obsidian-lean-coder 0.4.0 → 0.5.0

package/CHANGELOG.md

@@ -4,6 +4,42 @@ All notable changes to OLC (Obsidian Lean Coder) are recorded here.
 OLC is pre-release: versions are `0.x` until a controlled-testing release says otherwise.
 `package.json` shipped `"private": true` through 0.1.0; the flag was removed at publish time with explicit user approval, and 0.1.0 is live as [`obsidian-lean-coder`](https://www.npmjs.com/package/obsidian-lean-coder). Publishing a new version is still a separate explicitly approved action, never automatic.
 
+## [0.5.0] — 2026-10-14 — vault cross-linking for the second-brain graph
+
+Vault documents were isolated islands: no `[[wikilinks]]` connected intent to
+plan, checkpoints to history, or the daily log to the task it recorded. Since
+the vault is meant as the project's second brain, every bundled template now
+ships a `## Terkait` section with relative wikilinks to its natural
+neighbours (intent ↔ plan ↔ history ↔ checkpoint; explore reports and taste
+rules back to the project hub; daily logs to the hub and, per entry, to the
+task it concerns). `index.md` and `current-state.md` link to each other.
+`decision.md` takes a new `related` variable so its links reflect the actual
+decision (callers must supply it). `session-summary.md` stays link-free by
+design — it has no fixed vault location, so it is returned in chat, never
+written to a file.
+
+This only affects newly rendered documents; pre-existing vault files from
+earlier versions are not backfilled.
+
+- **Verification:** typecheck/build clean; full suite green (152 tests,
+  including a new end-to-end fixture that walks every markdown file in a
+  generated vault and resolves every `[[...]]` target on disk);
+  `npm pack --dry-run` file check.
+- **Upgrade:** `npm install obsidian-lean-coder@0.5.0`, then re-run any task
+  that still needs its docs drafted — existing docs keep their old content.
+
+## [0.4.1] — 2026-10-14 — verification tooling rule for minimal harnesses
+
+`agents/olc-build.md` gains a verification rule from a real field report: prefer
+tools guaranteed by OLC's own prerequisites (node, npm, git) over extra
+shellutils (cat/grep/awk/sed) a minimal harness may lack, and classify a check
+failure caused by a missing environment binary as an **environment limitation**,
+separate from a task failure. This fixes the *reporting* of such incidents, not
+the environment itself — no instruction can conjure a missing binary.
+
+- **Verification:** typecheck/build clean; full suite green; `npm pack --dry-run` file check.
+- **Upgrade:** `npm install obsidian-lean-coder@0.4.1`, then
+
 ## [0.4.0] — 2026-10-14 — default-like agents + project rules
 
 Agents `olc-plan`/`olc-build` rewritten per user request to behave exactly like

package/README.md

@@ -1,6 +1,6 @@
 # OLC — Obsidian Lean Coder
 
-OpenCode extension for project-scoped planning, memory, and safe Git checkpoints. On npm as `obsidian-lean-coder` (v0.4.0: default-like agents + project rules — see CHANGELOG).
+OpenCode extension for project-scoped planning, memory, and safe Git checkpoints. On npm as `obsidian-lean-coder` (v0.5.0: vault cross-linking for the second-brain graph — see CHANGELOG).
 
 ## Three separate steps
 
@@ -24,7 +24,7 @@ npm test         # tsc -p tsconfig.tests.json && node --test "dist-tests/tests/*
 npm run typecheck
 ```
 
-Expected: build clean, typecheck clean. `npm test` runs the full suite (currently 146 tests). `npm pack --dry-run` shows what would ship (`dist/`, `agents/`, `skills/`, `templates/`).
+Expected: build clean, typecheck clean. `npm test` runs the full suite (currently 152 tests). `npm pack --dry-run` shows what would ship (`dist/`, `agents/`, `skills/`, `templates/`).
 
 > Published on npm as `obsidian-lean-coder` (MIT). Republishing needs a separate explicit approval — never publish by accident.
 
@@ -115,6 +115,37 @@ Vault path precedence: explicit `vaultPath` > `OLC_VAULT_PATH` env > existing co
 
 Re-running init is always safe: run it again after pulling updates; `created` becomes `preserved`. To preview without writing: `initProject({ ..., dryRun: true })`.
