@ericrisco/rsc 0.1.13 → 0.1.15

package/manifest.json

@@ -1,5 +1,5 @@
 {
-  "version": "0.1.13",
+  "version": "0.1.15",
   "counts": {
     "skills": 231
   },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ericrisco/rsc",
-  "version": "0.1.13",
+  "version": "0.1.15",
   "description": "Eric Risco's agent-skills catalog as a granular, self-recommending CLI installer.",
   "type": "module",
   "bin": {

package/skills/harness/SKILL.md

@@ -11,7 +11,7 @@ profiles: [minimal, core, full]
 The **harness** is the control plane of a workspace. A workspace need not be code: it can be a company, an ops desk, a legal archive, a personal knowledge vault. Whatever it is, the harness is the durable apparatus that keeps it operable and legible, made of three parts:
 
 - **`01-TOOLS/<PROVIDER>/`** — the operational tooling layer. One folder per external provider, co-locating credentials (`.env`) with the scripts that consume them. Each tool ships a working `test_connection` against the real API.
-- **`02-DOCS/`** — the **Karpathy chaos→knowledge engine**: a domain-agnostic LLM wiki (`inbox/` + `raw/` + `wiki/` + `wiki/index.md` + `wiki/log.md` + `wiki/gaps.md` + `wiki/scores.json`), fully embedded in this skill. The user drops **any raw file in any format** (PDF, image, CSV, JSON, txt, html…) into `inbox/`; an **Inbox Sweep** ("the agent goes for a walk") extracts, classifies by content, cross-links, and compiles it into the wiki — see `references/ingest-formats.md` for the multiformat Fetch and `references/wiki-protocol.md` for the protocol. Topics are inferred from content (`finanzas/`, `legal/`, `crm/`…), never hardcoded to software docs. The wiki **self-improves continuously**: every Ingest, Sweep and Query triggers a Maintenance Pass (deterministic lint, score recomputation, gap detection, See Also sweep), and every N interactions a Micro-Improve runs (rewrite 1 low-scoring article, fill 1 gap, preserving old versions in `_archive/`). Deep Improve runs on explicit request or via scheduled cron. No external skill needed.
+- **`02-DOCS/`** — the **Karpathy chaos→knowledge engine**: a domain-agnostic LLM wiki (`inbox/` + `raw/` + `raw/worklog/` + `wiki/` + `wiki/index.md` + `wiki/log.md` + `wiki/gaps.md` + `wiki/scores.json` + the `.base` views), fully embedded in this skill. It feeds from **two on-ramps**: (1) the user drops **any raw file in any format** (PDF, image, CSV, JSON, txt, html…) into `inbox/`, and an **Auto-Ingest Sweep** ("the agent goes for a walk") extracts, classifies, cross-links, and compiles it — and goes further, **discovering un-ingested documents anywhere in the workspace** (a stray `/transcripts`, `/facturas`…), bounded by `.rscignore` and de-duplicated by the `wiki/.ingested.json` ledger, fired automatically at SessionStart (when the inbox has material) and by the daily cron. Ingest is non-destructive (originals are copied, never moved); removing an emptied folder needs explicit consent; (2) a **Worklog Sweep** captures **what we do** — every meaningful session of work is itself a `raw` source (`raw/worklog/`) compiled into the wiki, fired by a `PreCompact`/`SessionEnd` hook, by an explicit milestone (a commit), or by the daily curation automation (`references/daily-curation-automation.md`). The wiki is **Obsidian-native**: wikilinks, YAML frontmatter (Properties), readable filenames, and `.base` views give a real graph + live tables — structure, **not vector DB / embeddings / RAG**. The agent writes it; the human reads it in Obsidian. See `references/wiki-protocol.md` for the protocol, `references/ingest-formats.md` for the multiformat Fetch, `references/wiki-worklog-template.md` for the work capture, and `references/obsidian-scaffolding.md` for the vault scaffolding. Topics are inferred from content (`finanzas/`, `legal/`, `crm/`…), never hardcoded. The wiki **self-improves continuously**: every Ingest, Sweep and Query triggers a Maintenance Pass (deterministic lint, score recomputation, gap detection, Related/See-Also sweep), and every N interactions a Micro-Improve runs. Deep Improve runs on explicit request or via the scheduled daily curation. No external skill needed.
 - **The Knowledge map** — the `## Knowledge map` section of the root `CLAUDE.md` that indexes the wiki (including the `harness/` topic) and is read by every other skill before it works in its area.
 
 `harness` is the **protagonist concept**. `init` is the bootstrap front door — it gauges the user, drafts the profile, and hands off the first scaffold. THIS skill (`harness`) is the **ongoing control**: it audits, migrates, scaffolds, sweeps the inbox, and keeps the wiki, the tooling and the Knowledge map honest over the life of the workspace. It also generates root `CLAUDE.md` and `AGENTS.md`, and migrates legacy `XX-*` numbered folders into the canonical layout.
@@ -55,6 +55,8 @@ phase, last verdict, next steps, risks, commands and skill_resolution. This is
 not a replacement for the wiki; it is the recovery note that lets the next
 agent resume without trusting chat history.
 
