@dreamteamer/knowledge 0.0.0-stage → 0.1.0

package/README.md

@@ -1,3 +1,39 @@
-# Temporary Holding Version
+# @dreamteamer/knowledge
 
-This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
+Optional knowledge capture and synthesis for personal or shared workspaces. No records or
+machine-specific configuration are shipped.
+
+```sh
+npm install @dreamteamer/core @dreamteamer/knowledge
+npm install dreamteamer
+npx dreamteamer compile
+npx dreamteamer check
+```
+
+Install core, never core alongside its individual component modules. Engine >=0.43.0 and core
+>=0.9.0 are required. `working-with-knowledge` routes requests among:
+
+- `blurbs`: raw thoughts, preserving `raw_input` separately from interpreted content; draft → digested.
+- `briefs`: reader-facing documents with audience, purpose and sources; draft/final editorial state.
+- `meta-analyses`: cross-record, cross-domain period synthesis with evidence coverage and a regeneration prompt.
+
+Meeting preps and transcript-grounded meeting critiques remain in core. A collection's presence
+never authorizes publication, cross-repository sharing or automatic task creation.
+
+## Existing local schemas
+
+This package preserves dated IDs and the `meta-analysis` filename suffix. It preserves title vs
+name, field names and statuses of the source schema family. It is not an automatic migration.
+Before enabling it in an existing workspace:
+
+1. Back up the workspace and record a hash inventory of its data files.
+2. Compare all local fields, ID patterns, storage paths/suffixes and statuses with this package.
+   A same-name local full definition collides; don't leave both active.
+3. On a disposable copy, replace matching local full schemas with this module, leaving custom
+   display choices as overlays. Run compile and check against the unchanged records.
+4. Investigate every validation failure. Do not rename, drop fields or rewrite records to silence
+   it. Verify hashes unchanged and representative source references still resolve.
+5. Adopt only after review. Keep the old configuration for rollback; no data deletion is required.
+
+The earlier unreleased personal-workspace prototype used different semantics and identifiers.
+Its records, if any were created for a local experiment, need explicit mapping, not a blind switch.

package/collections/blurbs.collection.yaml

@@ -0,0 +1,48 @@
+name: blurbs
+description: >-
+  A raw thought captured before its destination is known. Digest it later into a task, document or
+  other record; do not mistake an idea for a commitment.
+use_when: >-
+  A raw thought captured before its destination is known. Digest it later into a task, document or
+  other record; do not mistake an idea for a commitment.
+storage:
+  path: data/blurbs
+  format: md
+ids:
+  from: '{{ created | date }}--{{ name | slug }}'
+  pattern: ^\d{4}-\d{2}-\d{2}--[a-z0-9-]+$
+fields:
+  name:
+    type: string
+    description: A short title for finding the thought again.
+    required: true
+  status:
+    type: string
+    description: draft is raw; digested means it has been processed into its intended outcome.
+    default: draft
+    enum:
+      - draft
+      - digested
+  tags:
+    type: string
+    many: true
+    pattern: ^[a-z0-9]+(-[a-z0-9]+)*(:[a-z0-9]+(-[a-z0-9]+)*)?$
+    description: Free labels, lowercase and hyphenated, optionally key:value with no space around the colon.
+  raw_input:
+    type: markdown
+    description: >-
+      Verbatim user-provided input, kept separate from interpretation. Never reconstruct an exact
+      quote from memory.
+  content:
+    type: markdown
+    body: true
+    description: Working interpretation or organization of the thought; older records may contain only this.
+display:
+  nav:
+    icon: symbol-snippet
+  list:
+    columns:
+      - name
+      - status
+      - last_modified
+      - tags

