Bm Checkpoint logo

Bm Checkpoint

OrganizationPopular
basicmachines-co
bm-checkpoint

Save a deliberate work checkpoint to Basic Memory with the story, changed files, verification, decisions, blockers, and the next action. Use when the user asks to checkpoint, wrap up, hand off, or remember the state of the work.

Overview

Publisherbasicmachines-co
Repositorybasic-memory
Skill namebm-checkpoint
Stars
4K
Forks
283
Bundled files
Instructions only
LicenseAGPL-3.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by basicmachines-co on GitHub. Read the source before you install it.

Installation

Install the Bm Checkpoint AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/basicmachines-co/basic-memory.git /tmp/basic-memory
mkdir -p .claude/skills
cp -r /tmp/basic-memory/plugins/claude-code/skills/bm-checkpoint .claude/skills/bm-checkpoint
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Bm Checkpoint in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Bm Checkpoint on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Bm Checkpoint is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Checkpoint Claude Work

Create a durable handoff note for the current work. Use this when the user asks to checkpoint, wrap up, hand off, remember the state of the work, or before a long context transition. This is the deliberate, high-signal counterpart to the automatic PreCompact checkpoint.

Gather

Resolve config: read the basicMemory block with the same precedence the hooks use. For the user-level base, append settings.json to the literal value of CLAUDE_CONFIG_DIR when that environment variable is present; do not trim it, expand ~, or treat an empty value as unset. Use ~/.claude/settings.json only when the variable is absent. Then the project's .claude/settings.json and .claude/settings.local.json override it per key:

  • primaryProject, default omitted (Basic Memory's default project)
  • captureFolder, default sessions
  • placementConventions, optional
  • sessionProfile, default general
  • repository, required when sessionProfile is coding

Apply the bm-writing skill before drafting the note.

Gather evidence:

  • the original problem or goal and why it mattered
  • the approach taken and why it solves the problem
  • the current system state and practical impact
  • tradeoffs, sharp edges, useful simplifications, and intentionally parked work
  • the durable lesson, if one exists — what future work should know or avoid
  • git status --short
  • current branch
  • repository root and current working directory
  • current Git SHA
  • current pull request number, title, URL, state, base, and head when one exists
  • changed files you touched
  • tests or checks actually run
  • failures or skipped checks
  • decisions made in this session
  • unresolved blockers
  • next action
  • current username, hostname, and timestamp
  • exact host-provided session identifier, when available; never infer one

Do not claim a test passed unless you ran it or the user supplied the result.

Write

A checkpoint is a durable handoff, not a status dump or commit-by-commit changelog. Tell the story for a human or agent returning later.

Write the note with write_note, routed to primaryProject (pass it as project, or as project_id if it's an external_id UUID). For the general profile:

  • title: Claude checkpoint - <short topic>
  • directory: configured captureFolder
  • tags: ["claude", "checkpoint"]
  • note_type: session
  • metadata (frontmatter):
    • status: open
    • project: <primaryProject if known>
    • cwd: <current cwd>
    • started: <current timestamp>
    • username: <current username>
    • hostname: <current hostname>
    • capture: deliberate
    • agent: claude-code
    • session_id: <exact host-provided session id>, when available

New notes use agent and session_id, not claude_session_id. Existing notes with the legacy field remain valid; do not rewrite them. Session identity is the pair of agent and session ID, never a bare ID shared across hosts.

For the coding profile, write note_type: coding_session (frontmatter type: coding_session) and use the same common frontmatter plus these schema-required fields:

  • repository: <confirmed stable repository identifier>
  • repo_root: <git rev-parse --show-toplevel>
  • cwd: <current cwd>
  • branch: <git rev-parse --abbrev-ref HEAD>
  • git_sha: <git rev-parse HEAD>

When the current branch has a pull request, also add the typed optional fields pull_request_number, pull_request_title, pull_request_url, lowercase pull_request_state, pull_request_base, and pull_request_head. Resolve the pull request with a read-only GitHub query (e.g. gh pr view --json ...); omit those fields when no PR exists. Write the number as a quoted string, for example pull_request_number: "123", so exact metadata queries behave consistently across storage backends. Never infer or copy repository/PR identity only from conversation text. Stop if the required coding fields cannot be proven.

Begin the body with # <exact note title>.

Use these sections, omitting optional ones that add no value:

  • ## Summary: one concrete sentence that does not merely repeat the title
  • ## Story: problem -> approach -> current state and impact in substantive prose
  • ## Project Memory, when the work surfaced a durable lesson future readers need — the constraint discovered, the boundary made explicit, the shortcut to avoid
  • ## Changed Files, when paths are useful for resuming
  • ## Verification, for checks actually run and their outcomes
  • ## Observations
  • ## Relations, when the session has an obvious graph target

Use observations to distill durable facts for structured recall rather than duplicating every narrative sentence:

  • [result] for concrete outcomes
  • [decision] for each decision made or preserved
  • [blocker] for each unresolved blocker
  • [next_step] for the next concrete action; include at least one
  • [verification] or [changed_file] only when the item is itself important project memory, not merely supporting detail

Do not create separate Decisions, Blockers, or Next Action sections with plain bullets. Omit empty categories instead of writing placeholder text such as "None."

Relations are not observations. Put them under ## Relations using Basic Memory relation syntax, for example - relates_to [[Exact existing note title]]. Never write [relates_to] or a bare memory:// URL as an observation. Only add a relation when its target is an existing task, decision, spec, issue, or PR note.

Confirm

Reply with the permalink and the one next action the checkpoint preserves.

Frequently asked questions

What does the Bm Checkpoint AI skill do?

Save a deliberate work checkpoint to Basic Memory with the story, changed files, verification, decisions, blockers, and the next action. Use when the user asks to checkpoint, wrap up, hand off, or remember the state of the work.

Why use Bm Checkpoint on TypingMind?

Because you install it once and use it with any model. Bm Checkpoint is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Bm Checkpoint in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/basicmachines-co/basic-memory/tree/main/plugins/claude-code/skills/bm-checkpoint. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Bm Checkpoint?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Bm Checkpoint?

As many as you like. As long as a model supports skills, you can use Bm Checkpoint with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Bm Checkpoint AI skill free?

Yes. It is published on GitHub by basicmachines-co under the AGPL-3.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