Update to Basic Memory Cloud

{
  "description": "Write Basic Memory notes that render as charts, tables, boards, and live views with `::bm-*` Comark components. Use when a note reads better as a figure or board, or already has `::bm-*` directives.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 363
    },
    {
      "relative_path": "references/components.md",
      "size_in_bytes": 33063
    }
  ],
  "name": "memory-comark",
  "skill_md_contents": "---\nname: memory-comark\ndescription: \"Write Basic Memory notes that render as charts, tables, boards, and live views with `::bm-*` Comark components. Use when a note reads better as a figure or board, or already has `::bm-*` directives.\"\n---\n\n# Memory Comark\n\nBasic Memory notes can carry small, safe components written as Comark block\ndirectives (`::bm-bars{...}` … `::`). Basic Memory Cloud draws them in Notes, the\nWiki, and shared pages. Everywhere else (a local editor, GitHub, an agent's raw\nread) the same Markdown still reads as an ordinary list, table, or paragraph, so\na component never hides content.\n\nUse a component when its shape says more than a sentence or a list would. Keep\nprose primary: one figure per idea, at most two per section, with a sentence\nbetween them that says what the figure shows.\n\nThe full vocabulary, every prop, and every data grammar are in\n[references/components.md](references/components.md). Read it before using a\ncomponent you have not used before.\n\n## Pick a component\n\nMost notes need only these. The reference has the rest (39 in total).\n\n| You want to show | Use |\n| --- | --- |\n| One or a few headline numbers | `bm-stat` |\n| A comparison of a few values | `bm-bars` |\n| A short ranked list | `bm-rank` |\n| A small data table | `bm-table` |\n| Progress toward a whole | `bm-meter` |\n| A trend without axes | `bm-spark` |\n| Steps that drop off | `bm-funnel` |\n| A task list | `bm-check` |\n| Work in lanes (now, next, later) | `bm-kanban` |\n| Events in order | `bm-timeline` |\n| A framed claim or summary | `bm-frame` |\n| A short path (`a → b → c`) | `bm-flow` |\n| A real graph, sequence, or state machine | a fenced ```mermaid block (not a component) |\n\nIf nothing fits, write prose or a plain Markdown table. Never invent a\ncomponent: an unknown name such as `::bm-pie` is not drawn and shows as raw text.\n\n## Write a component\n\nEvery component opens with `::bm-name{attributes}` on its own line and closes\nwith `::` on its own line.\n\n```markdown\nThe queue rewrite more than doubled steady-state throughput.\n\n::bm-bars{title=\"Throughput\" caption=\"Before and after the queue rewrite\"}\n- Before: 42/s\n- **After: 96/s**\n::\n```\n\nRules that matter:\n\n- **Data goes in the body as Markdown**, not in an attribute: `- label: value`\n  rows, a table for grids. **Bold** marks the current or chosen row, *italic*\n  recedes, ` — note` adds a side note.\n- **Attribute values are double-quoted and cannot contain `\"`.** There is no\n  escape. Inside a `query`, quote phrases with single quotes:\n  `query=\"search 'queue rewrite'\"`.\n- **Task state stays in Markdown.** Use `- [ ]` and `- [x]` inside `bm-check` and\n  `bm-kanban`; people can tick boxes from the rendered note.\n- **Never put HTML, scripts, styles, or classes in a note**, and never wrap a\n  component in a code fence (a fenced block stays literal code).\n- **Use only registered props.** An unknown attribute produces `prop-unknown`\n  in a warning frame with the original source and registered props; it is never\n  silently dropped.\n\nA checklist and a board:\n\n```markdown\n::bm-check{title=\"Launch\"}\n- [x] Freeze the schema\n- [ ] Write the migration note — due Friday\n::\n\n::bm-kanban{title=\"Roadmap\"}\n### Now\n- **Billing page**\n- [ ] Usage emails\n\n### Next\n- *Partner portal*\n::\n```\n\n## Live views over the knowledge base\n\nA component with a `query` attribute shows current data from the project it\nlives in. Prefer a live view to a hand-copied number when the note describes its\nown knowledge base, because copied counts go stale.\n\n```markdown\n::bm-stat{label=\"Open decisions\" query=\"notes | where type == decision and status == open | count\"}\n::\n\n::bm-bars{title=\"Open decisions by owner\" query=\"notes | where type == decision and status != done | group owner | sort count desc\"}\n::\n\n::bm-table{title=\"Due soon\" query=\"notes | where tags contains infra and due < 2026-10-15 | fields title, owner, due | sort due\"}\n::\n\n::bm-timeline{title=\"Recent edits\" query=\"activity 7d\"}\n::\n\n::bm-rank{title=\"Related notes\" query=\"search 'checksum conflict' | limit 5\"}\n::\n```\n\nA pipeline reads left to right: a source (`notes`, `search 'phrase'`,\n`activity 7d|30d`), then optional stages in order: `where`, `group`, `count` or\n`sum field`, `fields`, `sort`, `limit` (1–50). The reference lists every\noperator and which components accept which shapes.