 
+### Vault linking rules (second brain)
+
+The vault is meant as the project's long-term memory, so every document a
+template produces must be a node in Obsidian's link graph, not an isolated
+file. Rules:
+
+- Every bundled template carries a `## Terkait` section with **relative**
+  `[[wikilinks]]` to its natural neighbours — e.g. `intent.md` ↔ `plan.md` ↔
+  `history.md` ↔ `checkpoint.md`; `explore-report.md` and `taste-rule.md`
+  back to the project hub (`index.md`, `current-state.md`); each daily log
+  file links to the hub, and each logged event with a `taskId` links to that
+  task's `intent.md`/`plan.md`.
+- `index.md` and `current-state.md` link to each other — they are the hub a
+  reader lands on first.
+- Links are relative path-form (`[[../T-1/plan]]`, not a bare `[[plan]]`),
+  so they resolve correctly whether Obsidian's vault root is `<vault>` or
+  `<vault>/OLC/Projects/<id>`, and stay correct across multiple projects
+  sharing one vault (a bare `[[plan]]` would be ambiguous between projects).
+- `decision.md` takes a `related` template variable — the caller supplies
+  the links relevant to that specific decision; there is no fixed set.
+- `session-summary.md` is the one exception: it has no fixed vault location
+  (it's returned in chat for the user to read, never written to a file), so
+  it carries no links by design.
+- **Limitation:** linking only applies to newly rendered documents. Older
+  vault files written by a previous OLC version are not backfilled with
+  links (see D-005 in `docs/TASKS.md`).
+- Writing to `memory/<name>.md` via the `olc-memory` skill is free-form
+  (no template), but should still include a `## Terkait` section linking
+  back to `[[../index]]`/`[[../current-state]]` and any related task or
+  decision, to keep the memory graph connected.
+
 ## Updating
 
 1. `git pull` the OLC checkout.

package/agents/olc-build.md

@@ -32,6 +32,11 @@ checks. You never expand scope on your own.
   activation needs explicit user approval).
 - Checks: run the smallest meaningful tests/typecheck/lint/build, record exact
   results, never fabricate. Bounded retries, then stop with evidence + options.
+  Prefer tools guaranteed by OLC's own prerequisites (node, npm, git) over
+  extra shellutils (cat/grep/awk/sed) that a minimal harness may lack. If a
+  check fails because an environment binary/dependency is missing — not
+  because of the code — report it explicitly as an **environment limitation**,
+  separate from a task failure.
 - Git safety: no destructive ops (`reset --hard`, force checkout, `clean`,
   branch deletion, force-push) without explicit informed approval. Stage
   explicit paths only — never `git add .` / `-A`, never touch changes outside

package/dist/index.d.ts

@@ -10,5 +10,5 @@
  * `plugin.ts` tool `execute` (deferred to call time), so this re-export
  * cycle is safe under ESM live bindings.
  */
-export declare const OLC_VERSION = "0.4.0";
+export declare const OLC_VERSION = "0.5.0";
 export { default } from "./plugin.js";

package/dist/index.js

@@ -10,5 +10,5 @@
  * `plugin.ts` tool `execute` (deferred to call time), so this re-export
  * cycle is safe under ESM live bindings.
  */
-export const OLC_VERSION = "0.4.0";
+export const OLC_VERSION = "0.5.0";
 export { default } from "./plugin.js";

package/dist/init/index.js

@@ -26,8 +26,10 @@ const AGENTS_STUB = (projectId) => `# OLC project instructions — ${projectId}\
     "Bila bertentangan dengan perilaku default OpenCode, default yang menang — agen akan\n" +
     "menunjukkan konfliknya dan meminta keputusan Anda sebelum lanjut.\n\n" +
     "- (belum ada — tambahkan sesuai kebutuhan proyek)\n";
-const VAULT_INDEX = (projectId) => `# ${projectId}\n\nOLC project memory root. See current-state.md for the latest status.\n`;
-const VAULT_STATE = (projectId) => `# Current state — ${projectId}\n\nInitialized. No activity recorded yet.\n`;
+const VAULT_INDEX = (projectId) => `# ${projectId}\n\nOLC project memory root. See current-state.md for the latest status.\n` +
+    `\n## Terkait\n\n- [[current-state]] — status terbaru proyek\n`;
+const VAULT_STATE = (projectId) => `# Current state — ${projectId}\n\nInitialized. No activity recorded yet.\n` +
+    `\n## Terkait\n\n- [[index]] — hub memori proyek\n`;
 /** Derive a safe project_id proposal from the repository directory name. */
 export function proposeProjectId(repoRoot) {
     const cleaned = basename(resolve(repoRoot))

package/dist/log/index.js

@@ -70,7 +70,8 @@ function ensureDayFile(path, projectId, day) {
     // "wx" fails when another process won the race — that is fine, the file
     // (with an identical header) already exists and we just append below.
