checkpoint-save
Saves session progress to a state file, naming it automatically from the session's work.
Skill by rm0nroe
Chalk Labs
Need help deploying checkpoint-save?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
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/rm0nroe-checkpoint-save
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
Use cases: productivity
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/checkpoint-save
What's inside
- Skill name
checkpoint-save- Description
- Saves session progress to a state file, naming it automatically from the session's work. Usage: /checkpoint-save [optional_name]
- License
- MIT
- Instructions
- 40 lines in SKILL.md · view on GitHub
Agent Skills spec issues
- body: contains an absolute path from the author's machine
Files (1)
- SKILL.md
SKILL.md (first lines)
# Save Session State
1. **Determine the project**: the basename of the git repository root (`git rev-parse --show-toplevel`), or of the working directory outside a repo, so subfolders of one repo share state (e.g. working in `/Users/me/projects/my-app/src` → project is `my-app`). Two repos with the same folder name share one cabinet. State files live in `~/.claude_states/<project>/`.
2. **Determine the `<session_name>`. Do NOT ask for one; derive it.**
- If a name was passed after `/checkpoint-save`, use it verbatim (slugified: lowercase, kebab-case).
- Otherwise generate one, matching **this project's existing convention**, which you must observe rather than assume: run `ls ~/.claude_states/<project>/ | tail -30` and copy the dominant shape.
- Absent any existing files, default to `<topic-slug>-MM-DD-YY`.
- Rules for the topic slug:
- **Lead with the identifier the work is tracked under** if one exists: a ticket ID (`auth142`, `ops58`), else the subsystem (`billing`, `catalog`, `dashboard`). That prefix is what makes a checkpoint findable months later.
- Follow with **2–4 words naming the outcome, not the activity**: `shipped`, `closed`, `live`, `retired`, `half-done`, `plan-ready`. `auth142-shipped` beats `working-on-auth142`.
- Prefer the state the session *ended* in. A session that shipped one ticket and left a migration half-applied is `auth142-shipped-migration-half-done`, not `auth142`.
- Never use content-free names: `before-we-start`, `checkpoint`, `session`, `wip`, `latest`, `update`, `notes`.
- Keep the whole filename under ~50 chars.
- Use today's date in the project's existing date format (`MM-DD-YY` if that dominates, `YYYY-MM-DD` if that does).
- **Collision**: if that exact file exists and belongs to a *different* session, append a short distinguishing word rather than overwriting (`auth142-shipped-2-07-30-26`). Re-checkpointing the *same* session should overwrite, and should note in the body which earlier checkpoints it supersedes.
- The `sendit-` prefix is reserved for runtime views; reject it for manual checkpoint names. Do not overwrite or migrate existing files with that prefix.
- State the chosen name in the reply so it can be corrected.
3. **Review all actions, file changes, and decisions made during this session.**
4. Create or overwrite `~/.claude_states/<project>/<session_name>.md` with a highly compressed summary. First line is `# Checkpoint: <session_name>`. Sections:
- **Domain/Goal:** what this specific session is responsible for.
- **Working Directory:** the full path to the project.
- **Completed Steps:** what was just finished in this exact session.
- **Key Files Modified:** files changed or created.
- **Open Blockers:** unresolved errors or logic gaps.
- **Next Actions:** exact instructions for the next step, as a numbered list with one action per line. Keep the action itself on that line and put any detail it needs on indented sub-lines beneath it: `/checkpoint-load` echoes these back as single-line sub-bullets, so an action whose first line runs into rationale reads as noise there. When the session states the next actions, keep exactly that list in that order; add no steps of your own.
**Lead with any state that is unsafe to forget** (halted daemons, disabled flags, live production state, open migrations, money or data at risk) in its own section above Completed Steps, including the exact command to reverse it. A checkpoint's worst failure mode is a fresh session resuming without knowing the system was left in a non-default state.
**Record what was measured, not only what was concluded**, and explicitly flag any conclusion a later step invalidated. Numbers and falsified assumptions are exactly the parts a fresh context cannot re-derive.
