* feat: add Loops component type with referenced components
Introduce "loops" as a new first-class component type — autonomous
agentic workflows (goal + interval + stop condition) inspired by Loop
Engineering and the Forward Future Loop Library. A loop references other
components (agents, skills, hooks, commands) and the dashboard renders
those references as clickable cards on the loop detail page.
- Add 10 starter loops under cli-tool/components/loops/
(engineering, evaluation, operations)
- Catalog generator: scan loops + write per-component content files
- Dashboard: new type in icons/types/data, /loops page, and a
"Referenced Components" section on the loop detail page
- CLI: --loop flag installs the loop to .claude/loops/ and
auto-installs its referenced components
- Docs: CLAUDE.md and cli-tool rule updated
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01T9ynMiCHt8JwdMd8KjCkES
* feat: add 8 more loops (goal/loop/schedule patterns)
Expand the loops catalog to 18, drawing ideas from community "Loop
Engineering" patterns. Adds loops across the three trigger verbs
(/goal, /loop, /schedule) and emphasizes the verifier-in-the-loop and
budget / anti-spin guardrails:
- engineering/five-minute-maintainer-loop
- engineering/build-test-fix-loop
- engineering/anti-spin-build-loop
- engineering/completion-contract-loop (goal)
- engineering/goal-refiner-loop (goal)
- engineering/adversarial-review-loop (cross-model)
- operations/overnight-pr-routine-loop (schedule)
- evaluation/human-approval-loop (human-in-the-loop)
All referenced components verified to exist; catalog regenerated.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01T9ynMiCHt8JwdMd8KjCkES
---------
Co-authored-by: Claude <noreply@anthropic.com>
Adds a browser game (Dino Runner, Snake, Flappy Bird) that launches
automatically when Claude enters Plan Mode. Users can play while
Claude thinks, with a live activity panel showing tool calls in real time.
- EnterPlanMode / /plan → starts Node.js server on port 3456, opens browser
- Read/Bash/Edit/etc → streams tool activity to the activity panel
- Write to ~/.claude/plans/ → also triggers the game (for /plan command)
- AskUserQuestion → pauses the game until user responds
- ExitPlanMode → shows "Claude Termino!" overlay with confetti
Features: 3 games with high score persistence, signed score sharing URLs,
pause overlay when Claude needs input, activity feed with tool icons.
Requires Node.js and curl. macOS only (uses 'open' command).
Install globally in ~/.claude/settings.json to work across all projects.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds the memex MCP server under mcps/integration. memex watches git
repos and builds a temporal knowledge graph of modules, symbols,
decisions, and open problems (Graphiti + Neo4j), serving it to AI
coding agents over MCP. 12 read/write tools.
npx -y stifler-memex-mcp serve --repo .
Adds a new skill in creative-design/luma-imagegen that generates images
via the Luma AI Dream Machine API (Photon models). Includes interactive
prompt collection, API key detection from env/.env, async polling, and
image download. Uses only Python stdlib — no extra dependencies.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Remove all hardcoded absolute paths (/Users/cam/VAULT01/...) and replace with relative paths rooted at ./
- Replace VAULT01-specific references with generic vault terminology
- Add model: sonnet field to frontmatter
- Add Grep and Edit to tools list; remove redundant LS
- Expand description with 3 concrete <example>/<commentary> blocks
- Add Native Fallback Workflow using Glob+Grep (no external script dependency)
- Make Script-Assisted Workflow optional with clear instructions to run from vault root
- Add MOC Hierarchy section explaining LYT three-tier system (Home/Topic/Sub-MOC)
- Add optional Dataview block to MOC template with note to remove if plugin absent
- Add vault-structure validation guidance before writing MOCs
Automated review cycle | Co-Authored-By: Claude Code <noreply@anthropic.com>