+**Opt-out marker.** When a workspace has no profile yet, a freshly-installed session auto-starts `init` (via the `suggest` Onboarding gate and claude's SessionStart hook). A `.rsc/.no-harness` file is the explicit opt-out: present → never auto-start onboarding in this repo, even without a profile. `harness` treats it as canonical and never overwrites or deletes it.
+
 ### 4. Decision pattern — "siempre 3 opciones"
 
 For **any significant decision** (deploy target, database, framework, hosting, tooling…):
@@ -98,12 +100,15 @@ harness/
 │   ├── agents-md-template.md         ← root AGENTS.md template
 │   ├── tools-readme-template.md      ← 01-TOOLS/README.md catalog template
 │   ├── audit-report-template.md      ← exact format of the audit report shown to user
-│   ├── wiki-protocol.md              ← embedded wiki protocol + Inbox Sweep + Continuous Improvement
+│   ├── wiki-protocol.md              ← embedded wiki protocol + Inbox/Worklog Sweep + Continuous Improvement
 │   ├── ingest-formats.md             ← multiformat Fetch (PDF, image, CSV, JSON, html…)
 │   ├── inbox-readme-template.md      ← the inbox/README.md drop-zone contract
 │   ├── wiki-raw-template.md          ← format for raw/<topic>/*.md
-│   ├── wiki-article-template.md     ← format for wiki/<topic>/*.md
-│   ├── wiki-index-template.md       ← format for wiki/index.md (with Score column)
+│   ├── wiki-worklog-template.md      ← format for raw/worklog/*.md (work-driven capture)
+│   ├── wiki-article-template.md     ← format for wiki/<topic>/*.md (Obsidian frontmatter + wikilinks)
+│   ├── wiki-index-template.md       ← format for wiki/index.md (machine catalog; .base = human nav)
+│   ├── obsidian-scaffolding.md      ← .base views, .gitignore, attachments, .obsidian config
+│   ├── daily-curation-automation.md ← portable scheduled compounding pass (the third trigger)
 │   ├── wiki-archive-template.md     ← format for archived query answers
 │   └── wiki-gaps-template.md        ← format for wiki/gaps.md (Knowledge Gaps log)
 ├── assets/
@@ -321,12 +326,15 @@ If any of these occur, stop and report:
 - `references/tools-readme-template.md` — `01-TOOLS/README.md` catalog template.
 - `references/audit-report-template.md` — text summary format for the in-conversation audit summary.
 - `references/audit-report-template.html` — HTML format for the full per-run audit artifact written to `02-DOCS/audits/`.
-- `references/wiki-protocol.md` — embedded protocol for the `02-DOCS/` chaos→knowledge layer (initialization, **Inbox Sweep**, ingest, query, lint, **Continuous Improvement**: Maintenance Pass, Micro-Improve, Deep Improve).
+- `references/wiki-protocol.md` — embedded protocol for the `02-DOCS/` chaos→knowledge layer (initialization, **Inbox Sweep**, **Worklog Sweep**, ingest, query, lint, **Continuous Improvement**: Maintenance Pass, Micro-Improve, Deep Improve; Obsidian-native conventions).
 - `references/ingest-formats.md` — multiformat Fetch: how any input (PDF, image, CSV/Excel, JSON/API, html, docx, email, unknown binary) becomes `raw/` markdown with the original preserved in `_originals/`.
 - `references/inbox-readme-template.md` — the `inbox/README.md` drop-zone contract shown to the user.
 - `references/wiki-raw-template.md` — format for `02-DOCS/raw/<topic>/*.md`.
-- `references/wiki-article-template.md` — format for `02-DOCS/wiki/<topic>/*.md`.
-- `references/wiki-index-template.md` — format for `02-DOCS/wiki/index.md` (with Score column).
+- `references/wiki-worklog-template.md` — format for `02-DOCS/raw/worklog/*.md` (the work-driven capture; the session as a raw source).
+- `references/wiki-article-template.md` — format for `02-DOCS/wiki/<topic>/*.md` (Obsidian frontmatter + wikilinks + `## Related`).
+- `references/wiki-index-template.md` — format for `02-DOCS/wiki/index.md` (machine catalog; the `.base` views are the human navigation).
+- `references/obsidian-scaffolding.md` — the vault scaffolding: `.base` views, `.gitignore`, `attachments/`, `.obsidian` config (no vector DB / RAG).
+- `references/daily-curation-automation.md` — portable scheduled compounding pass; on Claude Code wired via the `schedule` skill.
 - `references/wiki-archive-template.html` — HTML format for archived query answers (point-in-time, never edited).
 - `references/wiki-dashboard-template.html` — HTML format for the live wiki dashboard, regenerated by Maintenance Pass.
 - `references/wiki-deep-improve-report-template.html` — HTML format for Deep Improve run reports.

package/skills/harness/references/daily-curation-automation.md

@@ -0,0 +1,66 @@
+# Daily curation automation — the compounding pass on a timer
+
+The third trigger of the work-driven path (the other two: the PreCompact/SessionEnd
+hook and explicit milestones). This is the harness's existing **Continuous
+Improvement** (Maintenance Pass / Micro-Improve / Deep Improve) promoted to a
+**first-class, portable, scheduled** automation — "the agent goes for a walk" on a
+timer, not only when asked.
+
+**Portable definition + Claude-Code reliability** (same pattern as the hooks): the
+spec below works on any assistant as a protocol; on Claude Code, wire a real daily
+cron via the `schedule` skill pointed at this prompt.
+
+## Purpose
+
+Run once per day, at a quiet hour, to keep the second brain alive:
+
+- run the **Auto-Ingest Sweep**: process `inbox/`, then scan the workspace (minus
+  `.rscignore`) for un-ingested documents, ingest the clearly-documentary folders,
+  record them in `wiki/.ingested.json`, and propose the ambiguous ones
+- process pending `raw/worklog/` material into `wiki/`
+- strengthen weak / low-score pages; fill one open gap
+- repair links and the graph (resolve broken wikilinks, add `## Related` where proper nouns recur)
+- mark `stale` pages whose sources moved on
+- record what changed in `wiki/log.md`
+
+## Suggested schedule
+
+Daily, at a quiet hour when the user is unlikely to be editing the vault.
+
+## Scope (paths)
+
+The **workspace root** (for Auto-Ingest discovery, bounded by `.rscignore`),
+`raw/worklog/`, `inbox/`, `raw/`, `wiki/`, `wiki/.ingested.json`, `wiki/harness/`.
+
+## Prompt
+
+```text
+Run the daily curation pass for this 02-DOCS second brain.
+
+Use the harness wiki-protocol.md as the contract. First run the Auto-Ingest Sweep:
+process inbox/, then scan the workspace (minus .rscignore) for un-ingested documents,
+ingest the clearly-documentary folders (copy originals, never move), record them in
+wiki/.ingested.json, and list ambiguous folders as proposals (do not grab them).
+Then compile pending raw/worklog/ material into wiki/ pages (update existing before
+creating new), keep frontmatter + wikilinks consistent, append significant decisions
+to wiki/harness/decisions.md, refresh the .base-backed navigation, recompute
+scores.json + the score: property, and append a dated entry to wiki/log.md.
+
+Be conservative. If the topic map no longer fits, record a recommendation in
+wiki/gaps.md or a page's Open Questions instead of moving large parts of the
+vault without explicit approval.
+```
+
+## Expected outcome (one of)
+
+- worklog/inbox material distilled; `status` flipped to processed
+- maintenance improvements to links, scores, `.base` views, or stale flags
+- a `wiki/log.md` entry explaining what was checked and what remains open
+- no file changes when there is genuinely nothing useful to do
+
+## Safety
+
+- Do not rewrite or delete raw sources.
+- Do not perform large taxonomy reorganizations without explicit approval.
+- Do not create pages from weak connections just to look busy.
+- Prefer small, compounding improvements over big sweeps.