     try {
-        const header = withFrontmatter({ project: projectId, type: "activity-log", date: day }, "");
+        const header = withFrontmatter({ project: projectId, type: "activity-log", date: day }, "") +
+            "## Terkait\n\n- [[../index]] — hub memori proyek\n- [[../current-state]] — status terbaru proyek\n\n";
         appendFileSync(path, header.endsWith("\n") ? header : header + "\n", { flag: "wx" });
     }
     catch (error) {
@@ -91,6 +92,12 @@ function formatEntry(event, iso) {
         lines.push(`- sha: ${event.gitSha}`);
     if (redacted)
         lines.push(`- redacted: true`);
+    // Obsidian graph links toward the task docs. parseEntries does not know the
+    // `terkait` key and safely ignores it; it exists purely so every log entry
+    // is a clickable node in the vault graph (second-brain navigation).
+    if (event.taskId) {
+        lines.push(`- terkait: [[../tasks/${event.taskId}/intent]], [[../tasks/${event.taskId}/plan]]`);
+    }
     return lines.join("\n") + "\n";
 }
 /**

package/dist/planning/index.js

@@ -112,7 +112,10 @@ function appendHistory(vaultRoot, projectId, taskId, kind, reason, date) {
         existing = "";
     }
     const entry = `## ${date} — ${kind}\n\n${reason}\n`;
-    writeVaultDoc(history, existing === "" ? `# History — ${taskId}\n\n${entry}` : `${existing}\n${entry}`, {
+    const header = `# History — ${taskId}\n\n## Terkait\n\n` +
+        `- [[../${taskId}/intent]] — maksud kerja\n` +
+        `- [[../${taskId}/plan]] — rencana kerja\n\n`;
+    writeVaultDoc(history, existing === "" ? `${header}${entry}` : `${existing}\n${entry}`, {
         overwrite: true,
     });
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "obsidian-lean-coder",
-  "version": "0.4.0",
+  "version": "0.5.0",
   "description": "Obsidian Lean Coder — OpenCode extension for project-scoped planning, memory, and safe Git checkpoints.",
   "type": "module",
   "main": "./dist/index.js",

package/skills/olc-explore/SKILL.md

@@ -1,35 +1,35 @@
 ---
 name: olc-explore
-description: Selidiki request yang kompleks/tidak pasti sebelum planning. Read-only terhadap app code; menghasilkan satu laporan eksplorasi di vault.
+description: Investigate a complex/uncertain request before planning. Read-only against app code; produces a single exploration report in the vault.
 ---
 
 # olc-explore
 
-Klarifikasi request yang kompleks atau tidak pasti **sebelum** `intent.md`/`plan.md` dibuat.
-Untuk task sederhana yang sudah jelas, lewati skill ini dan langsung ke Plan (direct Plan).
+Clarify a complex or uncertain request **before** `intent.md`/`plan.md` are created.
+For simple, already-clear tasks, skip this skill and go straight to Plan (direct Plan).
 
-## Kapan dipakai vs direct Plan
+## When to use vs direct Plan
 
-- Pakai Explore bila: request ambigu, menyentuh area yang belum dikenal, ada beberapa pendekatan dengan tradeoff nyata, atau risiko salah paham tinggi.
-- Direct Plan bila: perubahan kecil dan terlokalisasi, pendekatan sudah jelas, tidak ada pertanyaan material yang terbuka.
-- Pilihan user yang menang: kalau user minta direct Plan, jangan memaksa Explore, dan sebaliknya.
+- Use Explore when: the request is ambiguous, touches unfamiliar territory, has multiple approaches with real tradeoffs, or carries a high risk of misunderstanding.
+- Use direct Plan when: the change is small and localized, the approach is already clear, there are no material open questions.
+- The user's choice wins: if the user asks for direct Plan, don't force Explore, and vice versa.
 
-## Aturan keras
+## Hard rules
 
-1. **Read-only terhadap app source.** Boleh membaca file, menjalankan perintah baca (status, log, test baca), menelusuri kode. DILARANG mengubah, menambah, atau menghapus file app dalam bentuk apa pun — tidak ada fix "sekalian", tidak ada refactor kecil.
-2. Satu laporan aktif per task: `tasks/<task-id>/explore/report.md` di vault proyek yang sudah tervalidasi (OLC-021). Jangan membuat file laporan tandingan.
-3. Setiap klaim temuan harus punya bukti: path file + baris, output perintah, atau kutipan config. Tanpa bukti = pertanyaan terbuka, bukan temuan.
