kodelyth-ecc 1.2.1 → 1.3.0

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/KODELYTH.md CHANGED
@@ -2,68 +2,103 @@
2
2
 
3
3
  Kodelyth ECC is built and maintained by **Kodelyth** — a brand focused on making AI-powered developer tooling smarter, faster, and production-grade for real-world use at scale.
4
4
 
5
- ## What Kodelyth Added v1.2.0
5
+ ## What's New in v1.3.0
6
6
 
7
- ### Agents (6 exclusive)
7
+ ### Removed: Lens Dashboard
8
8
 
9
- | Agent | Added | Description |
10
- |---|---|---|
11
- | `kodelyth-advisor` | v1.0.0 | Master guide — reads your intent and points you to the right tool instantly |
12
- | `debug-detective` | v1.0.0 | Root-cause hunter — traces every bug through evidence, never guesses |
13
- | `ux-reviewer` | v1.0.0 | UX and WCAG 2.1 AA reviewer — catches what users feel before they feel it |
14
- | `api-guardian` | v1.1.0 | API contract protector — classifies every change as BREAKING/SAFE/WARN |
15
- | `pair-programmer` | v1.1.0 | Pre-code thinking partner — catches wrong approaches before a line is written |
16
- | `migration-guide` | v1.1.0 | Framework migration specialist — full playbooks for React, Next.js, Python, Java, Go |
9
+ The bundled Kodelyth Lens dashboard has been removed. It tried to track usage across Claude Code, Windsurf, Cursor, Codex, OpenCode, and Antigravity — but cloud-AI platforms (Windsurf, Antigravity, parts of Cursor) don't expose token/cost data locally, which led to inaccurate or hardcoded zeros in the dashboard.
17
10
 
18
- ### Skills (4 exclusive)
11
+ The toolkit is now leaner and **does only what it does honestly**: ship rules, agents, and skills your AI loads on every session.
19
12
 
20
- | Skill | Added | Description |
21
- |---|---|---|
22
- | `kodelyth-quickstart` | v1.0.0 | Plain-language onboarding guide — up and running in 10 minutes |
23
- | `smart-debug` | v1.0.0 | Structured 5-step debugging framework — ends guess-and-check forever |
24
- | `git-mastery` | v1.1.0 | Trunk-based dev, gitflow, Ship/Show/Ask, rebase, bisect, monorepos |
25
- | `observability` | v1.1.0 | Structured logging, Four Golden Signals, OpenTelemetry, SLOs |
13
+ ### Added: God-Tier Intent Routing
26
14
 
27
- ### Hooks (3 new in v1.1.0)
15
+ A new rule (`rules/common/agent-intent-routing.md`) auto-loads with every session and maps user intent → specialist agent across 10 priority tiers, 50+ pattern groups.
28
16
 
29
- | Hook | Trigger | What it does |
30
- |---|---|---|
31
- | `test-reminder` | After code file edit | Reminds you to write tests when no test file was touched |
32
- | `smart-suggest` | After each response | Reads session signals and suggests the next logical agent |
33
- | `branch-name-check` | Before `git checkout -b` | Blocks non-standard branch names |
17
+ Now the user just describes their problem in plain words and the AI:
18
+
19
+ 1. Announces which specialist is taking over (`→ Routing to <agent>`)
20
+ 2. Behaves as that agent
21
+ 3. Teaches the user the explicit form (`Tip: type "use <agent>"`)
22
+
23
+ No more memorizing 58 agent names.
34
24
 
35
- ### Kodelyth Lens AI Usage Dashboard (v1.2.0)
25
+ ### Added: 5 New Specialist Agents
36
26
 
37
- Built-in zero-dependency dashboard at `~/.kodelyth/lens/`. Reads from 6 platforms:
27
+ | Agent | What it does |
28
+ |---|---|
29
+ | `dependency-doctor` | Diagnoses npm/pip/cargo/maven dep hell, CVEs, lockfile drift |
30
+ | `git-rescue` | Recovers broken git states without destroying history |
31
+ | `release-captain` | Owns the release ritual — semver, tagging, publishing, rollback |
32
+ | `env-debugger` | "Works on my machine" hunter — env, config, secrets, layers |
33
+ | `flake-hunter` | Stabilizes flaky tests — never adds blind retries |
38
34
 
39
- - **Claude Code** full local JSONL session logs
40
- - **Codex CLI** — SQLite threads database
41
- - **Windsurf** — workspaceStorage and Cascade files
42
- - **Cursor** — workspaceStorage and chat files
43
- - **OpenCode** — local JSONL/JSON session files
44
- - **Antigravity** — ECC project detection (AI sessions server-side)
35
+ ### Added: 2 New Meta Skills
45
36
 
