{
  "markdown": "# Skillforge\n\n<img src=\"./assets/skillforge-banner.png\" alt=\"Skillforge — skill packs for Claude Code\" width=\"100%\" />\n\nSkill packs for Claude Code and other LLM agent systems, published by [Entelligentsia](https://github.com/Entelligentsia) ([entelligentsia.in](https://entelligentsia.in)).\n\n## Available Packages\n\n| Package | Type | Description |\n|---------|------|-------------|\n| [forge](https://github.com/Entelligentsia/forge) | Meta-generator | Self-enhancing AI software development lifecycle — scans your codebase, generates project-specific workflows, personas, templates, and tools |\n| [security-watchdog](./security-watchdog/) | Security plugin | Auto-scans newly installed/updated Claude Code plugins for prompt injection, malicious hook scripts, and data exfiltration |\n| [freshdesk-api](./freshdesk-api/) | API integration skill | Freshdesk helpdesk API — tickets, contacts, companies, knowledge base (Solutions), webhooks (1 skill) |\n| [doc-review](./doc-review/) | Workflow plugin | In-browser document annotation overlay with Claude-driven apply — annotate HTML/MD/TXT in browser, click Apply, edits land in source (1 skill, 4 commands) |\n| [design-patterns](./design-patterns/) | Reference skills | Canonical software design patterns — all 23 GoF + enterprise/DDD patterns (10 skills) |\n| [llm-patterns](./llm-patterns/) | Reference skills | LLM integration patterns — RAG, tool use, agents, guardrails, tool synthesis (9 skills) |\n| [harness-engineering](./harness-engineering/) | Reference skills | Agent harness components — memory/compaction, caching, sandboxing, permissions, telemetry, hooks, routing (13 skills) |\n| [meta-webxr-skills](./meta-webxr-skills/) | Reference skills | Meta Quest PWA XR engineering (8 skills) |\n| [threejs-skills](./threejs-skills/) | Reference skills | Three.js 3D development (10 skills) |\n| [lean-ctx-eval](./lean-ctx-eval/) | Eval skill | Measure whether the lean-ctx MCP saves more context tokens than it costs, from your own Claude Code session transcripts (1 skill) |\n\n## Installation (Claude Code)\n\n```\n/plugin marketplace add Entelligentsia/skillforge\n```\n\nThen install whichever packs you need:\n\n```\n/plugin install security-watchdog@skillforge\n/plugin install freshdesk-api@skillforge\n/plugin install doc-review@skillforge\n/plugin install design-patterns@skillforge\n/plugin install llm-patterns@skillforge\n/plugin install harness-engineering@skillforge\n/plugin install threejs-skills@skillforge\n/plugin install meta-webxr-skills@skillforge\n/reload-plugins\n```\n\n### Forge\n\nForge has its own repository. See [Entelligentsia/forge](https://github.com/Entelligentsia/forge) for installation instructions.\n\n## Forge\n\nForge is different from the reference skill packages. Instead of loading knowledge into context, it **generates** a complete project-specific engineering practice: agent personas, workflows, templates, review checklists, and tools — all tailored to your stack.\n\n```bash\n/forge init          # Bootstrap SDLC into your project\n/sprint-plan         # Start your first sprint (generated command)\n/engineer ACME-S01-T01   # Plan a task (generated command)\n```\n\nSee [Entelligentsia/forge](https://github.com/Entelligentsia/forge) for the full vision and design.\n\n## Skills Index\n\n### security-watchdog\n\n| Skill / Command | Purpose |\n|-----------------|---------|\n| `/security-watchdog:scan-plugin <plugin-id>` | Scan any installed plugin for prompt injection, malicious hooks, and data exfiltration |\n| `plugin-security` | Threat model and heuristics reference — attack taxonomy, severity guide, detection patterns |\n\nRuns automatically via `SessionStart` hook: detects newly installed or updated plugins and prompts Claude to scan before your first request.\n\n### design-patterns\n\n| Skill | Patterns Covered |\n|-------|-----------------|\n| `pattern-selection` | **Entry point** — decision tree mapping pain to pattern |\n| `creational` | Singleton, Builder, Factory Method, Abstract Factory, Prototype |\n| `structural` | Adapter, Facade, Decorator, Proxy, Composite, Flyweight, Bridge |\n| `behavioural` | Chain of Responsibility, Command, Strategy, State, Observer, Memento, Mediator, Visitor, Iterator, Template Method |\n| `domain-modeling` | Entity, Value Object, Aggregate, Aggregate Root |\n| `data-access` | Repository, Unit of Work, Data Mapper, Active Record |\n| `service-layer` | Application Service, Domain Service, Service Layer |\n| `domain-events` | Domain Events, Transactional Outbox, eventual consistency |\n| `cqrs` | Commands, Queries, Read Models, Projections |\n| `anti-corruption` | Anti-Corruption Layer, Gateway, Strangler Fig |\n\n### llm-patterns\n\n| Skill | Pain It Removes |\n|-------|----------------|\n| `pattern-selection` | **Entry point** — decision tree for LLM integration patterns |\n| `structured-generation` | Output breaks parsers, violates