-4. Revisi tidak menghapus penalaran penting: revisi ditambahkan sebagai bagian baru bertanggal di laporan yang sama; alasan perubahan pendekatan harus tercatat.
+1. **Read-only against app source.** Allowed: reading files, running read-only commands (status, log, read-only tests), tracing code. FORBIDDEN: changing, adding, or removing app files in any way — no "while I'm at it" fixes, no small refactors.
+2. One active report per task: `tasks/<task-id>/explore/report.md` in the already-validated project vault (OLC-021). Do not create a competing report file.
+3. Every finding claim needs evidence: a file path + line, command output, or a config quote. No evidence = open question, not a finding.
+4. Revisions never erase important reasoning: a revision is added as a new dated section in the same report; the reason for the approach change must be recorded.
 
-## Alur
+## Flow
 
-1. Validasi identitas proyek dan baca `project_id` dari `.olc/config.json` (jangan operasi vault sebelum ini lolos).
-2. Baca request user + konteks relevan secukupnya (index/current-state vault bila ada, file app terkait). Jangan membaca seluruh vault.
-3. Tulis laporan via template `explore-report` (var: `task_id`, `project_id`, `date`, `request`, `findings`, `options`, `recommendation`). Buat direktori `explore/` bila belum ada; jangan menimpa laporan yang sudah dikustom tanpa persetujuan (ikuti aturan create-only OLC-023).
-4. Bila request berubah atau temuan baru muncul: tambah bagian revisi bertanggal di laporan yang sama, catat alasan di `history.md` task.
-5. Serahkan ke Plan: laporan explore adalah input `draftTaskDocs`, bukan pengganti `intent.md`/`plan.md`. Tanpa kedua dokumen itu disetujui, tidak ada coding (OLC-032).
+1. Validate project identity and read `project_id` from `.olc/config.json` (no vault operation passes before this).
+2. Read the user's request plus just enough relevant context (vault index/current-state if present, related app files). Do not read the entire vault.
+3. Write the report via the `explore-report` template (vars: `task_id`, `project_id`, `date`, `request`, `findings`, `options`, `recommendation`). Create the `explore/` directory if missing; never overwrite an already-customized report without approval (follow OLC-023's create-only rule).
+4. If the request changes or new findings appear: add a new dated revision section to the same report, record the reason in the task's `history.md`.
+5. Hand off to Plan: the explore report is input to `draftTaskDocs`, not a replacement for `intent.md`/`plan.md`. No coding happens until both documents are approved (OLC-032).
 
-## Verifikasi skill ini
+## Verifying this skill
 
-- Setelah Explore: `git status --short` di repo app harus bersih (tidak ada perubahan app).
-- Laporan ada di path yang benar dan semua var template terisi (tidak ada `{{...}}` tersisa).
+- After Explore: `git status --short` in the app repo must be clean (no app changes).
+- The report exists at the correct path and every template var is filled (no leftover `{{...}}`).

package/skills/olc-init/SKILL.md

@@ -1,39 +1,40 @@
 ---
 name: olc-init
-description: Daftarkan satu repo app + vault ke OLC secara aman dan idempotent. Create-only — tidak pernah menimpa file yang sudah ada.
+description: Safely and idempotently register one app repo + vault with OLC. Create-only — never overwrites existing files.
 ---
 
 # olc-init
 
-Inisialisasi proyek OLC: mengusulkan `project_id`, mengonfirmasi vault path, menyiapkan `.olc/` dan struktur vault, lalu mendaftarkan pasangan repo↔vault di registry.
+Initialize an OLC project: propose a `project_id`, confirm the vault path, set up `.olc/` and the vault structure, then register the repo↔vault pair in the registry.
 
-## Aturan keras
+## Hard rules
 
-1. **Jelaskan dulu sebelum menulis.** Sebutkan file apa saja yang akan dibuat sebelum memanggil `initProject`.
-2. **Create-only.** File yang sudah ada dilaporkan `preserved` — tidak pernah ditimpa. Yang butuh keputusan manusia dikembalikan sebagai `status: "needs-decision"` dengan pesan yang bisa ditindaklanjuti.
-3. **Tidak ada tebakan vault.** Urutan: `vaultPath` eksplisit > env `OLC_VAULT_PATH` > config yang ada. Bila tidak ada, minta ke user — jangan membuat lokasi default diam-diam.
-4. **Tidak ada prompting di dalam modul.** Kumpulkan jawaban dulu, lalu panggil ulang dengan parameter lengkap.
+1. **Explain before writing.** State which files will be created before calling `initProject`.
