jeo-code 0.7.15 → 0.7.16

package/CHANGELOG.md

@@ -7,6 +7,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 The README mirrors the latest 5 entries — regenerate with `bun run changelog:sync`.
 
 
+## [0.7.16] - 2026-06-25
+_Bundled skills get a safe installer: `jeo skills sync` materializes the built-in workflow skills into `~/.jeo/skills`, preserving local edits by default, with a CI-friendly `--check` drift report and a `--force` overwrite._
+
+### Added
+- **`jeo skills sync` installs and reconciles the bundled workflow skills into `~/.jeo/skills`.** New `syncBundledSkills(dir, { check, force })` (`src/commands/skills.ts`) reconciles the built-in `SKILLS` against a target dir — defaulting to `userSkillsDir()` (`~/.jeo/skills`, honoring `JEO_CONFIG_DIR`/`$HOME`, matching the highest-precedence flat dir in `skillDirs`). Missing skills are installed; a differing local copy is **preserved** by default (gjc `setup defaults` parity) unless `--force` overwrites it. `--check` is a pure drift report that writes nothing and sets a non-zero exit code on drift, so CI can gate on bundled-skill staleness. `--json` emits the structured `SkillSyncResult` (per-skill `status`/`action`, `drift`, `wrote`, `mode`). A new `bundledSkillFileContent` helper (`src/skills/catalog.ts`) defines the canonical on-disk bytes (raw `SKILL.md` when present, decorated render otherwise) shared by `sync` and the existing `--write` path. The `skills` command summary/usage (`src/cli/runner.ts`) and the list footer now advertise the subcommand.
+
+### Verified
+- `bun run typecheck` clean; full suite green — 1960 pass / 0 fail across 233 files — including the new `test/skills-sync.test.ts` (fresh install, idempotent re-run, `--check` drift report without writing, default-preserve vs `--force` overwrite, single-file reinstall, non-zero `--check` exit via `JEO_CONFIG_DIR`, and `--json` shape).
+
+
 ## [0.7.15] - 2026-06-25
 _Rollback release: the runtime source and test suite are restored to the **0.7.9** state. The intervening 0.7.10–0.7.14 runtime changes (quiet-mode/`-q`, `/resume` transcript rework, `mutatedFiles` plumbing, Ollama `num_ctx`, tmux REPL reaping, and related engine/provider/TUI edits) are reverted. Published as a new version because npm cannot re-issue an existing version number; `latest` now serves code identical to 0.7.9._
 

package/README.ja.md

@@ -205,11 +205,11 @@ CI は `.github/workflows/npm-publish.yml` で公開します — GitHub リリ
 ## 変更履歴 (Changelog)
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.7.16]** (2026-06-25) — Bundled skills get a safe installer: `jeo skills sync` materializes the built-in workflow skills into `~/.jeo/skills`, preserving local edits by default, with a CI-friendly `--check` drift report and a `--force` overwrite.
 - **[0.7.15]** (2026-06-25) — Rollback release: the runtime source and test suite are restored to the **0.7.9** state. The intervening 0.7.10–0.7.14 runtime changes (quiet-mode/`-q`, `/resume` transcript rework, `mutatedFiles` plumbing, Ollama `num_ctx`, tmux REPL reaping, and related engine/provider/TUI edits) are reverted. Published as a new version because npm cannot re-issue an existing version number; `latest` now serves code identical to 0.7.9.
 - **[0.7.14]** (2026-06-25) — Quieter, cleaner output: a new `-q`/`--quiet` flag strips launch banners and courtesy logs, `/resume` no longer dumps walls of raw escaped JSON into scrollback, and a turn that edits files now reliably surfaces those paths to orchestrators.
 - **[0.7.9]** (2026-06-24) — Local model providers actually run, and OAuth/login hygiene: Ollama requests now carry an explicit `num_ctx` so jeo's large system prompt no longer overflows Ollama's small default window (the "ollama 모델이 안 돌아간다" blocker), LM Studio reasoning models that emit the whole reply on the reasoning channel are recovered instead of dying empty, GGUF chat templates that can't render the `tools` array are retried with native tools stripped, the OAuth success tab auto-closes with a clear hint, the effective-credential picker prefers a verified OAuth login over a stray API key, and abandoned idle `jeo` tmux REPL sessions are reaped on launch.
 - **[0.7.8]** (2026-06-24) — Long-session performance & resource hygiene: the detached memory-distill worker now always terminates itself (closing the `jeo memory-distill` orphan pileup that pinned each session's transcript in RSS), the live reasoning/output block re-wraps only its trailing window (O(tail) per frame instead of O(len²) over a long stream), the team loop stops reprinting the todo guide every iteration, and the readline caret stays aligned after an image attach.
-- **[0.7.7]** (2026-06-23) — Long-running-process hygiene & input robustness: backgrounded grandchildren (`next dev &`, daemons) are now reaped at the turn boundary via per-command process groups — closing the climbing `next-server` RSS leak — while the prompt gains recursive fuzzy `@path` search, ANSI-safe and idle-merged bracketed paste, a configurable OSC 52 clipboard cap, per-role ralplan model routing, and leaked-reasoning-tag stripping on salvaged answers.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/README.ko.md

