Skills
Reusable capability modules that commands and agents load on demand. Each skill is a SKILL.md file with structured instructions, operations, and usage examples. 141skills, generated directly from the framework's skill definitions.
| g-skl-3d-animation | Comprehensive 3D character animation, skeletal rigging, bone retargeting, and WebGL Three.js AnimationMixer runtime execution. Includes seated/typing/idle/walking pose management, blend trees, armature binding, and headless Blender animation transfer. Parent skill for the 3D pipeline family (generation via Meshy/Tripo, headless Blender processing) — see "Which specialization do I need?" below. |
| g-skl-adrs | Own and manage ADRs (Architecture Decision Records) — Context/Decision/Consequences/Alternatives-considered records with a proposed→accepted|rejected→superseded lifecycle. Distinct from a D-record (owner ruling). |
| g-skl-api-doc-gen | Auto-generate OpenAPI 3.1 specs from FastAPI/Express/Flask routes; fill docstring gaps for undocumented functions; update README API tables. Covers Python FastAPI, Express.js, and Flask. Also generates MCP tool descriptions for FastMCP plugins. |
| g-skl-argument-diagrams | Author a hand-built SVG diagram that makes a visual ARGUMENT (not a decorative box-and-line dump), then run a render-look-fix loop (screenshot, inspect, edit, repeat) against an original quality checklist before treating it as done. Complements the existing Mermaid subsystem/dependency graphs (g-skl-subsystems, g-skl-dependency-graph) -- use this for one-off explanatory/marketing/architecture diagrams that need real visual composition, not auto-generated relationship graphs. |
| g-skl-auto-triage | >- Triage non-code defects (spec_defect, policy_incongruity, design_gap): assess risk, auto-fix bounded safe issues, and log the rest as tracked bugs. |
| g-skl-backlog-curate | > Thorough backlog-vs-code curation. Walks live tasks/bugs against shipped code and proposes dispositions (complete-as-shipped, close-out-only, cancel-competes, leave-open, needs-owner). Never auto-cancels. |
| g-skl-browser-use | > Production browser harness for agentic web tasks requiring persistent sessions, stealth/anti-detect, CAPTCHA solving, and self-healing CDP code generation via the browser-use library (YC W25, $17M seed). Use for login-required sites, anti-bot environments, competitive intel, and long-running multi-step web tasks. |
| g-skl-bugs | Own and manage all bug data — BUGS.md index, bugs/ individual files, bug fixes, quality metrics. Single source of truth for everything bug and quality related. |
| g-skl-cartograph | >- Query Cartograph, gald3r's code-graph surface (`gald3r graph`). Suggest this whenever the user asks about the code graph, cartograph, blast radius, callers/callees, or "what would break if I change this" — do not wait for them to name Cartograph. Complements g-skl-subsystems and g-skl-dependency-graph (those are markdown/subsystem maps, not this tree-sitter artifact). |
| g-skl-carver-matrix | >- CARVER weighted prioritization matrix invoked by g-cm / g-carver-matrix: score items on Criticality, Accessibility, Recuperability, Vulnerability, Effect, and Recognizability with a documented 1–5 rubric and configurable weights, then render a ranked matrix. Accepts a user-supplied list or items pulled from the open gald3r backlog. |
| g-skl-cli-claude | Claude Code CLI (claude command) — headless flags, session continuation, MCP config, permissions, multi-agent via Agent SDK, and overnight/CI best practices. |
| g-skl-cli-codex | OpenAI Codex CLI (codex command) — approval modes, sandbox config, provider/model flags, session continuation, MCP tool integration, multi-agent coordination, and overnight/CI best practices. |
| g-skl-cli-copilot | GitHub Copilot CLI (gh copilot command) — natural language to shell/git/gh commands, agent mode, chat sessions, workspace instructions, and multi-agent coordination via .github/copilot-instructions.md. |
| g-skl-cli-cursor | Cursor CLI (agent command) — modes, Cloud Agent handoff, API mode, session management, and multi-agent coordination from the terminal. |
| g-skl-cli-jcode | jcode CLI platform skill — Rust-based coding agent with 14ms startup (245x faster than Claude Code), local embeddings, multi-session, MIT licensed. Use for low-overhead local tasks, batch loops, and resource-constrained environments. |
| g-skl-cli-opencode | OpenCode CLI (opencode command) — stub. OpenCode is an emerging AI coding tool from sst.dev. Full documentation pending stable release. |
| g-skl-code-review | Code review — security, quality, performance, reusability. Structured report with severity ratings and action items. Scales from quick scan to comprehensive architecture review. |
| g-skl-codename | Own the internal codename/nickname registry (.gald3r/config/CODENAMES.md) — map internal codenames (Valkyrie, CRASHV, Sunstone, ...) to a plain descriptive name, folder/repo/file refs, and a definition, so a codename never becomes a requirement for code, folder, or repo naming. Resolve fuzzy, description-shaped queries to the right codename. |
| g-skl-comfyui | ComfyUI V5 agent integration for local GPU image/video generation. Install, launch, manage workflows, and run AnimateDiff/SDXL pipelines via the ComfyUI REST API. Local GPU path — zero cloud costs. Use for example_desktop theme pack animation or any offline image/video generation. |