Major upstream rewrite with eval-driven iterative workflow:
- New agents/ dir (analyzer, comparator, grader)
- New eval-viewer/ for benchmark review UI
- New scripts for eval runs, benchmarking, and description optimization
- Updated SKILL.md with new process-oriented approach
- Preserved local-only files (output-patterns.md, workflows.md, init_skill.py)
- Add tools and model frontmatter fields (WebSearch, WebFetch; sonnet)
- Rewrite description with three concrete examples using example/commentary tags
- Add imperative "When Invoked" seven-step workflow checklist
- Add iterative retrieval loop section with explicit stopping conditions
- Add source credibility evaluation framework as a scored table
- Expand contradiction handling into an explicit protocol with example format
Automated review cycle | Co-Authored-By: Claude Code <noreply@anthropic.com>
- Fix llms.txt skeleton to use official Markdown spec (H1, blockquote, H2 sections, link entries)
- Fix page entries to use standard `- [Title](url): description` format instead of invented key:value
- Add framework detection step (Astro, Nuxt, Next.js, Svelte, Hugo) to determine output path
- Add `model: haiku` to frontmatter for cost-efficient deterministic scanning
- Add `maxTurns: 20` to prevent runaway iteration on large sites
- Remove unconditional `git push` — commit only by default, user pushes manually
- Expand "AEO" to "AI Engine Optimization (AEO)" on first use in description and closing line
Automated review cycle | Co-Authored-By: Claude Code <noreply@anthropic.com>
CONTEXT_LIMIT was hardcoded at 200_000 tokens. Users on Claude Opus 4.7
with the 1M context window (model: "opus[1m]") saw the dashboard report
80–90% usage when actual usage was 16–18%, producing constant false
alarms and making the meter unusable on long sessions.
Replace the constant with _detect_context_limit(cwd) which resolves the
limit in this order:
1. CONTEXT_TIMELINE_LIMIT env var (explicit override).
2. "[1m]" / "[1000k]" / "[200k]" suffix in the model field of
<cwd>/.claude/settings.json or ~/.claude/settings.json.
3. 200_000 fallback (current behaviour).
The limit is resolved once per _Builder instance, not per update tick,
so there is no per-event filesystem cost.
No new dependencies (still stdlib only). Default behaviour is unchanged
for users without [1m] in settings.
Verified live on a real Opus 1M session of ~178K tokens: meter went
from 89% (red, false alarm) to 17.9% (cyan, correct).
The `_encode_cwd` function only replaced `/` with `-`, which fails on
Windows where paths use `\` and include drive letters like `D:`.
Use regex to replace any non-alphanumeric character with `-`.
This handles all edge cases across platforms in a single expression.
Before: D:\Claude\mcx -> lookup failed
After: D:\Claude\mcx -> D--Claude-mcx (matches Claude Code encoding)
- Add correct YAML frontmatter with name, description, tools, and model fields
- Target WCAG 2.2 explicitly with all 9 new criteria listed in reference table
- Mandate hybrid testing methodology (automated + manual) with 40% coverage note
- Restrict tools to Read, Grep, Glob, Bash — no Write/Edit (auditor pattern)
- Set model to sonnet for deep ARIA and compliance reasoning
- Add two example blocks with user/assistant/commentary tags
- Add legal compliance mapping to ADA, Section 508, and EAA (June 2025)
- Add structured finding format with WCAG criterion, severity, remediation, verification
- Add summary scorecard template with severity counts and WCAG 2.2 criterion status
- Add ARIA patterns reference (dialog, combobox, tabs, navigation, live regions)
- Add screen reader test matrix (VoiceOver, NVDA, TalkBack, JAWS)
Automated review cycle | Co-Authored-By: Claude Code <noreply@anthropic.com>
Persistent local HTTP daemon (port 7878, 1h watchdog) that reads Claude Code
JSONL transcripts and renders a git-graph timeline in the browser — one column
per agent (main + subagents), tool nodes color-coded by type, context-window
sidebar with cache/input/output breakdown. Pure stdlib Python, zero pip deps.
Daniel Avila