@@ -205,11 +205,11 @@ CI는 `.github/workflows/npm-publish.yml`로 배포합니다 — GitHub 릴리
 ## 변경 이력 (Changelog)
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.7.16]** (2026-06-25) — Bundled skills get a safe installer: `jeo skills sync` materializes the built-in workflow skills into `~/.jeo/skills`, preserving local edits by default, with a CI-friendly `--check` drift report and a `--force` overwrite.
 - **[0.7.15]** (2026-06-25) — Rollback release: the runtime source and test suite are restored to the **0.7.9** state. The intervening 0.7.10–0.7.14 runtime changes (quiet-mode/`-q`, `/resume` transcript rework, `mutatedFiles` plumbing, Ollama `num_ctx`, tmux REPL reaping, and related engine/provider/TUI edits) are reverted. Published as a new version because npm cannot re-issue an existing version number; `latest` now serves code identical to 0.7.9.
 - **[0.7.14]** (2026-06-25) — Quieter, cleaner output: a new `-q`/`--quiet` flag strips launch banners and courtesy logs, `/resume` no longer dumps walls of raw escaped JSON into scrollback, and a turn that edits files now reliably surfaces those paths to orchestrators.
 - **[0.7.9]** (2026-06-24) — Local model providers actually run, and OAuth/login hygiene: Ollama requests now carry an explicit `num_ctx` so jeo's large system prompt no longer overflows Ollama's small default window (the "ollama 모델이 안 돌아간다" blocker), LM Studio reasoning models that emit the whole reply on the reasoning channel are recovered instead of dying empty, GGUF chat templates that can't render the `tools` array are retried with native tools stripped, the OAuth success tab auto-closes with a clear hint, the effective-credential picker prefers a verified OAuth login over a stray API key, and abandoned idle `jeo` tmux REPL sessions are reaped on launch.
 - **[0.7.8]** (2026-06-24) — Long-session performance & resource hygiene: the detached memory-distill worker now always terminates itself (closing the `jeo memory-distill` orphan pileup that pinned each session's transcript in RSS), the live reasoning/output block re-wraps only its trailing window (O(tail) per frame instead of O(len²) over a long stream), the team loop stops reprinting the todo guide every iteration, and the readline caret stays aligned after an image attach.
-- **[0.7.7]** (2026-06-23) — Long-running-process hygiene & input robustness: backgrounded grandchildren (`next dev &`, daemons) are now reaped at the turn boundary via per-command process groups — closing the climbing `next-server` RSS leak — while the prompt gains recursive fuzzy `@path` search, ANSI-safe and idle-merged bracketed paste, a configurable OSC 52 clipboard cap, per-role ralplan model routing, and leaked-reasoning-tag stripping on salvaged answers.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/README.md