schemas, varies in shape |\n| `rag` | LLM hallucinates, lacks domain knowledge, gives stale answers |\n| `tool-use` | LLM needs live data, calculations, or side effects |\n| `agent-loop` | Task requires autonomous multi-step reasoning |\n| `guardrails` | Output contains harmful content, PII, or policy violations |\n| `prompt-engineering` | Prompts are ad-hoc, untested, unversioned |\n| `graceful-degradation` | Model is down, slow, or over budget |\n| `evaluation-harness` | No way to measure quality or detect regressions |\n| `tool-synthesis` | LLM called repeatedly for tasks codifiable as deterministic tools |\n\n### harness-engineering\n\n| Skill | Concern |\n|-------|---------|\n| `memory-compaction` | Hot/warm/cold tiering; lossy compression that preserves load-bearing facts |\n| `session-persistence` | Append-only event log, idempotent replay, fork/branch semantics |\n| `prompt-caching` | Cache breakpoint placement, TTL, silent-regression traps |\n| `subagent-orchestration` | Fork vs fresh, briefing rules, isolation, result merge |\n| `tool-sandboxing` | Filesystem jail, egress allowlist, scoped credentials, safety vs security |\n| `permission-gates` | Risk classification, scope-bound consent, irreversibility detection |\n| `streaming-io` | Partial JSON parsing, mid-stream tool dispatch, cancellation |\n| `concurrency-control` | Conflict graphs, intent-order feed, partial-failure semantics |\n| `telemetry-tracing` | Span trees, cost attribution, replay from trace |\n| `hook-system` | Event taxonomy, blocking vs observation, failure isolation |\n| `workspace-state` | Read-before-edit invariant, external-mutation detection |\n| `model-routing` | Capability/cost routing, distinct from graceful degradation |\n| `rate-limiting` | Token buckets per dimension, fair allocation, 429 with Retry-After |\n\n### meta-webxr-skills\n\n| Skill | Trigger |\n|-------|---------|\n| `webxr-session` | WebXR session lifecycle, requestSession, feature flags |\n| `webxr-rendering` | XR render loop, reference spaces, frame timing |\n| `webxr-input` | Controller input, hand tracking, hit testing |\n| `webxr-passthrough` | AR/MR passthrough, plane/mesh detection |\n| `webxr-anchors` | Persistent spatial anchors |\n| `webxr-layers` | WebXR Layers API, compositing |\n| `webxr-ratk` | Reality Accelerator Toolkit (Three.js wrapper) |\n| `webxr-pwa-quest` | PWA manifest, service worker, Meta Quest packaging |\n\n### threejs-skills\n\n| Skill | Trigger |\n|-------|---------|\n| `threejs-fundamentals` | Scene setup, cameras, renderer, Object3D hierarchy |\n| `threejs-geometry` | Built-in shapes, BufferGeometry, custom geometry, instancing |\n| `threejs-materials` | PBR materials, shader materials, material properties |\n| `threejs-lighting` | Light types, shadows, environment lighting |\n| `threejs-textures` | Texture types, UV mapping, environment maps |\n| `threejs-animation` | Keyframe animation, skeletal animation, morph targets |\n| `threejs-loaders` | GLTF loading, texture loading, async patterns |\n| `threejs-shaders` | GLSL, ShaderMaterial, uniforms, custom effects |\n| `threejs-postprocessing` | EffectComposer, bloom, DOF, screen effects |\n| `threejs-interaction` | Raycasting, controls, mouse/touch input, object selection |\n\n### freshdesk-api\n\n| Skill | Trigger |\n|-------|---------|\n| `freshdesk-api` | Freshdesk API: tickets, contacts, companies, knowledge base articles, webhooks, automation |\n\nFive reference files cover the full API surface: tickets, contacts/companies, Solutions (knowledge base), webhooks/automation, and SDK examples.\n\n### doc-review\n\n| Command | Purpose |\n|---------|---------|\n| `/doc-review:review-doc <file>` | Start in-browser annotation overlay on HTML/MD/TXT; runs auto-apply loop |\n| `/doc-review:apply-review [<json>]` | Guided apply of pending review comments back into source |\n| `/doc-review:review-doc-stop` | Drop this session's references; servers exit when last ref drops |\n\nMulti-session aware: two sessions reviewing the same doc share one server. Apply button in browser queues edits; Claude applies within ~60s. Stores comments under `<project-root>/.doc-review/`.\n\n## Acknowledgements\n\n- `threejs-skills` originally sourced from [pinkforest/threejs-playground](https://github.com/pinkforest/threejs-playground) (MIT)\n- `doc-review` bundles [marked](https://github.com/markedjs/marked) (MIT), © 2011-2024 Christopher Jeffrey\n\n## License\n\n[MIT](LICENSE) © Entelligentsia. Each plugin directory carries its own copy of\nthe licence so installed plugins stay self-describing; plugins that bundle or\nderive from third-party code note it at the bottom of their `LICENSE`.\n\n",
  "bytes": 9436,
  "sha": "2245745d2c92e529616cfdb7b9ea0df5e56f5fbba76fdb200980b68c076aefb9",
  "repo_slug": "entelligentsia/skillforge",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_entelligentsia_skillforge_security_watch_483de444/readme"
}