46
- Features:
47
- - Agent leaderboard with call counts and platform breakdown
48
- - Intent detection 12 categories, 50+ patterns infers agent usage from natural language
49
- - ECC savings banner shows actual cost vs. estimated without-ECC
50
- - 30-day cost chart, cache hit rate, today vs. all-time stats
51
- - Only shows installed platforms — nothing shown for platforms you don't have
37
+ | Skill | What it does |
38
+ |---|---|
39
+ | `intent-routing` | Documents the routing system, priority tiers, design rules for new agents |
40
+ | `agent-handoff` | Standard multi-agent chain protocol with 10 documented chains |
41
+
42
+ ---
43
+
44
+ ## Kodelyth Exclusives — The 11 Agents That Define ECC
45
+
46
+ | Agent | Job |
47
+ |---|---|
48
+ | `kodelyth-advisor` | Master guide — picks the right tool when you don't know where to start |
49
+ | `pair-programmer` | The engineer who sits next to you **before** you write the code |
50
+ | `debug-detective` | Never guesses — traces every bug to root cause through evidence |
51
+ | `silent-failure-hunter` | Finds bugs that don't throw errors |
52
+ | `ux-reviewer` | Reviews UX behavior + WCAG 2.1 AA accessibility |
53
+ | `api-guardian` | Detects breaking API changes before they ship |
54
+ | `migration-guide` | Framework / language version upgrades, phase by phase |
55
+ | `dependency-doctor` | Dep hell — CVE triage, lockfile diagnosis, safe upgrade plans |
56
+ | `git-rescue` | Broken git states, lost commits, bad rebases — without destroying history |
57
+ | `release-captain` | Owns the release ritual, semver, rollback rehearsal |
58
+ | `env-debugger` | "Works on my machine" — environment, config, secrets layers |
59
+ | `flake-hunter` | Flaky test stabilization — root cause, not retries |
60
+
61
+ ---
52
62
 
53
- ### Install Improvements (v1.2.0)
63
+ ## Kodelyth-Authored Skills
54
64
 
55
- - `npx github:sifxprime/kodelyth-ecc` works from any machine with Node.js 18+
56
- - `--target windsurf-project` and `--target windsurf-home` — full Windsurf support
57
- - Auto-generates `.windsurfrules` from all common rules (Cascade reads it automatically)
58
- - Non-interactive install works (CI, Docker, headless environments)
59
- - `go` normalized to `golang` in language argument parser
65
+ | Skill | Description |
66
+ |---|---|
67
+ | `kodelyth-quickstart` | Plain-language onboarding guide |
68
+ | `smart-debug` | Structured 5-step debugging framework |
69
+ | `git-mastery` | Trunk-based dev, rebase, bisect, monorepos |
70
+ | `observability` | Structured logging, Four Golden Signals, OpenTelemetry, SLOs |
71
+ | `intent-routing` | How auto-routing works, priority tiers, design rules |
72
+ | `agent-handoff` | Standard multi-agent chain protocol |
73
+
74
+ ---
75
+
76
+ ## Hooks (15+)
77
+
78
+ | Hook | Trigger | What it does |
79
+ |---|---|---|
80
+ | `session-start` | Session boot | Loads context from previous session |
81
+ | `pre-commit` | Before `git commit` | Catches `console.log`, secrets, bad commit messages |
82
+ | `quality-gate` | After file edit | Runs type checks and formatting |
83
+ | `git-push-reminder` | Before `git push` | Prompts review |
84
+ | `config-protection` | Before linter config edit | Blocks weakening of standards |
85
+ | `desktop-notify` | Long task complete | macOS notification |
86
+ | `mcp-health-check` | Before MCP call | Validates server health |
87
+ | `test-reminder` | Code edit without tests | Prompts to write tests |
88
+ | `smart-suggest` | After response | Suggests next logical agent |
89
+ | `branch-name-check` | Before branch create | Blocks non-standard names |
90
+
91
+ ---
60
92
 
61
93
  ## Kodelyth Philosophy
62
94
 
63
95
  - **Smart defaults** — the system guides you to the right tool without you needing to know everything upfront
64
96
  - **Production quality** — every enhancement is battle-tested, not theoretical
65
97
  - **Zero friction** — good tools get out of the way and let you build
66
- - **Real data** — Lens shows your actual usage, no hardcoding, no placeholders
98
+ - **No magic, no telemetry** — everything is markdown your AI reads on every session; nothing phones home
99
+ - **Honest about limits** — if we can't measure something accurately, we don't pretend (this is why Lens was removed)
100
+
101
+ ---
67
102
 
68
103
  ## Credits
69
104