+2. **Create-only.** Existing files are reported as `preserved` — never overwritten. Anything needing a human decision comes back as `status: "needs-decision"` with an actionable message.
+3. **No guessing the vault.** Precedence: explicit `vaultPath` > env `OLC_VAULT_PATH` > existing config. If none exist, ask the user — never silently create a default location.
+4. **No prompting inside the module.** Gather answers first, then call again with the full parameters.
 
-## Alur
+## Flow
 
-1. Panggil `initProject({ projectRoot })` — dapatkan usulan `project_id` (turunan nama repo yang disanitasi).
-2. Bila `status: "needs-decision"`, tanyakan ke user: konfirmasi `projectId`, isi `vaultPath`, atau selesaikan duplikat id sesuai pesan.
-3. Panggil ulang dengan `{ projectRoot, projectId, vaultPath }` hingga `status: "ready"`.
-4. Tanyakan ke user: "Apakah ada aturan tambahan proyek selain yang ada di
-   AGENTS.md utama?" Bila ada, tulis ke bagian "Aturan tambahan proyek" di
-   `<repo>/.olc/AGENTS.md` (file itu create-only — bila sudah ada dan sudah
-   berisi aturan, jangan ditimpa; gabungkan hanya bila user meminta).
-5. Laporkan ringkasan `created` / `preserved` / `attention` ke user dalam bahasa sederhana.
-5. Untuk pratinjau tanpa menulis: `initProject({ ..., dryRun: true })` — `created` berarti "akan dibuat".
+1. Call `initProject({ projectRoot })` — get a proposed `project_id` (sanitized derivative of the repo name).
+2. If `status: "needs-decision"`, ask the user: confirm `projectId`, fill in `vaultPath`, or resolve a duplicate id per the message.
+3. Call again with `{ projectRoot, projectId, vaultPath }` until `status: "ready"`.
+4. Ask the user: "Are there any additional project rules beyond what's in the
+   main AGENTS.md?" If yes, write them to the "Additional project rules"
+   section of `<repo>/.olc/AGENTS.md` (that file is create-only — if it
+   already exists and already has rules, don't overwrite; merge only if the
+   user asks).
+5. Report a `created` / `preserved` / `attention` summary to the user in plain language.
+5. To preview without writing: `initProject({ ..., dryRun: true })` — `created` means "would be created".
 
-## Yang dibuat (hanya bila belum ada)
+## What gets created (only if missing)
 
-- `<repo>/.olc/config.json` — `{ schemaVersion, projectId, vaultPath }`. Kunci berbau kredensial (`secret|token|password|api-key`) ditolak.
-- `<repo>/.olc/AGENTS.md` — stub instruksi, minta user meninjau dan menyesuaikan.
+- `<repo>/.olc/config.json` — `{ schemaVersion, projectId, vaultPath }`. Keys that look like credentials (`secret|token|password|api-key`) are rejected.
+- `<repo>/.olc/AGENTS.md` — instruction stub, asks the user to review and adjust.
 - `<vault>/OLC/Projects/<project_id>/{index.md, current-state.md, decisions/, tasks/, logs/, memory/, taste/}`
-- Entri `{ projects: { "<id>": { repoRoot } } }` di `<vault>/OLC/registry.json`.
+- An entry `{ projects: { "<id>": { repoRoot } } }` in `<vault>/OLC/registry.json`.
 
-## Verifikasi skill ini
+## Verifying this skill
 
-- Jalankan init dua kali di fixture repo/vault sementara: larian kedua semua `preserved`, tidak ada byte berubah.
-- Init dengan config + instruksi kustom yang sudah ada: file kustom tetap identik.
+- Run init twice against a temporary fixture repo/vault: the second run is all `preserved`, no byte changes.
+- Init with existing config + custom instructions: the custom file stays byte-identical.

package/skills/olc-log/SKILL.md

@@ -1,26 +1,26 @@
 ---
 name: olc-log
-description: Catat event penting, checkpoint, dan ringkasan sesi ke log aktivitas harian yang tahan lama. Terverifikasi anti rusak dan anti bocor rahasia.
+description: Record important events, checkpoints, and session summaries into a durable daily activity log. Verified crash-safe and secret-leak-safe.
 ---
 
 # olc-log
 
-Log aktivitas abadi per proyek di `<vault>/OLC/Projects/<project_id>/logs/YYYY-MM-DD.md` (satu file per hari — itulah rotasinya, tanpa rotasi ukuran untuk MVP).
+Durable per-project activity log at `<vault>/OLC/Projects/<project_id>/logs/YYYY-MM-DD.md` (one file per day — that's its only rotation, no size-based rotation for MVP).
