kairos-chain 3.29.5 → 3.29.6
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.
- checksums.yaml +4 -4
- data/CHANGELOG.md +15 -0
- data/lib/kairos_mcp/version.rb +1 -1
- data/templates/knowledge/skill_authoring_patterns/skill_authoring_patterns.md +32 -12
- metadata +1 -1
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 2fde5a098059d86d9dc9f10cbf27911ff0f6e3defede0c118ac36825964e4d55
|
|
4
|
+
data.tar.gz: a5b4283909071dba96fa4919a59ef29881f084f76a8fd8cc3335fcbb9efdb8d9
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: a863d1090ee8f1bffaf6a47848d01f113983d2b2efefd7d7e1e6ecec89105fbdde85851681f163be8d93d79734d6bfaf23dfa0080436ec9a15e0e2e258db2d52
|
|
7
|
+
data.tar.gz: 93690b91d30df69e0dc634b135dc3492a436089e6af02906ed4cfaf22e8fb42f5e174fa500dfb1fc906096b3b6fbb275cf043399042a642f9847a28a471147b8
|
data/CHANGELOG.md
CHANGED
|
@@ -4,6 +4,21 @@ All notable changes to the `kairos-chain` gem will be documented in this file.
|
|
|
4
4
|
|
|
5
5
|
This project follows [Semantic Versioning](https://semver.org/).
|
|
6
6
|
|
|
7
|
+
## [3.29.6] - 2026-06-05
|
|
8
|
+
|
|
9
|
+
### Changed — `skill_authoring_patterns` readability + structure (v1.4)
|
|
10
|
+
|
|
11
|
+
- Source citation is now a direct URL link to Anthropic's post (was a bare
|
|
12
|
+
local reference path); the local reference remains as a provenance / facts
|
|
13
|
+
index pointer.
|
|
14
|
+
- §A nine categories are itemized as a numbered list with one-line glosses
|
|
15
|
+
(was a flat inline string).
|
|
16
|
+
- §C ("KairosChain extension") translated from Japanese to English for
|
|
17
|
+
consistency with §A/§B (entry is LLM-primary).
|
|
18
|
+
- Added a "When to reference this entry" section: when to open it (authoring,
|
|
19
|
+
classifying, layer placement, mis-trigger, positioning) and which sibling
|
|
20
|
+
skill to use instead (lifecycle, quality gates, scaffolding mechanics).
|
|
21
|
+
|
|
7
22
|
## [3.29.5] - 2026-06-05
|
|
8
23
|
|
|
9
24
|
### Changed — `skill_authoring_patterns` ships a source pointer, not the article full text
|
data/lib/kairos_mcp/version.rb
CHANGED
|
@@ -1,27 +1,47 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: skill_authoring_patterns
|
|
3
|
-
description: Use
|
|
4
|
-
version: "1.
|
|
3
|
+
description: Use before authoring or classifying a KairosChain SkillSet — to pick a category, apply authoring craft, and place norm/meta skills. Anthropic's nine categories + craft, extended with the categories Anthropic omits.
|
|
4
|
+
version: "1.4"
|
|
5
5
|
tags: [skills, authoring, taxonomy, meta, provenance]
|
|
6
6
|
---
|
|
7
7
|
|
|
8
8
|
# Skill Authoring Patterns
|
|
9
9
|
|
|
10
|
-
Distilled from Anthropic's
|
|
10
|
+
Distilled from Anthropic's post [Lessons from Building Claude Code: How We Use Skills](https://claude.com/blog/lessons-from-building-claude-code-how-we-use-skills) (Thariq Shihipar, 2026-06-03). Local provenance / facts index: `references/anthropic_skills_lessons_2026-06-03.md` (pointer; full archive in `log/`, not shipped). SkillSet→category map: `references/kairoschain_skillset_category_map.md`.
|
|
11
11
|
|
|
12
|
-
##
|
|
13
|
-
|
|
12
|
+
## When to reference this entry
|
|
13
|
+
A reference to open *before* creating or classifying a skill — not a runtime tool:
|
|
14
|
+
- New SkillSet / L1 knowledge → pick a §A category, then apply §B craft.
|
|
15
|
+
- Auditing or classifying existing skills → §A + the category-map reference.
|
|
16
|
+
- "SkillSet vs core vs instruction mode?" → §C (tool / norm / meta).
|
|
17
|
+
- A skill mis-triggers or goes unused → §B (description-as-trigger, don't-railroad, usage logging).
|
|
18
|
+
- Explaining KairosChain's positioning (grants, design rationale) → §C "why this bet".