package/skills/harness/references/ingest-ignore-defaults.md

@@ -0,0 +1,48 @@
+# Ignore defaults — what the Auto-Ingest Sweep must never scan
+
+The Auto-Ingest Sweep discovers un-ingested documents across the whole workspace.
+`02-DOCS/.rscignore` scopes that scan so it never grabs code, build output, tooling,
+the wiki's own internals, or non-document files. This is the single guardrail that
+keeps "automatic, workspace-wide" from turning into "tragar lo que no toca".
+
+`02-DOCS/.rscignore` is **tracked** (not gitignored): the team shares the same scan
+boundary. Users append their own patterns freely.
+
+## Default `02-DOCS/.rscignore`
+
+```gitignore
+# 02-DOCS/.rscignore — paths the Auto-Ingest Sweep must NEVER scan as a knowledge source.
+# Tracked (the team shares it). Append your own patterns below.
+node_modules/
+.git/
+.next/
+dist/
+build/
+.venv/
+__pycache__/
+.dart_tool/
+.pytest_cache/
+.ruff_cache/
+01-TOOLS/
+02-DOCS/inbox/_processed/
+02-DOCS/raw/
+02-DOCS/wiki/
+.DS_Store
+# Only document-like files are candidates anyway — pdf, docx, doc, md, txt, csv,
+# tsv, json, html, rtf, and doc-like images (png/jpg of scans). Everything else
+# (code, archives, binaries, lockfiles) is ignored even outside these paths.
+```
+
+## App source dirs are ignored at scan time
+
+Beyond the file above, the sweep adds every detected **app source directory** to the
+effective ignore set: any directory carrying a manifest (`package.json`,
+`pyproject.toml`, `pubspec.yaml`, `go.mod`, `Cargo.toml`) and its tree. Source code is
+never a knowledge source — `02-DOCS/wiki/` is where understanding about the code lives,
+written deliberately, not scraped from the repo.
+
+## Why `02-DOCS/raw/` and `02-DOCS/wiki/` are ignored
+
+They are *outputs* of ingestion, not inputs. Scanning them would re-ingest the wiki
+into itself. `inbox/` is the only `02-DOCS/` subtree the sweep reads as a source
+(plus discovered external folders).

package/skills/harness/references/obsidian-scaffolding.md

@@ -0,0 +1,96 @@
+# Obsidian scaffolding — make `02-DOCS/` open as a vault
+
+Materialized during **Initialization** (see `wiki-protocol.md`). Turns the wiki
+into a native Obsidian vault: graph, backlinks, Properties, and Bases — with **no
+vector DB, no embeddings, no RAG**. Navigation is structure (links + frontmatter),
+not semantic search. Do NOT add embedding/semantic-search plugins; that breaks the
+thesis. Obsidian core only.
+
+The point a human opens the vault at (the "base folder") is `02-DOCS/` itself —
+not the repo root. Obsidian writes its config into `02-DOCS/.obsidian/`.
+
+## `02-DOCS/.gitignore` (additions)
+
+```gitignore
+# Obsidian per-machine state (the rest of .obsidian/ — themes, bases, config — is tracked)
+.obsidian/workspace*.json
+.obsidian/cache
+# regenerated / per-run surfaces
+wiki/dashboard.html
+audits/*.html
+inbox/_processed/
+```
+
+## `02-DOCS/attachments/README.md`
+
+```markdown
+# attachments
+
+Shared binaries referenced from wiki pages (images, PDFs, diagrams) when they are
+not bundled with a specific raw source. Link them from the page that depends on
+them. Source-bundled originals stay in `raw/<topic>/_originals/`.
+```
+
+## `02-DOCS/.obsidian/app.json` (attachment folder)
+
+```json
+{
+  "attachmentFolderPath": "attachments",
+  "newLinkFormat": "shortest",
+  "useMarkdownLinks": false
+}
+```
+
+`useMarkdownLinks: false` makes new internal links wikilinks. `userIgnoreFilters`
+(Settings → Files & Links → Excluded files) should hide noise from search/graph:
+`raw/_originals/`, `inbox/_processed/`, `audits/`.
+
+## `.base` views — the human navigation (replace hand-maintained nav)
+
+These supersede manual upkeep of `index.md`/`dashboard.html` for humans;
+`index.md` and `scores.json` remain as the machine layer + fallback.
+
+### `02-DOCS/wiki/Articles.base`
+
+```yaml
+filters:
+  and:
+    - type == "article"
+views:
+  - type: table
+    name: All Articles
+    order: [file.name, topic, status, score, updated]
+    sort:
+      - property: score
+        direction: DESC
+```
+
+### `02-DOCS/wiki/Worklog.base`
+
+```yaml
+filters:
+  and:
+    - type == "worklog"
+views:
+  - type: table
+    name: Worklog
+    order: [file.name, topic, date, status]
+    sort:
+      - property: date
+        direction: DESC
+```
+
+### `02-DOCS/wiki/Decisions.base`
+
+```yaml
+filters:
+  and:
+    - type == "decision"
+views:
+  - type: table
+    name: Decisions
+    order: [file.name, topic, status, updated]
+    sort:
+      - property: updated
+        direction: DESC
+```

package/skills/harness/references/wiki-article-template.md

@@ -1,7 +1,19 @@
+---
+title: {Title}
+aliases: [{old-slug}]
+tags: []
+type: article
+topic: {topic}
+status: draft
+sources: []
+updated: {YYYY-MM-DD}
+score: 0.0
+---
+
 # {Title}
 