 
-## Aturan keras
+## Hard rules
 
-1. **Hanya event eksplisit.** Modul tidak hook ke tool call — tidak ada noise per-tool-call secara konstruksi. Catat hanya milestone, checkpoint, verifikasi, dan ringkasan sesi.
-2. **Rahasia disensor, tidak disimpan.** Pola kredensial yang dikenal di summary/details diganti `[redacted]` dan entri ditandai.
-3. **Append aman.** Satu write `O_APPEND` per entri; interupsi tidak pernah merusak entri sebelumnya. File tanpa newline akhir (sisa crash) dinormalisasi sebelum append berikutnya.
-4. **Identitas dulu.** Vault salah → throw dengan nol file dibuat.
+1. **Explicit events only.** The module never hooks tool calls — no per-tool-call noise by construction. Log only milestones, checkpoints, verifications, and session summaries.
+2. **Secrets are redacted, never stored.** Known credential patterns in summary/details are replaced with `[redacted]` and the entry is flagged.
+3. **Safe appends.** One `O_APPEND` write per entry; an interruption never corrupts earlier entries. A file missing its trailing newline (crash remnant) is normalized before the next append.
+4. **Identity first.** Wrong vault → throw with zero files created.
 
-## Alur
+## Flow
 
-1. Tulis: `logEvent({ projectRoot, config, event: { kind, summary, taskId?, details?, gitSha?, date? } })` — `kind` mis. `milestone|checkpoint|verify|session`; `summary` non-kosong wajib.
-2. Baca: `readLogDay(vaultRoot, projectId, day)` atau `readLastEvents(vaultRoot, projectId, limit)` (urutan terbaru dulu — verifikasi terakhir terlihat di sana).
-3. Akhir sesi normal: panggil `buildSessionSummary` (modul `session`) — teks Indonesia via template `session-summary` + footer jujur hard-kill, klaim faktual hanya dari `VerifyOutcome`/SHA/checkpoint nyata — lalu tampilkan ke user. Sesi yang mati mendadak tidak menghasilkan ringkasan; event log terakhir adalah titik pemulihan (lihat resume).
+1. Write: `logEvent({ projectRoot, config, event: { kind, summary, taskId?, details?, gitSha?, date? } })` — `kind` e.g. `milestone|checkpoint|verify|session`; non-empty `summary` is required.
+2. Read: `readLogDay(vaultRoot, projectId, day)` or `readLastEvents(vaultRoot, projectId, limit)` (newest first — the last verification shows up there).
+3. Normal session end: call `buildSessionSummary` (module `session`) — rendered via the `session-summary` template with an honest hard-kill footer, factual claims sourced only from a real `VerifyOutcome`/SHA/checkpoint — then show it to the user. A session that dies abruptly produces no summary; the last log event is the recovery point (see resume).
 
-## Verifikasi skill ini
+## Verifying this skill
 
-- Rahasia pola `sk-...`, `ghp_...`, `AKIA...` di summary tidak pernah sampai ke vault (jadi `[redacted]`).
-- Interupsi di tengah append tidak merusak entri sebelumnya.
+- Secret patterns like `sk-...`, `ghp_...`, `AKIA...` in a summary never reach the vault (they become `[redacted]`).
+- An interruption mid-append never corrupts earlier entries.

package/skills/olc-memory/SKILL.md

@@ -1,26 +1,26 @@
 ---
 name: olc-memory
-description: Cari dan perbarui catatan memori proyek secara progresif — baca yang kecil dulu, hanya yang relevan dengan query. Tidak pernah memindai seluruh vault.
+description: Search and update project memory notes progressively — read small first, only what's relevant to the query. Never scans the whole vault.
 ---
 
 # olc-memory
 
-Ingatan berlingkup proyek: baca `index.md` + `current-state.md` dulu, lalu hanya dokumen `memory/` + `decisions/`, doc task, dan event log yang relevan dengan query. Tulis balik hanya di scope proyek yang tervalidasi.
+Project-scoped memory: read `index.md` + `current-state.md` first, then only the `memory/` + `decisions/` documents, task docs, and log events relevant to the query. Write back only within the validated project scope.
 
-## Aturan keras
+## Hard rules
 
-1. **Identitas dulu.** `validateIdentity` harus lolos sebelum baca/tulis apa pun. Proyek salah atau vault tak tersedia → gagal dengan nol file tersentuh.
-2. **Tidak memindai folder proyek lain.** Semua path di-resolve di dalam `<vault>/OLC/Projects/<id>/`; folder lain tidak tersentuh secara konstruksi.
