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/equwal-memstate-precompact
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/memstate-precompact
What's inside
- Skill name
memstate-precompact- Description
- Run right before /compact. Saves this session's working state to memstate, then prints the /compact instruction line to paste, so the next context window starts from memstate instead of a lossy summary.
- License
- MIT
- Instructions
- 46 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Before a compact A compact replaces this context with a summary. What the next context window needs and memstate does not hold is lost. Do the two steps below in order, then stop. Do not start new work after step 2. ## 1. Write the session state to memstate Target: the session's project (the cwd project, or the pinned project). Use `memstate_remember` for prose and `memstate_set` for one fact. Write only what the code and the git history cannot answer. Write these keypaths. Skip one only when it has nothing new. | Keypath | Content | |---|---| | `task.summary.<YYYY_MM_DD>` | One paragraph: the task as the user stated it, what is done with its evidence (tests run, commits made), what is not done, and the exact next step. | | `decisions.<topic>` | One keypath per choice made this session, with the reason. | | `gotchas.<topic>` | One keypath per trap found: what failed, the cause, the fix. | | `todo` | Unfinished items, the ones the user asked for and the ones you found. | | `questions` | Open questions for the user, with the context an answer needs. | | `notes.<topic>` | Facts you had to look up and would need again: how a subsystem works, the command that verified a step, a URL. | Rules: - Refer to code by symbol name, never by file and line. - Update an existing keypath. Do not create a sibling for a new value of the same fact. The write response names the superseded version. - State that holds only on an unmerged branch goes under `branches.<branch_slug>.*`. - Never save a denied prompt. Never save a secret. - A fact about the user or this machine goes to `scope="user"` only when it holds in every repository. - Do not read anything back. The write response is the receipt. ## 2. Print the compact instructions Print the line below, filled in, as the last thing in your reply, inside one fenced block. The user pastes it as the argument of `/compact`. Keep it under 120 words. Name the keypaths exactly as you wrote them. ``` /compact Focus on: project <id>; task "<one line>"; done so far: <one line>; next step: <one line>; open questions: <list or none>. The session state is in memstate under <keypath list>. In the new context, before any other work, call memstate_get on each of those keypaths. Drop tool output, file contents and reasoning from the summary. Keep only what this line says. ```
From equwal/ai-config, MIT.
More from equwal/ai-config
- 0.570score
agent-creator
equwal
Create production-grade agent .md files aligned with the Anthropic 2026 spec (16-field schema).
SkillAFreshness: AClaude Code +9 - 0.567score
memstate
equwal
Persistent, versioned memory via the memstated daemon.
SkillAFreshness: AClaude Code +9 - 0.565score
impactful-writing
equwal
Write clear, emotionally resonant, and well-structured content that readers remember and act upon.
SkillAFreshness: AClaude Code +9 - 0.559score
dynamic-analysis
equwal
Find memory errors, undefined behaviour, data races, and parser crashes using sanitizers (ASan, UBSan, TSan, MSan, Miri) and coverage-guided fuzzing…
SkillAFreshness: AClaude Code +9 - 0.559score
regression-bisect
equwal
Find the exact commit that introduced a regression using git bisect.
SkillAFreshness: AClaude Code +9 - 0.559score
api-contract-testing
equwal
Property-based testing of HTTP APIs against an OpenAPI or GraphQL specification using Schemathesis.
SkillAFreshness: AClaude Code +9 - 0.559score
deploy-handoff
equwal
Gives the last step of a deploy, billing, or sign-in task to the user.
SkillAFreshness: AClaude Code +9 - 0.559score
agent-manager
equwal
Run a fleet of AI agents (Claude Code, Codex, Copilot, Antigravity) toward the user's tasks.
SkillAFreshness: AClaude Code +9 - 0.559score
call-transcript
equwal
Read the transcripts of the user's Discord voice calls.
SkillAFreshness: AClaude Code +9 - 0.559score
docs-search
grounded.tools
Search and query the Grounded Docs MCP Server documentation index.
SkillAFreshness: AClaude Code +9 - 0.555score
docs-manage
grounded.tools
Manage the Grounded Docs MCP Server documentation index.
SkillAFreshness: AClaude Code +9 - 0.555score
memstate-upgrade
equwal
Upgrade jose's memstate install (memstated daemon, memstate CLI, memstate-mcp proxy, skills) on the PC, desktop g and the honjimaku daemon to the latest…
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 11, 2026 (17 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/equwal/ai-config
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from equwal
- 0.570score
agent-creator
equwal
Create production-grade agent .md files aligned with the Anthropic 2026 spec (16-field schema).
SkillAFreshness: AClaude Code +9 - 0.567score
memstate
equwal
Persistent, versioned memory via the memstated daemon.
SkillAFreshness: AClaude Code +9 - 0.565score
impactful-writing
equwal
Write clear, emotionally resonant, and well-structured content that readers remember and act upon.
SkillAFreshness: AClaude Code +9 - 0.559score
agent-manager
equwal
Run a fleet of AI agents (Claude Code, Codex, Copilot, Antigravity) toward the user's tasks.
SkillAFreshness: AClaude Code +9 - 0.559score
regression-bisect
equwal
Find the exact commit that introduced a regression using git bisect.
SkillAFreshness: AClaude Code +9 - 0.559score
call-transcript
equwal
Read the transcripts of the user's Discord voice calls.
SkillAFreshness: AClaude Code +9
Cite this page
memstate-precompact skill. Agent Mart. https://agentmart.dev/skills/equwal-memstate-precompact (last updated 2026-10-24)