**Active controller-owned Send It task:** preserve its runtime authority. Follow `/checkpoint-load`'s sealed-bridge validation procedure for the exact `sendit-<run-uuid>` using the recorded platform's installed runtime. Only after successful authoritative JSON readback, reference its run, platform, snapshot sequence and sealed hash in this manual document, preserving all nine semantic fields and retained limits. Save these notes under an ordinary name; the seal remains authoritative and must be revalidated on a later sealed load. Do not overwrite the Git-local generated view, invent a replacement binding, or use operator `recover` to reset its limits. If no valid seal exists or validation fails, label this manual document as session notes, record the missing seal or validation error, and do not present it as an automatic worker recovery receipt; the bound controller must save/reconcile its current phase before automatic continuation. Saving or reading notes does not rebind, reset limits or authorize a restart.
5. Create the directory if it does not already exist.
6. Once the file is written, remind me to run `/clear` to free up the terminal context window.From rm0nroe/loadout, MIT.
More from rm0nroe/loadout
- 0.570score
clip
rm0nroe
macOS only.
SkillAFreshness: AClaude Code +9 - 0.570score
relay
rm0nroe
Write a message addressed to a specific person or team who was NOT part of this conversation, which the user will hand off, paste, forward, or read aloud to…
SkillAFreshness: AClaude Code +9 - 0.570score
write
rm0nroe
Drafts Slack messages, emails, GitHub PR titles and descriptions, PR review comments and replies, and git commit messages, release notes, incident…
SkillAFreshness: AClaude Code +9 - 0.566score
readme-banner
rm0nroe
Design a memorable hero banner for a repository README as one self-contained SVG, with the wordmark converted to font outlines so it renders identically on…
SkillAFreshness: AClaude Code +9 - 0.565score
simplify
rm0nroe
Rewrites the last assistant reply into a short everyday analog (one metaphor, numbered roles, mix-up named, one-sentence rule).
SkillAFreshness: AClaude Code +9 - 0.565score
docs-prune
rm0nroe
Analyze a repo's documentation corpus and prune/exclude stale docs so knowledge-graph and vault tooling (graphify, Obsidian sync) index only live, high-signal…
SkillAFreshness: AClaude Code +9 - 0.565score
break-it-down
rm0nroe
Rewrites the last assistant reply as a plain, short, direct explanation: no metaphors.
SkillAFreshness: AClaude Code +9 - 0.565score
exec-recap
rm0nroe
Summarizes what shipped in a version range, date range or commit range as a plain-language TLDR for a non-technical executive who has never seen the project.
SkillAFreshness: AClaude Code +9 - 0.561score
roundtable
rm0nroe
Run a multi-perspective roundtable on a decision: 4 parallel background agents argue it against the actual code with file:line evidence, then their positions…
SkillAFreshness: AClaude Code +9 - 0.553score
checkpoint-load
rm0nroe
Instantly loads a specific session checkpoint.
SkillAFreshness: AClaude Code +9 - 0.553score
checkpoint-list
rm0nroe
Lists all saved checkpoints for the current project.
SkillAFreshness: AClaude Code +9 - 0.539score
slop-radar
rm0nroe
Use when the user asks to audit, critique, or detect AI slop, vibe-coded sameness, generic template aesthetics, or unmotivated frontend design choices in a…
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 20, 2026 (10 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/rm0nroe/loadout
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from rm0nroe
- 0.597score
Catalyst Edge Research
rm0nroe
Source-linked market intelligence for AI agents researching public-company tickers.
ConnectorAFreshness: ALocalRead-only↓ 47/wkClaude Code +5 - 0.570score
clip
rm0nroe
macOS only.
SkillAFreshness: AClaude Code +9 - 0.570score
relay
rm0nroe
Write a message addressed to a specific person or team who was NOT part of this conversation, which the user will hand off, paste, forward, or read aloud to…
SkillAFreshness: AClaude Code +9 - 0.570score
write
rm0nroe
Drafts Slack messages, emails, GitHub PR titles and descriptions, PR review comments and replies, and git commit messages, release notes, incident…
SkillAFreshness: AClaude Code +9 - 0.566score
readme-banner
rm0nroe
Design a memorable hero banner for a repository README as one self-contained SVG, with the wordmark converted to font outlines so it renders identically on…
SkillAFreshness: AClaude Code +9 - 0.565score
docs-prune
rm0nroe
Analyze a repo's documentation corpus and prune/exclude stale docs so knowledge-graph and vault tooling (graphify, Obsidian sync) index only live, high-signal…
SkillAFreshness: AClaude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get checkpoint-save running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
checkpoint-save skill. Agent Mart. https://agentmart.dev/skills/rm0nroe-checkpoint-save (last updated 2026-10-24)