 > Sources: {Author1, YYYY-MM-DD; Author2, YYYY-MM-DD}
-> Raw: [{source1}](../../raw/{topic1}/{filename1}.md); [{source2}](../../raw/{topic2}/{filename2}.md)
+> Raw: [[{raw-source-1}]]; [[{raw-source-2}]]
 
 ## Overview
 
@@ -11,10 +23,21 @@
 
 {Synthesize a coherent structure from the source material. Do not copy source text verbatim; distill and reorganize. Use blockquotes sparingly for particularly important original phrasing.}
 
-{OPTIONAL — include this section only when cross-references exist:}
-
-## See Also
-
-{Cross-references to related wiki articles. Maintained during lint. Use relative links:
-- Same topic: [Other Article](other-article.md)
-- Different topic: [Other Article](../other-topic/other-article.md)}
+## Related
+
+{OPTIONAL — include only when cross-references exist. Maintained during lint. Use wikilinks so the link survives renames and shows in the Obsidian graph + backlinks:
+- Same or different topic: `- [[Other Article]] — why it connects.`
+- With display text: `- [[Other Article|how we say it here]] — why it connects.`}
+
+<!--
+FRONTMATTER CONTRACT (this is what powers Obsidian Properties + Bases + graph):
+- title    human-readable; the H1 matches it.
+- aliases  keep the old kebab slug here so pre-migration links/wikilinks still resolve.
+- tags     cross-cutting qualities ONLY — never the main category (that is `topic`).
+- type     article | decision | worklog | brief | spec | profile
+- topic    the wiki/<topic>/ folder, one level. Inferred from content, never hardcoded.
+- status   draft | stable | stale  (lifecycle; filterable in a Base).
+- sources  wikilinks to raw/ pages this was distilled from (cite your evidence).
+- updated  YYYY-MM-DD of the last meaningful edit.
+- score    mirror of scores.json, written by the Maintenance Pass (single writer → no drift).
+-->

package/skills/harness/references/wiki-index-template.md

@@ -1,13 +1,18 @@
 # Knowledge Base Index
 
+> **Human navigation lives in the `.base` views** (`Articles.base`, `Worklog.base`,
+> `Decisions.base`) — live tables over frontmatter, sorted by score/status/date.
+> This index is the **machine-readable catalog + fallback** the agent maintains;
+> it is not the primary surface a human browses in Obsidian.
+
 ## {topic-name}
 
 {One-line description of this topic.}
 
 | Article | Summary | Updated | Score |
 |---------|---------|---------|-------|
-| [{Article Title}]({topic-name}/{article}.md) | {One-line summary} | {YYYY-MM-DD} | {N.N} |
-| [{Archived Article}]({topic-name}/{archived}.md) | [Archived] {One-line summary} | {YYYY-MM-DD} | {N.N} |
+| [[{Article Title}]] | {One-line summary} | {YYYY-MM-DD} | {N.N} |
+| [[{Archived Article}]] | [Archived] {One-line summary} | {YYYY-MM-DD} | {N.N} |
 
 ## {another-topic}
 
@@ -15,7 +20,7 @@
 
 | Article | Summary | Updated | Score |
 |---------|---------|---------|-------|
-| [{Article Title}]({another-topic}/{article}.md) | {One-line summary} | {YYYY-MM-DD} | {N.N} |
+| [[{Article Title}]] | {One-line summary} | {YYYY-MM-DD} | {N.N} |
 
 > **Score**: composite quality score (inbound links, source count, cited count,
 > freshness; minus conflicts and orphan penalty). Regenerated on every

package/skills/harness/references/wiki-protocol.md

@@ -12,9 +12,12 @@ skill required. When the parent skill (`SKILL.md`) reaches Phase 4 step 8
 
 - [The paradigm — chaos in, knowledge out](#the-paradigm--chaos-in-knowledge-out)
 - [Architecture](#architecture) · [Format split: markdown for state, HTML for surfaces](#format-split-markdown-for-state-html-for-surfaces)
+- [The vault is Obsidian-native](#the-vault-is-obsidian-native)
 - [Initialization](#initialization)
 - [Ingest](#ingest) · [Fetch](#fetch-raw) · [Compile](#compile-wiki) · [Cascade Updates](#cascade-updates) · [Post-Ingest](#post-ingest)
 - [Inbox Sweep — "el agente sale a pasear"](#inbox-sweep--el-agente-sale-a-pasear)
+- [Worklog Sweep — work-driven capture](#worklog-sweep--work-driven-capture)
+- [Auto-Ingest Sweep — automatic, workspace-wide](#auto-ingest-sweep--automatic-workspace-wide)
 - [Query](#query)
 - [Lint](#lint) · [Deterministic Checks](#deterministic-checks-auto-fix) · [Heuristic Checks](#heuristic-checks-report-only)
 - [Continuous Improvement](#continuous-improvement) — Maintenance Pass / Micro-Improve / Deep Improve
@@ -144,6 +147,18 @@ Templates live alongside this file in `references/`:
 