@@ -207,11 +207,11 @@ Huge thanks to [gajae-code](https://github.com/Yeachan-Heo/gajae-code) for the i
 ## Changelog
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.7.16]** (2026-06-25) — Bundled skills get a safe installer: `jeo skills sync` materializes the built-in workflow skills into `~/.jeo/skills`, preserving local edits by default, with a CI-friendly `--check` drift report and a `--force` overwrite.
 - **[0.7.15]** (2026-06-25) — Rollback release: the runtime source and test suite are restored to the **0.7.9** state. The intervening 0.7.10–0.7.14 runtime changes (quiet-mode/`-q`, `/resume` transcript rework, `mutatedFiles` plumbing, Ollama `num_ctx`, tmux REPL reaping, and related engine/provider/TUI edits) are reverted. Published as a new version because npm cannot re-issue an existing version number; `latest` now serves code identical to 0.7.9.
 - **[0.7.14]** (2026-06-25) — Quieter, cleaner output: a new `-q`/`--quiet` flag strips launch banners and courtesy logs, `/resume` no longer dumps walls of raw escaped JSON into scrollback, and a turn that edits files now reliably surfaces those paths to orchestrators.
 - **[0.7.9]** (2026-06-24) — Local model providers actually run, and OAuth/login hygiene: Ollama requests now carry an explicit `num_ctx` so jeo's large system prompt no longer overflows Ollama's small default window (the "ollama 모델이 안 돌아간다" blocker), LM Studio reasoning models that emit the whole reply on the reasoning channel are recovered instead of dying empty, GGUF chat templates that can't render the `tools` array are retried with native tools stripped, the OAuth success tab auto-closes with a clear hint, the effective-credential picker prefers a verified OAuth login over a stray API key, and abandoned idle `jeo` tmux REPL sessions are reaped on launch.
 - **[0.7.8]** (2026-06-24) — Long-session performance & resource hygiene: the detached memory-distill worker now always terminates itself (closing the `jeo memory-distill` orphan pileup that pinned each session's transcript in RSS), the live reasoning/output block re-wraps only its trailing window (O(tail) per frame instead of O(len²) over a long stream), the team loop stops reprinting the todo guide every iteration, and the readline caret stays aligned after an image attach.
-- **[0.7.7]** (2026-06-23) — Long-running-process hygiene & input robustness: backgrounded grandchildren (`next dev &`, daemons) are now reaped at the turn boundary via per-command process groups — closing the climbing `next-server` RSS leak — while the prompt gains recursive fuzzy `@path` search, ANSI-safe and idle-merged bracketed paste, a configurable OSC 52 clipboard cap, per-role ralplan model routing, and leaked-reasoning-tag stripping on salvaged answers.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/README.zh.md

@@ -205,11 +205,11 @@ CI 通过 `.github/workflows/npm-publish.yml` 发布 — GitHub 发布 release
 ## 更新日志 (Changelog)
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.7.16]** (2026-06-25) — Bundled skills get a safe installer: `jeo skills sync` materializes the built-in workflow skills into `~/.jeo/skills`, preserving local edits by default, with a CI-friendly `--check` drift report and a `--force` overwrite.
 - **[0.7.15]** (2026-06-25) — Rollback release: the runtime source and test suite are restored to the **0.7.9** state. The intervening 0.7.10–0.7.14 runtime changes (quiet-mode/`-q`, `/resume` transcript rework, `mutatedFiles` plumbing, Ollama `num_ctx`, tmux REPL reaping, and related engine/provider/TUI edits) are reverted. Published as a new version because npm cannot re-issue an existing version number; `latest` now serves code identical to 0.7.9.
 - **[0.7.14]** (2026-06-25) — Quieter, cleaner output: a new `-q`/`--quiet` flag strips launch banners and courtesy logs, `/resume` no longer dumps walls of raw escaped JSON into scrollback, and a turn that edits files now reliably surfaces those paths to orchestrators.
 - **[0.7.9]** (2026-06-24) — Local model providers actually run, and OAuth/login hygiene: Ollama requests now carry an explicit `num_ctx` so jeo's large system prompt no longer overflows Ollama's small default window (the "ollama 모델이 안 돌아간다" blocker), LM Studio reasoning models that emit the whole reply on the reasoning channel are recovered instead of dying empty, GGUF chat templates that can't render the `tools` array are retried with native tools stripped, the OAuth success tab auto-closes with a clear hint, the effective-credential picker prefers a verified OAuth login over a stray API key, and abandoned idle `jeo` tmux REPL sessions are reaped on launch.
 - **[0.7.8]** (2026-06-24) — Long-session performance & resource hygiene: the detached memory-distill worker now always terminates itself (closing the `jeo memory-distill` orphan pileup that pinned each session's transcript in RSS), the live reasoning/output block re-wraps only its trailing window (O(tail) per frame instead of O(len²) over a long stream), the team loop stops reprinting the todo guide every iteration, and the readline caret stays aligned after an image attach.
-- **[0.7.7]** (2026-06-23) — Long-running-process hygiene & input robustness: backgrounded grandchildren (`next dev &`, daemons) are now reaped at the turn boundary via per-command process groups — closing the climbing `next-server` RSS leak — while the prompt gains recursive fuzzy `@path` search, ANSI-safe and idle-merged bracketed paste, a configurable OSC 52 clipboard cap, per-role ralplan model routing, and leaked-reasoning-tag stripping on salvaged answers.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jeo-code",
-  "version": "0.7.15",
+  "version": "0.7.16",
   "description": "Clean, highly optimized AI coding agent using spec-first loop",
   "type": "module",
   "main": "src/cli.ts",