package/collections/briefs.collection.yaml

@@ -0,0 +1,94 @@
+name: briefs
+description: >-
+  A document prepared for a reader: a deck outline, positioning thesis, progress report or
+  reference. Use meeting-preps instead when preparing for one specific meeting.
+use_when: >-
+  A document prepared for a reader: a deck outline, positioning thesis, progress report or
+  reference. Use meeting-preps instead when preparing for one specific meeting.
+storage:
+  path: data/briefs
+  format: md
+ids:
+  from: '{{ created | date }}--{{ name | slug }}'
+  pattern: ^\d{4}-\d{2}-\d{2}--[a-z0-9-]+$
+fields:
+  name:
+    type: string
+    description: Document title.
+  date:
+    type: date
+    description: Date the document covers or is written for.
+  company:
+    type: companies
+    description: Company this document concerns, when relevant.
+  meeting:
+    type: meetings
+    description: Related meeting, not a substitute for meeting-preps.
+  person:
+    type: string
+    description: Subject person when not represented by a contact record.
+  projects:
+    type: string
+    description: Project or workstream labels.
+    many: true
+  language:
+    type: string
+    description: Language of the document.
+  audience:
+    type: string
+    description: Who will read this document.
+  purpose:
+    type: string
+    description: What the document should help its reader do.
+  description:
+    type: string
+    description: One-line summary.
+  sources:
+    type: string
+    description: Source record references or provenance paths; preserve original source locations.
+    many: true
+  content:
+    type: markdown
+    body: true
+    description: The document itself.
+  status:
+    type: string
+    description: Editorial state, not publication permission.
+    enum:
+      - draft
+      - final
+  author:
+    type: string
+    description: The agent that wrote this record; absent when a person wrote it.
+  creation_context:
+    type: string
+    description: Why and how this record was produced, including important uncertainty.
+  source:
+    type: string
+    title: Information source
+    description: Where the information came from, such as a record reference, file path or URL.
+  confidence:
+    type: string
+    description: Confidence in this record.
+    default: normal
+    enum:
+      - low
+      - normal
+      - high
+display:
+  nav:
+    icon: file-text
+  list:
+    columns:
+      - name
+      - last_modified
+      - date
+      - company
+      - status
+  form:
+    sections:
+      - title: Brief
+        fields: [name, date, company, meeting, person, projects, language, audience, purpose, description, sources, content, status]
+      - title: Provenance
+        fields: [author, creation_context, source, confidence]
+        collapsed: true

package/collections/meta-analyses.collection.yaml

@@ -0,0 +1,69 @@
+name: meta-analyses
+description: >-
+  Cross-record synthesis over a period: recurring patterns, themes, blind spots and strategy. May
+  combine thought captures, documents, sessions and meeting analyses; not a replacement for
+  transcript-grounded meeting critique.
+use_when: >-
+  Cross-record synthesis over a period: recurring patterns, themes, blind spots and strategy. May
+  combine thought captures, documents, sessions and meeting analyses; not a replacement for
+  transcript-grounded meeting critique.
+storage:
+  path: data/meta-analyses
+  format: md
+  suffix: meta-analysis
+ids:
+  from: '{{ created | date }}--{{ title | slug }}'
+  pattern: ^\d{4}-\d{2}-\d{2}--[a-z0-9-]+$
+fields:
+  title:
+    type: string
+    description: The synthesis title, ideally the claim it makes.
+    required: true
+  content:
+    type: markdown
+    body: true
+    description: >-
+      Synthesis with evidence, interpretations, counterexamples and limitations clearly
+      distinguished.
+  description:
+    type: string
+    description: One-line finding.
+  period_start:
+    type: date
+    description: First date included.
+  period_end:
+    type: date
+    description: Last date included.
+  sources_description:
+    type: string
+    description: What was actually read, how selected, and which coverage gaps remain.
+  sources_count:
+    type: integer
+    description: Count of records actually read, not merely discovered.
+  sources_token_count:
+    type: integer
+    description: Text volume read in tokens; identify any estimation method in sources_description.
+  author:
+    type: string
+    description: The agent that wrote this record; absent when a person wrote it.
+  creation_context:
+    type: string
+    description: Why and how this record was produced, including important uncertainty.
+  prompt:
+    type: string
+    description: Self-contained regeneration instructions identifying scope and sources, never credentials.
+  tags:
+    type: string
+    many: true
+    pattern: ^[a-z0-9]+(-[a-z0-9]+)*(:[a-z0-9]+(-[a-z0-9]+)*)?$
+    description: Free labels, lowercase and hyphenated, optionally key:value with no space around the colon.
+display:
+  nav:
+    icon: notebook
+  list:
+    columns:
+      - title
+      - period_start
+      - period_end
+      - sources_count
+singular: meta-analysis