-3. **Tulis terkontrol.** `recordMemory` butuh nama polos (huruf/angka/dash/underscore, tanpa path/ekstensi) + konten non-kosong; file yang sudah ada tidak ditimpa diam-diam (`overwrite` eksplisit).
+1. **Identity first.** `validateIdentity` must pass before any read/write. Wrong project or unavailable vault → fail with zero files touched.
+2. **Never scans other projects' folders.** Every path resolves inside `<vault>/OLC/Projects/<id>/`; other folders are untouched by construction.
+3. **Controlled writes.** `recordMemory` requires a plain name (letters/digits/dash/underscore, no path/extension) plus non-empty content; existing files are never silently overwritten (`overwrite` must be explicit).
 
-## Alur
+## Flow
 
-1. Panggil `recallMemory({ projectRoot, config, query?, taskId?, includeLogs?, logLimit?, limit? })` — tanpa query = konteks dasar (index + state + log terbaru, tanpa filter).
-2. Dengan query: token query difilter stopwords, skor per dokumen `memory/` + `decisions/` (bobot nama file), kutipan dibatasi 1200 karakter dengan flag truncasi, urutan skor terbaik dulu.
-3. Dengan `taskId`: sertakan status doc intent/plan task itu (hilang → null, bukan error).
-4. Untuk menyimpan: `recordMemory({ projectRoot, config, name, content, overwrite? })` → `memory/<name>.md` atomik.
+1. Call `recallMemory({ projectRoot, config, query?, taskId?, includeLogs?, logLimit?, limit? })` — no query = base context (index + state + recent log, unfiltered).
+2. With a query: query tokens are filtered for stopwords, each `memory/` + `decisions/` document is scored (filename weighted), excerpts are capped at 1200 characters with a truncation flag, best score first.
+3. With `taskId`: include that task's intent/plan doc status (missing → null, not an error).
+4. To save: `recordMemory({ projectRoot, config, name, content, overwrite? })` → atomic write to `memory/<name>.md`. `content` must include a `## Terkait` section with relative wikilinks to `[[../index]]`, `[[../current-state]]`, and any related task/decision document — this keeps the memory note connected in the vault graph (second brain) instead of an isolated island.
 
-## Verifikasi skill ini
+## Verifying this skill
 
-- Ingatan proyek A tidak pernah muncul saat recall di proyek B (isolasi multi-proyek).
-- Setiap klaim relevansi bisa ditelusur ke dokumen + barisnya, bukan karangan.
+- Project A's memory never shows up on recall in project B (multi-project isolation).
+- Every relevance claim is traceable to a document + line, not invented.

package/skills/olc-taste/SKILL.md

@@ -1,26 +1,26 @@
 ---
 name: olc-taste
-description: Pelajari preferensi khas proyek tanpa mengubah perilaku diam-diam. Sinyal berulang jadi usulan pending; hanya persetujuan user yang mengaktifkan.
+description: Learn project-specific preferences without silently changing behavior. Repeated signals become pending proposals; only user approval activates them.
 ---
 
 # olc-taste
 
-Selera proyek dipelajari tanpa silent behavior change: sinyal berulang dicatat sebagai proposal `pending` lengkap dengan contoh/bukti, dan hanya keputusan user eksplisit yang mengubahnya jadi `active`. Penolakan memarkirnya sebagai `rejected` (riwayat, tidak pernah dipakai). Rule `pending` tidak pernah diperlakukan sebagai aktif oleh konsumen mana pun.
+Project taste is learned without silent behavior change: a repeated signal is recorded as a `pending` proposal complete with examples/evidence, and only an explicit user decision turns it `active`. Rejection parks it as `rejected` (kept as history, never used). A `pending` rule is never treated as active by any consumer.
 
-## Aturan keras
+## Hard rules
 
-1. **Hanya user yang mengaktifkan.** Tidak ada jalur otomatis pending→active. Review terminal: `active`/`rejected` tidak bisa di-review ulang (`BAD_TRANSITION`).
-2. **Scope satu proyek.** Semua path di dalam `taste/` proyek tervalidasi; proyek lain tidak tersentuh. Modul ini tidak pernah menyentuh `tasks/` — intent/plan yang disetujui tidak bisa tertimpa dari sini.
-3. **Usulan butuh bukti.** Tanpa ≥1 contoh/sinyal berulang, usulan ditolak langsung (`INVALID_INPUT`). Duplikat tidak digabung diam-diam (`ALREADY_EXISTS`).