package/src/cli/runner.ts

@@ -103,8 +103,8 @@ export const COMMANDS: readonly CommandSpec[] = [
   },
   {
     name: "skills",
-    summary: "List bundled workflow skills (jeo skills <name> for details).",
-    usage: "skills [name]",
+    summary: "List bundled workflow skills, or sync them into ~/.jeo/skills (jeo skills <name> for details).",
+    usage: "skills [list|read <name>|sync [--check|--force] [dir]|--write [dir]] [--json]",
     loader: async () => {
       const m = await import("../commands/skills");
       return args => m.runSkillsCommand(args);

package/src/commands/skills.ts

@@ -1,7 +1,109 @@
 import * as fs from "node:fs/promises";
 import * as path from "node:path";
-import { SKILLS, getSkillFrom, formatSkill, loadSkills, skillDirs } from "../skills/catalog";
+import * as os from "node:os";
+import { SKILLS, getSkillFrom, formatSkill, loadSkills, skillDirs, bundledSkillFileContent } from "../skills/catalog";
 import { getLocalJeoDir } from "../agent/state";
+import { jeoEnv } from "../util/env";
+
+/** User-level bundled-skills install dir (`~/.jeo/skills`, honoring JEO_CONFIG_DIR /
+ *  $HOME overrides) — the destination `jeo skills sync` reconciles against, matching
+ *  the highest-precedence flat dir in {@link skillDirs}. */
+export function userSkillsDir(): string {
+  const userHome = process.env.HOME || os.homedir();
+  const home = jeoEnv("CONFIG_DIR") || path.join(userHome, ".jeo");
+  return path.join(home, "skills");
+}
+
+export type SkillSyncStatus = "missing" | "up-to-date" | "differs";
+export type SkillSyncAction = "installed" | "overwritten" | "preserved" | "unchanged" | "none";
+
+export interface SkillSyncEntry {
+  name: string;
+  /** State of the on-disk file BEFORE any write this run. */
+  status: SkillSyncStatus;
+  /** What sync did (or, for --check, "none"). */
+  action: SkillSyncAction;
+  path: string;
+}
+
+export interface SkillSyncResult {
+  dir: string;
+  entries: SkillSyncEntry[];
+  /** true when any bundled skill is missing or differs from its on-disk copy. */
+  drift: boolean;
+  /** Number of files actually written this run. */
+  wrote: number;
+}
+
+/** Reconcile the bundled workflow skills against `dir` (default {@link userSkillsDir}).
+ *  Default install preserves existing local files (gjc `setup defaults` parity):
+ *  missing skills are installed, differing ones preserved unless `force`. `check`
+ *  is a pure drift report — it writes nothing. Returns a structured result so the
+ *  CLI wrapper owns all console output / exit codes. */
+export async function syncBundledSkills(
+  dir: string,
+  opts: { check?: boolean; force?: boolean } = {},
+): Promise<SkillSyncResult> {
+  const entries: SkillSyncEntry[] = [];
+  let wrote = 0;
+  if (!opts.check) await fs.mkdir(dir, { recursive: true });
+  for (const s of SKILLS) {
+    const file = path.join(dir, `${s.name}.md`);
+    const want = bundledSkillFileContent(s);
+    let have: string | null = null;
+    try { have = await fs.readFile(file, "utf-8"); } catch { /* missing */ }
+    const status: SkillSyncStatus = have === null ? "missing" : have === want ? "up-to-date" : "differs";
+    let action: SkillSyncAction = "none";
+    if (!opts.check) {
+      if (status === "missing") {
+        await fs.writeFile(file, want, "utf-8");
+        action = "installed";
+        wrote++;
+      } else if (status === "differs") {
+        if (opts.force) {
+          await fs.writeFile(file, want, "utf-8");
+          action = "overwritten";
+          wrote++;
+        } else {
+          action = "preserved";
+        }
+      } else {
+        action = "unchanged";
+      }
+    }
+    entries.push({ name: s.name, status, action, path: file });
+  }
+  return { dir, entries, drift: entries.some(e => e.status !== "up-to-date"), wrote };
+}
+
+async function runSkillsSync(cleanArgs: string[], isJson: boolean, cwd: string): Promise<void> {
+  const check = cleanArgs.includes("--check");
+  const force = cleanArgs.includes("--force");
+  const dirArg = cleanArgs.slice(1).find(a => !a.startsWith("--"));
+  const dir = dirArg ? path.resolve(cwd, dirArg) : userSkillsDir();
+  const result = await syncBundledSkills(dir, { check, force });
+
+  if (isJson) {
+    console.log(JSON.stringify({ ...result, mode: check ? "check" : force ? "force" : "install" }, null, 2));
+  } else if (check) {
+    console.log(`\n=== jeo skills sync --check ===\nTarget: ${dir}\n`);
+    for (const e of result.entries) console.log(`  ${e.status.padEnd(11)} ${e.name}`);
+    console.log(
+      result.drift
+        ? `\nDrift detected. Run 'jeo skills sync' to install missing skills, or '--force' to overwrite differing local copies.`
+        : `\nAll ${result.entries.length} bundled skills are in sync.`,
+    );
+  } else {
+    console.log(`\n=== jeo skills sync ===\nTarget: ${dir}\n`);
+    for (const e of result.entries) console.log(`  ${e.action.padEnd(11)} ${e.name}`);
+    const preserved = result.entries.filter(e => e.action === "preserved").length;
+    console.log(`\nWrote ${result.wrote} file(s)${preserved ? `, preserved ${preserved} local copy(ies) (use --force to overwrite)` : ""}.`);
+  }
+
+  // --check is CI-usable: non-zero exit signals drift (gjc `setup defaults --check` parity).
+  if (check && result.drift) process.exitCode = 1;
+}
+
 
 function editDistance(a: string, b: string): number {
   const m = a.length, n = b.length;
@@ -40,12 +142,20 @@ export async function runSkillsCommand(args: string[] = []): Promise<void> {
     await fs.mkdir(dir, { recursive: true });
     for (const s of SKILLS) {
       const file = path.join(dir, `${s.name}.md`);
-      await fs.writeFile(file, s.raw || `# ${s.name}\n\n${formatSkill(s)}\n`, "utf-8");
+      await fs.writeFile(file, bundledSkillFileContent(s), "utf-8");
     }
     console.log(`Wrote ${SKILLS.length} skill docs to ${dir}`);
     return;
   }
 
+  // `jeo skills sync [--check|--force] [dir]` — bundled skill inspection & safe install
+  // (gjc `setup defaults --check/--force` parity): install missing bundled skills into
+  // ~/.jeo/skills, preserve existing local copies by default, and report drift.
+  if (cleanArgs[0] === "sync") {
+    await runSkillsSync(cleanArgs, isJson, cwd);
+    return;
+  }
+
   // List/lookup over the MERGED set (bundled + ~/.jeo/skills + ~/.agents/skills + project dirs), matching the REPL /skill.
   const skills = await loadSkills(cwd);
   const command = cleanArgs[0];
@@ -60,6 +170,7 @@ export async function runSkillsCommand(args: string[] = []): Promise<void> {
         console.log(`  ${s.name.padEnd(16)} ${s.summary}`);
       }
       console.log("\nInvoke: /skill <name> [intent]  ·  $<name> [intent]  ·  skill-owned slash aliases (e.g. /speckit.plan)");
+      console.log("Install/inspect bundled defaults: 'jeo skills sync' (preserve local) · '--check' (drift report) · '--force' (overwrite local)");
       console.log("Discovery dirs (later wins on name clash; JEO_SKILLS_DIR adds more):");
       for (const d of skillDirs(cwd)) console.log(`  ${d}`);
       console.log("");

package/src/skills/catalog.ts

@@ -70,6 +70,13 @@ export function formatSkill(s: SkillDoc): string {
   ].filter(Boolean).join("\n");
 }
 
+/** Canonical on-disk content for a bundled skill — the exact bytes `jeo skills
+ *  --write` / `jeo skills sync` materialize into a `<name>.md` file. Prefers the
+ *  original SKILL.md text; falls back to a decorated render for skills without raw. */
+export function bundledSkillFileContent(s: SkillDoc): string {
+  return s.raw || `# ${s.name}\n\n${formatSkill(s)}\n`;
+}
+
 function compactSkillExecutionBrief(skill: SkillDoc): string {
   const aliases = skillSlashAliases(skill);
   let details = skill.details;