 ---
 
+## The vault is Obsidian-native
+
+`02-DOCS/` opens directly as an Obsidian vault (the human's "base folder" is
+`02-DOCS/` itself, not the repo root). The agent writes the wiki; the human
+**reads** it in Obsidian — graph, backlinks, Properties, and `.base` views — which
+is exactly the Karpathy split. Navigation is **structure** (wikilinks + frontmatter
++ Bases), not semantic similarity: **no vector DB, no embeddings, no RAG**, and no
+embedding/search plugins. The `.base` views become the human navigation; `index.md`
++ `scores.json` remain the machine layer + fallback. The exact scaffolding (the
+`.base` files, `.obsidian/app.json`, attachments, `.gitignore`, excluded files) is
+materialized from `obsidian-scaffolding.md` during Initialization.
+
 ## Initialization
 
 Triggers on the first Ingest. Check whether `02-DOCS/raw/` and `02-DOCS/wiki/`
@@ -158,15 +173,28 @@ exist. Create only what is missing; never overwrite existing files:
 - `02-DOCS/wiki/log.md` — heading `# Wiki Log`, empty body
 - `02-DOCS/wiki/gaps.md` — heading `# Knowledge Gaps`, empty body
 - `02-DOCS/wiki/scores.json` — `{}` (populated by the first Maintenance Pass)
+- `02-DOCS/wiki/.ingested.json` — `{}` (the Auto-Ingest Sweep's seen-ledger)
+- `02-DOCS/.rscignore` — the scan boundary for the Auto-Ingest Sweep, from
+  `ingest-ignore-defaults.md` (tracked, not gitignored)
 - `02-DOCS/wiki/reports/` directory (with `.gitkeep`) — will hold Deep Improve reports
 - `02-DOCS/audits/` directory (with `.gitkeep`) — holds per-run audit reports
   from the parent skill (`audit-YYYY-MM-DD-HHMM.html`). May already exist if
   Phase 2 of the parent skill created it eagerly to write the first audit.
+- `02-DOCS/raw/worklog/` directory (with `.gitkeep`) — the **work-driven on-ramp**
+  (see "Worklog Sweep"); holds `YYYY-MM-DD-<slug>.md` captures of what we did.
+- `02-DOCS/attachments/` directory + `README.md` — shared binaries (Obsidian's
+  default attachment folder), from `obsidian-scaffolding.md`.
+- `02-DOCS/wiki/Articles.base`, `Worklog.base`, `Decisions.base` — the human
+  navigation views, from `obsidian-scaffolding.md`.
+- `02-DOCS/.obsidian/app.json` — sets the attachment folder and wikilink defaults,
+  from `obsidian-scaffolding.md`.
 - `02-DOCS/.gitignore` — at minimum:
   - `wiki/dashboard.html` (regenerated by Maintenance Pass)
   - `audits/*.html` (per-run audit artifacts from the parent skill)
   - `inbox/_processed/` (sweep archive — bulky, derivable; the durable copy
     of every binary lives in `raw/<topic>/_originals/`)
+  - `.obsidian/workspace*.json` and `.obsidian/cache` (Obsidian per-machine state;
+    the rest of `.obsidian/` — themes, bases, config — is tracked)
   - `wiki/reports/*.html` is **not** gitignored — Deep Improve reports are
     durable historical records, unlike the per-run audit snapshots.
 
@@ -341,6 +369,123 @@ Append to `02-DOCS/wiki/log.md`:
 