+1. **Only the user activates.** There is no automatic pending→active path. Review is terminal: `active`/`rejected` cannot be re-reviewed (`BAD_TRANSITION`).
+2. **Single-project scope.** Every path lives inside the validated project's `taste/`; other projects are untouched. This module never touches `tasks/` — approved intent/plan can't be overwritten from here.
+3. **Proposals need evidence.** Without ≥1 example/repeated signal, a proposal is rejected outright (`INVALID_INPUT`). Duplicates are never silently merged (`ALREADY_EXISTS`).
 
-## Alur
+## Flow
 
-1. Usulkan: `proposeTasteRule({ projectRoot, config, name, rule, examples[] })` → `taste/<name>.md` status `pending` via template `taste-rule`, write atomik create-only.
-2. Baca: `getTasteRule(...)` / `listTasteRules({ status? })` — hanya di scope tervalidasi, urut nama.
-3. Review eksplisit bersama user: `reviewTasteRule({ projectRoot, config, name, approve })` — `true` = pending→active, `false` = pending→rejected. Nama tak dikenal → `UNKNOWN_RULE`.
+1. Propose: `proposeTasteRule({ projectRoot, config, name, rule, examples[] })` → `taste/<name>.md` with status `pending`, via the `taste-rule` template, atomic create-only write.
+2. Read: `getTasteRule(...)` / `listTasteRules({ status? })` — only within the validated scope, name-sorted.
+3. Explicit review with the user: `reviewTasteRule({ projectRoot, config, name, approve })` — `true` = pending→active, `false` = pending→rejected. Unknown name → `UNKNOWN_RULE`.
 
-## Verifikasi skill ini
+## Verifying this skill
 
-- Rule `pending` tidak muncul di daftar `active`.
-- `approve` mengaktifkan, `reject` memarkir; review ulang status terminal gagal.
-- Intent/plan task byte-identik sebelum dan sesudah siklus propose→review.
+- A `pending` rule never appears in the `active` list.
+- `approve` activates, `reject` parks; re-reviewing a terminal status fails.
+- Task intent/plan stay byte-identical before and after a propose→review cycle.

package/templates/activity-log.md

@@ -5,6 +5,11 @@ date: {{date}}
 
 # Activity log — {{date}}
 
+## Terkait (dokumen terkait)
+
+- [[../index]] — hub memori proyek
+- [[../current-state]] — status terbaru proyek
+
 ## {{event}} ({{timestamp}})
 
 {{details}}

package/templates/checkpoint.md

@@ -22,3 +22,9 @@ date: {{date}}
 ## Timestamp
 
 {{date}}
+
+## Terkait (dokumen terkait)
+
+- [[../{{task_id}}/intent]] — maksud kerja
+- [[../{{task_id}}/plan]] — rencana kerja
+- [[../{{task_id}}/history]] — riwayat revisi keputusan

package/templates/decision.md

@@ -17,3 +17,7 @@ date: {{date}}
 ## Consequences (akibat / yang harus diikuti)
 
 {{consequences}}
+
+## Terkait (dokumen terkait)
+
+{{related}}

package/templates/explore-report.md

@@ -21,3 +21,9 @@ date: {{date}}
 ## Recommendation (pilihan yang disarankan + alasan)
 
 {{recommendation}}
+
+## Terkait (dokumen terkait)
+
+- [[../../{{task_id}}/intent]] — maksud kerja
+- [[../../{{task_id}}/plan]] — rencana kerja
+- [[../../index]] — hub memori proyek

package/templates/intent.md

@@ -21,3 +21,9 @@ project: {{project_id}}
 ## Open questions
 
 {{open_questions}}
+
+## Terkait (dokumen terkait)
+
+- [[../{{task_id}}/plan]] — rencana kerja untuk intent ini
+- [[../../index]] — hub memori proyek
+- [[../../current-state]] — status terbaru proyek

package/templates/plan.md

@@ -25,3 +25,10 @@ project: {{project_id}}
 ## Risks and open questions
 
 {{risks}}
+
+## Terkait (dokumen terkait)
+
+- [[../{{task_id}}/intent]] — maksud dari rencana ini
+- [[../{{task_id}}/history]] — riwayat revisi keputusan
+- [[../../index]] — hub memori proyek
+- [[../../current-state]] — status terbaru proyek

package/templates/taste-rule.md

@@ -12,3 +12,8 @@ status: {{status}}
 ## Examples / evidence (contoh + bukti)
 
 {{examples}}
+
+## Terkait (dokumen terkait)
+
+- [[../index]] — hub memori proyek
+- [[../current-state]] — status terbaru proyek