\n\nShapes that work:\n\n- **Charts, ranks, and bars need a number per row**: add `group field`, `count`,\n  or `sum field`, or use a `search` source.\n- **Sparks, KPIs, and plots need several rows** (`group`, `daily`, or `search`),\n  not a single `count`.\n- **`bm-table` takes `fields`**; `bm-timeline` takes `activity` without `daily`;\n  `bm-activity` takes `activity 30d | daily`.\n\nSemantics to state honestly in the surrounding prose:\n\n- `notes` views are exact counts over indexed frontmatter. Search views are\n  ranked and bounded; call them matches, never \"all\".\n- `!=` keeps notes that lack the field. `group` puts notes without the field in\n  a `none` group.\n- Frontmatter numbers are stored as text; comparisons and sums treat decimal text\n  as numbers and ignore everything else.\n\n### Check the rendered result before you quote it\n\nIn Basic Memory Cloud, `bm_read(view=\"text\")` returns the note as a reader\nsees it: every component as plain Markdown, with live queries run now. Read it\nback after writing a live view and confirm the view has rows before you state a\nnumber in prose. A query over a field no note has renders an empty view.\n\nElsewhere, only the browser runs the query, so run the equivalent read yourself\n(for example `bm_search` with metadata filters; see the\nmemory-metadata-search skill).\n\nA live chart's body is its **snapshot**: the rows readers see when live data\ncannot load. A new live view can leave the body empty; people fill it with\n**Freeze snapshot** in the editor. If you add snapshot rows yourself, label them\nas a snapshot in prose and keep them in the chart's row grammar.\n\n## Edit a note that already has components\n\n- Change one component with `bm_edit_find_replace` on its directive line or\n  rows. Do not rewrite the whole note to change one value.\n- The rich editor normalizes formatting when a person edits the note: attribute\n  order changes and YAML blocks come back as JSON. Values survive; do not fight\n  the formatting.\n- Keep the closing `::` lines. A block with a body but no close produces\n  `directive-unclosed`; the opener and YAML props appear in a warning while\n  following Markdown stays visible. Add the close at the intended boundary in\n  source; the editor never guesses it. Standalone leaves with no body remain\n  valid, though new components should use explicit closes.\n\n## When a component shows an error\n\nA broken component renders a coded problem and a fix inside its own frame; the\nrest of the note still renders. In Basic Memory Cloud, `bm_write` and\nthe `bm_edit_*` tools list these problems under **Component problems** in their result\n(they never block the write), so fix them with `bm_edit_find_replace`. `comark(7)` in the manual\n(`memory://man/comark(7)`) is a quick reference to the syntax and codes. The\ncodes you will meet most:\n\n| Code | Meaning | Fix |\n| --- | --- | --- |\n| `prop-unknown` | A prop the component does not accept | Remove it or use a registered prop from the warning or reference |\n| `directive-unclosed` | A block with a body is missing its close | Add `::` at the intended boundary in Markdown source |\n| `query-measure` | The chart has no number per row | Add `group`, `count`, or `sum`, or use `search` |\n| `query-series` | A trend from one value | Use `group`, `daily`, or `search` with `limit` ≥ 2 |\n| `query-escaped-quote` | `\\\"` inside a query | Use single quotes inside the query |\n| `query-field` / `query-operator` | Bad field name or operator | Fields are `[A-Za-z0-9_-]` with dotted paths; ops are `== != < <= > >= contains` |\n| `query-stage-order` | Stages out of order | `where`, `group`, `count`/`sum`, `fields`, `sort`, `limit` |\n| `chart-body-empty` | No rows in the body | Add `- label: value` rows |\n| `chart-body-number` | A row without a number | Every row needs a numeric value |\n\nUnknown props preserve the original directive and data through nearby rich\nedits. Closed malformed components can use a reviewed agent replacement and\nundo; missing-close repairs require source editing. Canonical validation and\nstrict export reject malformed source until repaired. Never author internal\nrecovery nodes such as `bm-problem`. Check component names and props against the\nreference.\n\n## Checklist\n\n1. Does a figure beat a sentence here? If not, write the sentence.\n2. Is the component name and every prop in the reference?\n3. Is data in the body as Markdown rows or a table, with no `\"` inside values?\n4. For a live view: does the pipeline give the shape this component needs, and\n   did you check the numbers with a read of your own?\n5. Is there prose around the figure saying what it shows and whether a count is\n   exact or a ranking?\n"
}