 ---
 
+## Worklog Sweep — work-driven capture
+
+The Inbox Sweep turns files the user *dropped* into knowledge. The **Worklog
+Sweep** turns what we *did* into knowledge — the second on-ramp. A work session is
+just another source of chaos: it lands in `raw/worklog/` and is Compiled into
+`wiki/` through the exact same pipeline. No parallel system.
+
+### What it captures
+
+One `raw/worklog/YYYY-MM-DD-<slug>.md` per meaningful unit of work, in the format
+of `wiki-worklog-template.md` (frontmatter `type: worklog`, `topic`, `date`,
+`status: unprocessed`): what we did, why (intent + decisions), files touched
+(`path:line`), outcome with evidence, open questions/next, commands. This
+generalizes the SDD session-summary convention to *any* work.
+
+### Triggers
+
+- **Hook — `PreCompact` / `SessionEnd`** (Claude Code): fires right before context
+  is lost or the session ends. The hook only *reminds*; the agent writes the
+  worklog (Karpathy: the LLM writes). Wired by `targets/` → `.rsc/worklog-checkpoint.sh`.
+- **Explicit milestone**: after a commit or a shipped feature, capture a worklog.
+- **Daily curation automation**: distills any pending worklog raw on a timer (see
+  Continuous Improvement and `daily-curation-automation.md`).
+
+### Throttle — capture signal, not noise
+
+Write a worklog **only** when there is durable signal: files changed, a decision
+was made, or a milestone was hit. **Skip** pure read/answer turns and questions
+with no change. Never manufacture a worklog "to look busy" — that is the same
+safety rule the curation pass honors.
+
+### Steps
+
+1. **Capture.** Write the worklog to `raw/worklog/YYYY-MM-DD-<slug>.md`
+   (`status: unprocessed`). Infer `topic` from content.
+2. **Compile.** Run the normal Compile (Fetch is a no-op — the worklog is already
+   `raw`): distill durable points into `wiki/<topic>/` articles (update existing
+   before creating new), add wikilink cross-refs + `## Related`, cascade.
+3. **Route decisions.** Every significant decision is appended to
+   `wiki/harness/decisions.md` (append-only, the "siempre 3 opciones" shape).
+4. **Close.** Flip the worklog's `status` to `processed` (the raw stays as
+   evidence — never deleted or rewritten) and append to `wiki/log.md`.
+
+### Post-Sweep log
+
+Append to `02-DOCS/wiki/log.md`:
+
+```markdown
+## [YYYY-MM-DD] worklog | <slug> → <A> articles touched
+```
+
+### Worklog Sweep safety rails
+
+- **Evidence is immutable.** The worklog raw is never edited for style after the
+  fact; only its `status` flips. Distillation happens in `wiki/`, not in the raw.
+- **No invention.** Capture what actually happened (cite commands/outputs); do not
+  embellish outcomes. "Done" requires evidence.
+- **Throttled.** No worklog for trivial turns (see Throttle above).
+
+---
+
+## Auto-Ingest Sweep — automatic, workspace-wide
+
+The Inbox Sweep processes what the user *dropped* in `inbox/`. The **Auto-Ingest
+Sweep** extends it: it also **discovers** un-ingested documents anywhere in the
+workspace and ingests them on its own — so the user never has to remember to file
+sources. It runs automatically (SessionStart nudge + daily cron; see Triggers) and
+on request.
+
+The hard rule that keeps "automatic, workspace-wide" safe: **ingesting is
+non-destructive** (read + create wiki pages + copy the original into
+`raw/<topic>/_originals/`), so it is safe to do unattended. **Moving or deleting
+anything is destructive** and always requires explicit, quoted consent.
+
+### Steps
+
+1. **inbox/** — run the Inbox Sweep on `inbox/` (unchanged).
+2. **Scan** — walk the workspace minus `.rscignore` (and minus detected app source
+   dirs). Consider only document-like files. A file is **un-ingested** when its
+   `path` is absent from `wiki/.ingested.json` OR its content hash differs.
+3. **Classify** each candidate folder:
+   - **Clearly documentary** (a folder of docs/transcripts/pdfs, no code) →
+     **auto-ingest**: Compile to `wiki/`, copy the original into
+     `raw/<topic>/_originals/` (**copy, never move** — the source stays put).
+   - **Ambiguous** (mixed, or possibly app data) → **list it and propose**; do not
+     touch it. Log it to `gaps.md` so the next pass remembers.
+4. **Record** every ingested source in `wiki/.ingested.json`.
+5. **Offer cleanup** — for a folder the sweep fully ingested, **offer** to remove it
+   once empty, with the consent quoted back from the user. **Never auto-delete**, and
+   never a folder the sweep did not itself empty.
+
+### Ledger — `wiki/.ingested.json`
+
+The seen-ledger; the single source of truth for "what we've already ingested",
+which makes the workspace scan idempotent (the Inbox Sweep alone tracks
+processed-ness only by moving files to `inbox/_processed/`).
+
+```json
+{
+  "<relpath-from-workspace-root>": { "hash": "sha256:…", "ingested": "YYYY-MM-DD", "topic": "…" }
+}
+```
+
+Un-ingested = path absent OR hash changed. `inbox/` files still move to
+`inbox/_processed/` **and** get a ledger entry. Initialized as `{}`.
+
+### Auto-Ingest safety rails
+
+- **Ledger-idempotent.** Never reprocess a source whose path+hash is recorded.
+- **Scoped by `.rscignore`.** Never scan code, build output, `01-TOOLS/`, the
+  wiki's own `raw/`+`wiki/`, or non-document files (see `ingest-ignore-defaults.md`).
+- **Copy, don't move.** Discovered originals are copied; the source is left in place.
+- **Consent for destruction.** Removing an emptied folder is quoted-consent only.
+- **Ambiguity is proposed, not grabbed.**
+
+---
+
 ## Query
 
 Search the wiki and answer questions. Examples of triggers:
@@ -635,15 +780,25 @@ These apply to all Layers:
 
 ## Conventions
 
-- Standard markdown with relative links throughout.
+- **Obsidian-friendly markdown.** Internal cross-references use **wikilinks**
+  (`[[Page]]` / `[[Page|alias]]`) — resolved by filename, no path needed — so the
+  Obsidian graph + backlinks come for free and survive moves. (Legacy relative
+  markdown links still resolve in Obsidian; the migration pass converts them.)
+- **Frontmatter on curated articles.** Every `wiki/<topic>/` article carries the
+  YAML frontmatter from `wiki-article-template.md` (`title, aliases, tags, type,
+  topic, status, sources, updated, score`). This is what powers Obsidian
+  Properties + the `.base` views. Operational files (`index.md`, `log.md`,
+  `gaps.md`) may use simpler formats.
+- **Human-readable wiki filenames** (`Month-End Close.md`) for clean graph nodes;
+  the old kebab slug goes in `aliases` so prior links/wikilinks still resolve.
+  `raw/` and `raw/worklog/` filenames stay kebab + date (chaos, not display).
 - `wiki/` supports one level of topic subdirectories only. No deeper
   nesting.
 - Today's date for log entries, Collected dates, and Archived dates.
   Updated dates reflect when the article's knowledge content last changed.
   Published dates come from the source (use `Unknown` when unavailable).
-- Inside `wiki/` files, all markdown links use paths relative to the
-  current file. In conversation output, use project-root-relative paths
-  (e.g., `02-DOCS/wiki/topic/article.md`).
+- Inside `wiki/` files, internal references use wikilinks. In conversation
+  output, use project-root-relative paths (e.g., `02-DOCS/wiki/topic/Article.md`).
 - Ingest updates both `02-DOCS/wiki/index.md` and `02-DOCS/wiki/log.md`.
   Archive (from Query) updates both. Lint updates `02-DOCS/wiki/log.md`
   (and `02-DOCS/wiki/index.md` only when auto-fixing index entries). Plain