|
|
19
|
+
- Not for: lifecycle/maturation → [[agent_skill_evolution_guide]]; wiring/quality gates → skillset_implementation_quality_guide; scaffolding mechanics → skillset_creator.
|
|
20
|
+
|
|
21
|
+
## A. Nine categories (base-level / operational skills)
|
|
22
|
+
1. Library/API reference — use a library, CLI, or SDK correctly.
|
|
23
|
+
2. Product verification — test/verify code works (highest measured impact, per Anthropic).
|
|
24
|
+
3. Data fetching & analysis — query data and monitoring stacks.
|
|
25
|
+
4. Business-process & team automation — collapse repetitive workflows into one command.
|
|
26
|
+
5. Code scaffolding & templates — generate framework boilerplate.
|
|
27
|
+
6. Code quality & review — enforce standards, assist review.
|
|
28
|
+
7. CI/CD & deployment — fetch, build, push, deploy.
|
|
29
|
+
8. Runbooks — symptom → multi-tool investigation → structured report.
|
|
30
|
+
9. Infrastructure operations — routine maintenance / ops procedures.
|
|
31
|
+
|
|
32
|
+
These map ~1:1 onto KairosChain base-level SkillSets (see category-map reference).
|
|
14
33
|
|
|
15
34
|
## B. Authoring craft (universal principles — adopt as-is)
|
|
16
|
-
Don't state the obvious; the highest-signal content is the **Gotchas** section. **Progressive disclosure**: thin entry, load detail on demand; the folder *is* context engineering. **Don't railroad**: give What/Why, keep How flexible; narrow with JSON/YAML only when output drifts. `description` = the model's trigger condition, not a
|
|
35
|
+
Don't state the obvious; the highest-signal content is the **Gotchas** section. **Progressive disclosure**: thin entry, load detail on demand; the folder *is* context engineering. **Don't railroad**: give What/Why, keep How flexible; narrow with JSON/YAML only when output drifts. `description` = the model's trigger condition, not a summary. Validation first; ship scripts so the model composes. Hooks as guardrails; usage-logging for health checks. Distribution: repo for small scope, marketplace for scale. (`${CLAUDE_PLUGIN_DATA}`, PreToolUse, introspection_check are Claude/KairosChain mechanisms; the principles are universal, the mechanisms are not.)
|
|
17
36
|
|
|
18
37
|
## C. KairosChain extension (the categories Anthropic omits)
|
|
19
|
-
|
|
20
|
-
-
|
|
21
|
-
-
|
|
22
|
-
|
|
38
|
+
Anthropic's nine are all base-level *tool* skills — there is no norm or meta category. Not an oversight but a design difference: Anthropic keeps philosophy/norms in the core (weights, Constitution, system prompt), so it has no reason to make philosophy a kind of skill. KairosChain's structural self-referentiality (Prop 1) lets norms and meta-rules be expressed like ordinary skills, so it adds two:
|
|
39
|
+
- **Norm skill / instance constitution** — how an instance acts; philosophy lives here. e.g. masa mode.
|
|
40
|
+
- **Meta skill** — the evolution rules for skills. e.g. skillset_creator / knowledge_creator (SkillSet-type), [[agent_skill_evolution_guide]] (L1), the L0 tool skills_evolve (different layer); the L2→L1→L0 promotion path is a process, not an artifact.
|
|
41
|
+
|
|
42
|
+
The real difference is *where philosophy lives*: closed in the core (Anthropic) vs open in an external instruction mode (KairosChain). Anthropic having no philosophy category is consistent with — not proof of — masa mode's Scaffolding Stance.
|
|
23
43
|
|
|
24
|
-
**Why this bet (operational vs decorative).** Harness-hosted norms are readable
|
|
44
|
+
**Why this bet (operational vs decorative).** Harness-hosted norms are readable/recorded/revisable where weight-baked norms are opaque (Prop 10, provisional in masa.md) — arguably the right home even post-fine-tuning. Hiring a model ≠ cultivating one; the harness is the weight-less user's only cultivation surface. Norms must stay operational (hooks, introspection_check), not decorative (philosophy theater). Full dialogue: `references/why_harness_norms_bet.md`.
|
|
25
45
|
|
|
26
46
|
## Related
|
|
27
|
-
[[agent_skill_evolution_guide]]
|
|
47
|
+
[[agent_skill_evolution_guide]] · skillset_implementation_quality_guide · kairoschain_meta_philosophy (Prop 1; Prop 10 provisional in masa.md).
|