0.521score
graphify
chauipp
Build and query a knowledge graph of a codebase or document corpus with graphify.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 28, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/chauipp/dotagents
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from chauipp
- 0.570score
ch-learn-english
chauipp
Use when the user explicitly invokes $ch-learn-english on, off, status or sync, or when a trusted per-chat hook confirms English learning is enabled in the…
SkillAFreshness: AClaude Code +9 - 0.570score
ch-writing-prompts-map
chauipp
Use when the user explicitly invokes ch-writing-prompts-map with an optional level to create or improve a prompt for building a game map, level, or environment.
SkillAFreshness: AClaude Code +9 - 0.565score
ch-build-map
chauipp
Use when building or editing an Unreal Engine room, sector, or level from a detailed spatial specification, especially with UE/Blender assets, shared .umap…
SkillAFreshness: AClaude Code +9 - 0.561score
ch-research
chauipp
Research open-ended or time-sensitive questions with multiple GPT-6 Luna subagents, independent sources, cross-checking, adversarial review, and cited…
SkillAFreshness: AClaude Code +9 - 0.561score
ch-verifying-ui-with-playwright
chauipp
Dùng khi một task có đụng tới thứ người dùng nhìn thấy trên trình duyệt — component, trang, CSS, form, luồng bấm — TRƯỚC khi báo là xong.
SkillAFreshness: AClaude Code +9 - 0.559score
ch-no-clarify
chauipp
Use when the user explicitly invokes $ch-no-clarify and wants the current request executed directly without explanatory commentary.
SkillAFreshness: AClaude Code +9
Cite this page
ch-compacting-conversations skill. Agent Mart. https://agentmart.dev/skills/chauipp-ch-compacting-conversations (last updated 2026-10-24)