package/skills/harness/references/wiki-worklog-template.md

@@ -0,0 +1,56 @@
+# Worklog raw template — `raw/worklog/<YYYY-MM-DD>-<slug>.md`
+
+The **work-driven on-ramp**. A Worklog Sweep writes one of these per meaningful
+unit of work — what we *did*, not what someone dropped in `inbox/`. It is `raw`
+(evidence: preserved, never edited after the fact) and is Compiled into `wiki/`
+like any other source. The session is just another source of chaos.
+
+**Throttle — do NOT write a worklog for:** pure read/answer turns, questions with
+no file change, no decision, and no commit. Capture only durable signal. Never
+manufacture a worklog "to look busy."
+
+```yaml
+---
+type: worklog
+topic: {inferred-topic}
+date: {YYYY-MM-DD}
+status: unprocessed
+sources: []
+---
+```
+
+## What we did
+
+{Concrete changes / features / fixes — one bullet each. Plain, factual.}
+
+## Why
+
+{Intent and the decisions taken. Every *significant* decision is ALSO appended to
+`wiki/harness/decisions.md` (append-only, with the 3-options shape) during Compile.}
+
+## Files touched
+
+- `{path}:{line}` — {what changed}
+
+## Outcome
+
+{Verdict: shipped & verified / partial / blocked. Include the evidence — test
+output, command result, screenshot path. No "done" without proof.}
+
+## Open questions / next
+
+- {What remains, what to pick up next session.}
+
+## Commands
+
+- `{command}` — {result}
+
+<!--
+LIFECYCLE: written by the Worklog Sweep → status: unprocessed. The Compile step
+distills durable points into wiki/<topic>/ articles (update existing before
+creating new), cross-links, cascades, appends decisions to decisions.md, then
+flips this file's status to processed (the raw stays as evidence; it is not
+deleted or rewritten). Triggers that fire a Worklog Sweep: a PreCompact/SessionEnd
+hook (Claude Code), an explicit milestone (a commit / a shipped feature), and the
+daily curation automation. See `wiki-protocol.md` → "Worklog Sweep".
+-->

package/skills/init/SKILL.md

@@ -55,6 +55,10 @@ Before discovery, before any recommendation, write the profile to `02-DOCS` and
 
 If `02-DOCS/` does not yet exist (greenfield), create `02-DOCS/wiki/harness/` now — just enough to hold these two files. That, plus the `CLAUDE.md` Knowledge-map link, is everything `init` writes; ALL other `01-TOOLS/` + `02-DOCS/` scaffolding is the `harness` skill's job.
 
+### Opt-out marker — `.rsc/.no-harness`
+
+A freshly-installed session auto-starts `init` while `02-DOCS/wiki/harness/user-profile.md` is absent (the `suggest` Onboarding gate + claude's SessionStart hook). If the user does not want a harness in this repo (e.g. they installed only code skills), write an empty `.rsc/.no-harness` — this permanently silences the auto-start here even before a profile exists. Completing first contact (which writes `user-profile.md`) also silences it. The marker is project-local; commit it so the "no harness here" decision is shared by the team.
+
 ### How every skill adapts to the dial
 
 Every rsc skill MUST read `user-profile.md` at the start of a session and adapt:

package/skills/suggest/SKILL.md

@@ -24,6 +24,16 @@ Rules:
 - Never recommend something already installed (`npx @ericrisco/rsc list`).
 - One suggestion at a time. Don't interrupt the flow for nice-to-haves.
 
+## Onboarding gate (first contact)
+
+Before handling the first request of the session, check the workspace state:
+
+- If `02-DOCS/wiki/harness/user-profile.md` is **missing** AND `.rsc/.no-harness` is **missing**, the harness has never been set up here — your FIRST action is to invoke `init` (auto-onboarding), which opens with the two gauging questions (technical level + accompaniment dial). Don't wait to be asked; don't start the user's task until first contact is done.
+- If the user declines or says they don't want a harness here ("sin harness", "solo código", "no quiero esto") — create an empty `.rsc/.no-harness` and never auto-start `init` in this repo again. Confirm in one line.
+- Once `02-DOCS/wiki/harness/user-profile.md` exists, this gate is inert — never re-onboard.
+
+This is the universal layer every assistant reads. On Claude Code a SessionStart hook also prints this reminder deterministically, but the rule above is what makes it fire everywhere.
+
 ## Orientación (siempre)
 
 Cierra cada turno con el **bloque-brújula** (📍 dónde estás · ✅ qué hiciste · 🧭 por qué · ➡️ siguiente, terminando en pregunta), calibrado al dial de `02-DOCS/wiki/harness/user-profile.md`. **Nunca termines en seco.** Protocolo completo: skill `orient` → `skills/orient/references/orientation-contract.md`. (Defiere a `suggest` el "¿instalo la skill que falta?".)

package/targets/claude.js

@@ -1,22 +1,56 @@
-import { readFileSync, writeFileSync, existsSync, mkdirSync } from 'node:fs';
-import { dirname } from 'node:path';
+import { readFileSync, writeFileSync, existsSync, mkdirSync, copyFileSync, chmodSync } from 'node:fs';
+import { dirname, join } from 'node:path';
+import { fileURLToPath } from 'node:url';
 import { linkOrCopy } from './index.js';
 
 export function writeSkill(id, fromDir, toPath) {
   return linkOrCopy(fromDir, toPath);
 }
 
+// SessionStart runs a project-local session-start.sh: it cats suggest's always-on
+// body (preserving prior behavior) and appends an onboarding banner when the
+// workspace has no harness profile yet. We materialize the script next to the
+// shared base and point the hook at it. Any prior rsc SessionStart entry (the old
+// `cat …/suggest/SKILL.md` form or a previous script form) is dropped before we
+// add the current one — idempotent, and it migrates legacy hooks in place. Other
+// (non-rsc) SessionStart hooks are preserved.
 export function wireHook(paths) {
+  const scriptDest = join(paths.projectRoot, '.rsc', 'session-start.sh');
+  mkdirSync(dirname(scriptDest), { recursive: true });
+  copyFileSync(join(dirname(fileURLToPath(import.meta.url)), 'session-start.sh'), scriptDest);
+  chmodSync(scriptDest, 0o755);
+
+  const suggestMd = `${paths.skillDir('suggest')}/SKILL.md`;
+  const cmd = `bash "${scriptDest}" "${suggestMd}" "${paths.projectRoot}"`;
+
   const file = paths.hookTarget;
   const settings = existsSync(file) ? JSON.parse(readFileSync(file, 'utf8')) : {};
   settings.hooks ||= {};
   settings.hooks.SessionStart ||= [];
-  const cmd = `cat "${paths.skillDir('suggest')}/SKILL.md"`;
-  const already = JSON.stringify(settings.hooks.SessionStart).includes('skills/rsc/suggest');
-  if (!already) {
-    settings.hooks.SessionStart.push({ hooks: [{ type: 'command', command: cmd }] });
+  settings.hooks.SessionStart = settings.hooks.SessionStart.filter((e) => {
+    const s = JSON.stringify(e);
+    return !s.includes('skills/rsc/suggest') && !s.includes('.rsc/session-start.sh');
+  });
+  settings.hooks.SessionStart.push({ hooks: [{ type: 'command', command: cmd }] });
+
+  // Worklog checkpoint: PreCompact + SessionEnd run a project-local
+  // worklog-checkpoint.sh that reminds the agent to capture what we did this
+  // session into 02-DOCS/raw/worklog/ (the work-driven on-ramp). Silent when the
+  // workspace has no harness wiki. Registered idempotently on both events, with
+  // any prior rsc worklog-checkpoint entry dropped first.
+  const wlDest = join(paths.projectRoot, '.rsc', 'worklog-checkpoint.sh');
+  copyFileSync(join(dirname(fileURLToPath(import.meta.url)), 'worklog-checkpoint.sh'), wlDest);
+  chmodSync(wlDest, 0o755);
+  const wlCmd = `bash "${wlDest}" "${paths.projectRoot}"`;
+  for (const event of ['PreCompact', 'SessionEnd']) {
+    settings.hooks[event] ||= [];
+    settings.hooks[event] = settings.hooks[event].filter(
+      (e) => !JSON.stringify(e).includes('.rsc/worklog-checkpoint.sh'),
+    );
+    settings.hooks[event].push({ hooks: [{ type: 'command', command: wlCmd }] });
   }
+
   mkdirSync(dirname(file), { recursive: true });
   writeFileSync(file, JSON.stringify(settings, null, 2) + '\n');
-  return [file];
+  return [file, scriptDest, wlDest];
 }

package/targets/index.js

@@ -105,6 +105,7 @@ export function targetPaths(target, home = homedir(), cwd = process.cwd()) {
   const rootAbs = join(cwd, ...s.root.split('/'));
   return {
     root: rootAbs,
+    projectRoot: cwd,
     skillDir: (id) => (s.skillExt ? join(rootAbs, `${id}${s.skillExt}`) : join(rootAbs, id)),
     stateFile: join(rootAbs, '.rsc-state.json'),
     hookTarget: join(cwd, ...s.hook.split('/')),

package/targets/session-start.sh

@@ -0,0 +1,43 @@
+#!/usr/bin/env bash
+# rsc SessionStart payload (claude). Wired by targets/claude.js.
+#   $1 = absolute path to suggest's SKILL.md   $2 = absolute project root
+# Always emits suggest's always-on body; appends an onboarding banner when the
+# workspace has no harness profile yet, and an auto-ingest nudge when there is
+# un-ingested material waiting in the inbox.
+set -u
+
+cat "$1" 2>/dev/null
+
+profile="$2/02-DOCS/wiki/harness/user-profile.md"
+optout="$2/.rsc/.no-harness"
+
+if [ ! -f "$profile" ] && [ ! -f "$optout" ]; then
+  cat <<'BANNER'
+
+===== rsc onboarding =====
+Fresh setup: 02-DOCS/wiki/harness/user-profile.md is missing.
+ACTION: invoke `init` now (first contact: technical level + accompaniment dial) before the task.
+If the user does not want a harness here: create .rsc/.no-harness
+==========================
+BANNER
+fi
+
+# Auto-Ingest nudge: when a harness wiki exists and the inbox holds a real file
+# (anything other than README.md / dotfiles / the _processed archive), tell the
+# agent to run the Auto-Ingest Sweep. The hook only reminds; the agent does the
+# scan + ingest. Cheap signal here; the thorough workspace scan lives in the sweep.
+inbox="$2/02-DOCS/inbox"
+if [ -d "$2/02-DOCS/wiki" ] && [ -d "$inbox" ]; then
+  pending=$(find "$inbox" -maxdepth 1 -type f ! -name 'README.md' ! -name '.*' 2>/dev/null | head -1)
+  if [ -n "$pending" ]; then
+    cat <<'BANNER'
+
+===== rsc auto-ingest =====
+Un-ingested material is waiting in 02-DOCS/inbox/.
+ACTION: run the Auto-Ingest Sweep now — ingest inbox/, then scan the workspace
+(minus .rscignore) for un-ingested documents, recording them in wiki/.ingested.json.
+Originals are copied, never moved; deleting an emptied folder needs explicit consent.
+===========================
+BANNER
+  fi
+fi

package/targets/worklog-checkpoint.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+# rsc Worklog checkpoint payload (claude). Wired by targets/claude.js onto the
+# PreCompact and SessionEnd hooks.
+#   $1 = absolute project root
+# Reminds the agent to run a Worklog Sweep (capture what we did this session into
+# 02-DOCS/raw/worklog/ so the harness can compile it into the wiki). The hook only
+# REMINDS — the agent writes the worklog (Karpathy: the LLM writes the wiki).
+# Silent when this workspace has no harness wiki yet (nothing to document into).
+set -u
+
+root="${1:-$PWD}"
+
+# No harness wiki here → nothing to do. Stay silent.
+[ -d "$root/02-DOCS/wiki" ] || exit 0
+
+cat <<'BANNER'
+
+===== rsc worklog checkpoint =====
+If this session did meaningful work (files changed, a decision made, a commit),
+run a WORKLOG SWEEP before context is lost:
+  1. Write 02-DOCS/raw/worklog/<YYYY-MM-DD>-<slug>.md using the harness
+     wiki-worklog-template.md (what we did · why · files touched · outcome · next).
+  2. Compile it into wiki/ (update existing articles first; wikilinks + Related);
+     append significant decisions to 02-DOCS/wiki/harness/decisions.md.
+Skip entirely if this was a pure read/answer turn with no changes.
+==================================
+BANNER