jthingelstad/mb / skills/mb-cli
mb-cli
VerifiedUse this skill whenever you are using the mb CLI to read, post, manage follows, or build pipelines against micro.blog. It covers command selection, output modes, public boundaries, and cheap-versus-expensive mb workflows.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 6.5 KB |
SKILL.md
View on GitHub---
name: mb-cli
description: Use this skill whenever you are using the mb CLI to read, post, manage follows, or build pipelines against micro.blog. It covers command selection, output modes, public boundaries, and cheap-versus-expensive mb workflows.
---
# mb CLI
Use this skill whenever the task involves `mb`.
## Core rules
- Start most agent sessions with `mb heartbeat`.
- Use `mb inbox` when deciding whether something deserves a reply.
- Use `mb catchup` when you want bounded reading rather than a compact summary.
- Default to `agent` output for exploration and pipelines.
- Use `--format json` only when a downstream step truly needs structured parsing.
- Prefer cheap list/read commands first.
- Make expensive fan-out explicit with `mb lookup ...`.
- Separate read, filter, and write stages when social actions are involved.
## Cheap first, expensive second
Cheap reads:
```bash
mb heartbeat
mb inbox
mb catchup
mb whoami
mb timeline --count 10
mb user following
mb user discover
mb discover --list
mb discover --collection books
```
Explicit enrichment:
```bash
mb user following | mb lookup users --last-post
mb user following | mb lookup users --days-since-posting
```
## Heartbeat workflow
- `mb heartbeat` is the default session-start check for agent use.
- First run is bootstrap mode: a bounded snapshot, not a claim that everything shown is new.
- Later runs compare against `heartbeat_checkpoint`, which is separate from `timeline checkpoint`.
- Heartbeat advances by default; use `mb heartbeat --no-advance` to inspect without saving.
- Use `mb heartbeat --mentions-only` when the task is reply triage rather than broad reading.
- Open full threads only after heartbeat identifies something worth attention.
## Inbox and catchup
- `mb inbox` is the reply-triage surface.
- `mb catchup` is the bounded read-what-is-new surface.
- `mb inbox`, `mb catchup`, and `mb heartbeat` each use separate checkpoints.
- Use `mb checkpoint list` when cursor state is unclear.
- Use `mb checkpoint clear <name>` to reset a stuck workflow cursor.
- Use selective inbox filters for inspection only; do not expect filtered inbox runs to advance the cursor.
- Truncated inbox/catchup results cannot advance; increase the CLI count or use MCP paging plus explicit acknowledgement.
- Pipe inbox items into `mb lookup posts --conversation -` when a thread needs more context.
## Public boundaries
- `mb post ...` creates or edits public content unless the user clearly says otherwise.
- Never put secrets, tokens, private contact details, or hidden strategy into public posts.
## Safe social workflow
Show candidates before acting:
```bash
mb user following | mb lookup users --days-since-posting
mb discover --collection books
mb user discover
```
Act only after the intent is clear:
```bash
... | mb unfollow -
... | mb follow -
```
## Useful command patterns
Session start:
```bash
mb heartbeat
mb heartbeat --no-advance
mb heartbeat --mentions-only
mb heartbeat --count 3 --mention-count 3
mb inbox
mb inbox --reason thread-reply
mb inbox --fresh-hours 24
mb inbox --all
mb inbox --advance
mb catchup
mb catchup --advance
mb checkpoint list
mb checkpoint clear inbox
```
Identity and config:
```bash
mb whoami
mb profiles
mb blogs
mb doctor --offline
```
Posting:
```bash
mb post new "Text" --operation-id example-1
mb post short "Short text" --operation-id example-2
mb post new --content "Text" --operation-id example-3
mb post new --file post.md --operation-id example-4
mb post short --strict-300 "Short text" --operation-id example-5
mb post new --dry-run "Text"
mb media upload image.jpg --alt "Description" --operation-id example-6
mb post new "Text" --photo-url https://cdn.micro.blog/... --alt "Description" --operation-id example-7
mb post edit <id-or-url> --content "Updated" --operation-id example-8
mb post reply <id-or-url> "Reply text" --operation-id example-9
mb post publish <url> --source-hash HASH --operation-id example-10
```
Reading:
```bash
mb timeline
mb timeline mentions
mb conversation <id>
mb blog posts
mb blog search "query"
mb discover --list
mb discover --collection books
mb user show <username>
```
Lookup and follow management:
```bash
mb user following | mb lookup users --last-post
mb user following | mb lookup users --days-since-posting
mb lookup posts --post 12345
mb lookup posts --conversation 12345
mb inbox | mb lookup posts --conversation -
mb follow <username|->
mb unfollow <username|->
```
## Write requirements
- Persist the intended action's `--operation-id` and exact arguments before invoking a write. Reuse both on retries; never generate a fresh ID after an uncertain outcome. Ordinary human CLI use may omit the ID (one is generated and saved per invocation), so a plain rerun is a new write.
- Use one persistent state file (`--state-file`) shared with any MCP client on the same account.
- After `outcome=unknown`, run the printed `mb operation-status ...` command (or `mb operation-status --latest`) and read back from micro.blog. Do not resend. Recording the verified outcome with `mb operation-status ID --resolve applied|not_applied` is the human's decision; propose it, don't run it on your own judgment.
- `mb doctor` lists pending and unknown receipts with resolve hints.
- Edit, delete and publish need an exact owned URL or numeric ID. To publish a draft: `mb post get URL --format json`, review the source, then `mb post publish URL --source-hash HASH --operation-id ID`. Publishing moves the post to a new URL; use the result's `url` afterwards (`draft_url` is the retired one).
## Images
`mb media upload FILE --alt TEXT` uploads a local JPEG, PNG, GIF or WebP (up to 20 MiB) exactly as provided. It does not strip EXIF or GPS metadata; check with the user before uploading a photo that may carry location data. `mb media preview FILE --alt TEXT` shows type, size and sha256 without uploading; pass that `--sha256` to upload to refuse a changed file. HTTP 202 means accepted but possibly still processing. Attach the returned URL with `mb post new ... --photo-url URL --alt TEXT` under a different operation ID. Never repeat an upload merely because creating the post failed. Combined `--photo` and remote-URL uploads are removed.
## Other reads
`mb post replies`, `mb user show NAME --count N`, `mb discover --count N` and `mb conversation URL` are account reads. `mb blog posts/search/categories` are selected-blog reads over a bounded window, not a complete category audit. `mb user mute WORD --keyword` mutes a keyword. Metadata such as `coverage=` and `mb lookup users` errors go to stderr; stdout carries only records.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |