@dzhechkov/harness-cli 0.3.80 → 0.3.81
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +26 -4
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -147,7 +147,7 @@ dz publish # publish all changed packages
|
|
|
147
147
|
dz init --target claude-code --select design-thinking
|
|
148
148
|
|
|
149
149
|
# Curated set by topic (recommended):
|
|
150
|
-
dz setup --target claude-code --preset meta #
|
|
150
|
+
dz setup --target claude-code --preset meta # 12 development skills + self-learning
|
|
151
151
|
|
|
152
152
|
# Full toolkit with orchestrated pipeline:
|
|
153
153
|
npx @dzhechkov/keysarium init # 7-phase research + commands + memory
|
|
@@ -513,7 +513,7 @@ Phase 6: VALIDATE → Pilot with variance analysis: projected vs actual → Sc
|
|
|
513
513
|
|
|
514
514
|
### What's included vs what's optional
|
|
515
515
|
|
|
516
|
-
**Core DT** — the `meta` preset includes all required dependencies (
|
|
516
|
+
**Core DT** — the `meta` preset includes all required dependencies (12 skills):
|
|
517
517
|
|
|
518
518
|
```bash
|
|
519
519
|
dz setup --target claude-code --preset meta
|
|
@@ -580,7 +580,7 @@ dz pretrain # detects stack, recommends pres
|
|
|
580
580
|
dz recommend "work on this Node.js API" # suggests skills + toolkits
|
|
581
581
|
|
|
582
582
|
# 2. Install skills (choose your level)
|
|
583
|
-
dz setup --target claude-code --preset meta --memory agentdb #
|
|
583
|
+
dz setup --target claude-code --preset meta --memory agentdb # 12 skills (includes feature-adr)
|
|
584
584
|
dz setup --target claude-code --preset qe-engineer # + 20 QE skills
|
|
585
585
|
|
|
586
586
|
# Want the full feature-adr toolkit with /feature-adr command + governance?
|
|
@@ -614,6 +614,28 @@ Architecture-aware development: every skill knows the codebase structure.
|
|
|
614
614
|
|
|
615
615
|
---
|
|
616
616
|
|
|
617
|
+
## Example: AI-Assisted Reasoning & Self-Improvement
|
|
618
|
+
|
|
619
|
+
```
|
|
620
|
+
# Auto-select reasoning strategy:
|
|
621
|
+
"Compare 3 architectures" → structured-reasoning: Tree-of-Thought (branches + scoring)
|
|
622
|
+
"Debug this test" → structured-reasoning: Chain-of-Thought (linear trace)
|
|
623
|
+
"We've been looping" → structured-reasoning: Reflection-Suppression (break loop)
|
|
624
|
+
|
|
625
|
+
# Self-review before delivering:
|
|
626
|
+
"Write a migration and verify" → reflection-loop: draft → critique → revise (max 3 rounds)
|
|
627
|
+
|
|
628
|
+
# Manage long sessions:
|
|
629
|
+
"Context is getting long" → context-window-management: checkpoint + prune + continue
|
|
630
|
+
|
|
631
|
+
# Learn from success:
|
|
632
|
+
"Extract this as a skill" → skill-crystallizer: trace → reusable SKILL.md
|
|
633
|
+
```
|
|
634
|
+
|
|
635
|
+
All included in `meta` preset.
|
|
636
|
+
|
|
637
|
+
---
|
|
638
|
+
|
|
617
639
|
## Self-Learning: JSONL vs AgentDB
|
|
618
640
|
|
|
619
641
|
DZ Harness supports two memory backends for self-learning:
|
|
@@ -879,7 +901,7 @@ npx @dzhechkov/p-replicator init
|
|
|
879
901
|
|
|
880
902
|
## Status
|
|
881
903
|
|
|
882
|
-
`v0.3.
|
|
904
|
+
`v0.3.80` — published on npm. Also available as [Claude Plugin](#claude-plugin). Part of [DZ Harness Hub](https://github.com/djd1m/dz-harness-hub).
|
|
883
905
|
|
|
884
906
|
## Claude Plugin
|
|
885
907
|
|
package/package.json
CHANGED