package/package.json

@@ -1,6 +1,38 @@
 {
   "name": "@dreamteamer/knowledge",
-  "version": "0.0.0-stage",
-  "stub": true,
-  "description": "Temporary package placeholder for staged publishing"
-}
+  "version": "0.1.0",
+  "description": "Optional raw thought capture, reader-facing briefs and cross-record synthesis for dreamteamer workspaces.",
+  "license": "Apache-2.0",
+  "type": "module",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/dreamteamer/dreamteamer-modules.git",
+    "directory": "modules/knowledge"
+  },
+  "keywords": [
+    "dreamteamer",
+    "dreamteamer-module",
+    "knowledge"
+  ],
+  "files": [
+    "README.md",
+    "collections",
+    "skills",
+    "ui-views"
+  ],
+  "peerDependencies": {
+    "@dreamteamer/core": ">=0.9.0"
+  },
+  "dreamteamer": {
+    "title": "Knowledge",
+    "description": "Capture raw thoughts, prepare reader-facing briefs and synthesize evidence across records and periods. Optional knowledge schemas and procedures; ships no personal records.",
+    "engine": ">=0.43.0",
+    "dependencies": [
+      "@dreamteamer/core"
+    ],
+    "peer_collections": [
+      "companies",
+      "meetings"
+    ]
+  }
+}

package/skills/working-with-knowledge/SKILL.md

@@ -0,0 +1,57 @@
+---
+name: working-with-knowledge
+description: Capture a raw thought for later digestion, prepare a general reader-facing brief, or synthesize patterns across a period of mixed workspace records. Uses blurbs, briefs and meta-analyses. For preparation or transcript-grounded critique of specific meetings, use the meeting skills instead.
+---
+
+# Working with knowledge
+
+Load `using-dreamteamer` and the target collection's compiled descriptor first. Use the workspace
+CLI to read/write records. Update an existing record when the request continues it; a new title
+is not by itself evidence of a new document. These collections do not grant sharing permission.
+
+## Capture and digest a thought
+
+Use `blurbs` when an idea is worth keeping but has no established destination yet. Store the
+user-provided wording in `raw_input` without editing it, with interpretation in `content`.
+For an older record with no raw_input, do not manufacture a verbatim original from its content.
+
+A request to capture a thought is not a request to implement it or generate tasks. Leave it
+`draft`. When asked to digest it, identify the intended outcome (document, task, design or
+retained insight), preserve the original, create or update that outcome within the authorized
+scope, and link it in the blurb body. Set `digested` only after that outcome exists or the owner
+explicitly chooses no further action. Re-running digestion should follow that link rather than
+create duplicates. Do not automatically create tasks from possibilities mentioned in a thought.
+
+## Prepare a document for a reader
+
+Use `briefs` for a report, deck outline, thesis or reference someone will act on. Use
+`meeting-preps` for one upcoming meeting and an R&D collection for its formal lifecycle.
+Capture `audience` and `purpose` from the request; resolve an important ambiguity before writing.
+Read the relevant sources, retain their references in `sources`, and separate sourced facts from
+interpretation and open questions. The content can be in the requested language.
+
+`final` means the document is finished for its intended use, not that it has been sent or approved
+for publication. Writing a brief does not authorize contacting its audience or publishing it.
+
+## Synthesize patterns across a period
+
+Use `meta-analyses` when the question is what recurs across sources, not just how one call went.
+Sources may include briefs, thought captures, meeting analyses, project documents and authorized
+session or message records. Reuse `meeting-analyses` for a transcript-grounded critique of an arc
+of calls; don't duplicate it merely because several meetings are involved.
+
+1. Inspect the previous relevant synthesis and its scope. For “since last time,” use its coverage
+   end, not its file timestamp. State any overlap or gap; ask when ambiguity changes the scope.
+2. Inventory permitted sources in the agreed period. Exclude restricted/private material outside
+   the authorized scope; do not treat repository accessibility as permission to redistribute it.
+3. Read selected evidence, including contradictory examples. For meeting claims, inspect the raw
+   transcript and speaker/ASR caveats as needed; summaries are not verbatim evidence.
+4. Write patterns supported by traceable source links, distinguish fact, inference and advice,
+   and state counterexamples, missing sources and attribution limits.
+5. Record actual coverage in `sources_description`, `sources_count` and the period fields.
+   Count records read, not search hits. Leave token count absent unless measured or explicitly
+   estimated with the method stated. `prompt` should reproduce scope and method, not contain
+   secrets or imply authorization to read a source that is no longer permitted.
+
+Recommendations remain in the analysis. Creating tasks, changing project records or transferring
+material to other repositories is a separate deliberate step under the owner's instructions.