| g-skl-compliance | SCA/license compliance scanning — wraps ORT, FOSSA, Snyk, and PMD CPD behind a unified gald3r interface with SCAN/REPORT/GATE/STATUS operations. Produces structured compliance reports with pass/warn/fail verdicts. |
| g-skl-compress-memory | Compress the NON-gald3r sections of AGENTS.md/CLAUDE.md (and *memory*.md) to cut token overhead, while strictly preserving the install-managed gald3r SECTION ranges, code blocks, and URLs. Dry-run by default; apply only after confirmation. |
| g-skl-computer-use | > OS-level computer-use harness for Gald3r — capture the screen and issue mouse/keyboard input at OS coordinates to drive ANY native desktop app (not just a browser) end-to-end without a human. Built on the five existing computer-use tools (screenshot/click/type/scroll/key) plus a deterministic, opt-in UI-test driver. Use for autonomous GUI testing of the Throne desktop app and any native-window verification the g-go loop needs. |
| g-skl-constraints | Own and manage .gald3r/CONSTRAINTS.md — add, update, deprecate, check, and surface project constraints at session start. |
| g-skl-context-builder | Dynamic context assembly from live .gald3r/ state. Builds a token-budgeted agent context block including active tasks, constraints, subsystem specs, and session memory. Use at session start or when handing off between agents in g-go pipelines. |
| g-skl-crawl | Native crawl4ai web crawler — FETCH, BATCH, and SETUP operations. Produces clean LLM-optimized Markdown from any URL. No Docker, no MCP required. Shared crawl primitive used by g-skl-ingest-docs, g-skl-ingest-url, and g-skl-harvest. |
| g-skl-cross-review | Cross-vendor code review — a reviewer on a DIFFERENT LLM vendor judges only the diff + acceptance contract; blocking findings re-dispatch the implementer. |
| g-skl-crr | Clean-Room Rewrite pipeline. Orchestrates 4 phases via independent background subagents — capture a source repo as a whole-system, consumer-neutral functional spec centralized in the shared vault (research/CRR_FunctionalSpecs/), write all findings to IDEA_BOARD (mandatory), triage tasks, and produce a gald3r-native clean-room implementation spec. |
| g-skl-curator | > Autonomous skill library curator — grades every SKILL.md against a structured rubric (recency, clarity, scope overlap, token efficiency, last-invoked rate), proposes consolidations and archive-candidate tags for low-scoring or overlapping skills, and writes a structured audit log. Never deletes — only proposes. Safe for unattended scheduled runs. |
| g-skl-delegate | Engineering team delegation workflow skill — delegation patterns adapted for AI agent orchestration. Task briefs, code review request templates, quality gates, and handoff protocols. |
| g-skl-dependency-audit | Scan package files for outdated or vulnerable dependencies. Generates a severity-ranked report with CVE references and upgrade commands. Supports Python (requirements.txt/pyproject.toml), JavaScript/Node (package.json/package-lock.json), and Rust (Cargo.toml/Cargo.lock). |
| g-skl-dependency-graph | Generate and update .gald3r/DEPENDENCY_GRAPH.md from task file dependencies. Auto-triggered when tasks are created or updated with dependency changes. Also callable via @g-dependency-graph. |
| g-skl-design | UI/UX design engineering — turns functional AI output into production-grade interfaces: oklch color, typography, tokens, dark themes, motion, anti-AI-slop rules, six-step workflow. |
| g-skl-drift | Compute a deterministic, code-computed drift score per task-claim territory and per dev from local .gald3r state, and emit it as a local report or a g-skl-json-output-compatible JSON payload. Serves D15 (territory leasing + onboarding + drift skills). |
| g-skl-features | Own and manage all feature data — FEATURES.md index, features/ individual files, staging lifecycle (staging→specced→committed→shipped), harvest source collection, and feature promotion. Single source of truth for everything feature-related. |
| g-skl-ghostwrite | Rewrite supplied text so it reads like a person wrote it, not a model. Varies sentence length, cuts hedging and throat-clearing openers, removes anything that exists only to sound smart, bans em dashes, and swaps in the simpler word every time. Both @g-gw and @g-ghostwrite activate this skill. |
| g-skl-gilfoyle | >- Senior-expert answer mode invoked by g-gilf / g-gilfoyle: skip 101-level context, lead with tradeoffs edge cases and the judgment call a beginner would miss, stay honest concise factual, no pleasantries. |
| g-skl-git-commit | Create well-structured git commits following gald3r conventions, with proper type prefixes, task references, and clean trailers (no AI co-author footers per C-021). |
| g-skl-github-pr | GitHub PR/issue lifecycle for software_development projects — open/ready/comment/close PRs, link issues. Gated on github_integration; no-op when disabled. gh-CLI transport. |
| g-skl-graphify | > NOT YET IMPLEMENTED (BUG-898, T837 Phase A) — target design doc only. No graphify-cli, no gitNexus integration, and no `.mcp.json` entry exist in this codebase; the "T871 wired gitNexus into .mcp.json" claim below was false and has been struck. Documents a PLANNED code-graph capability for 71x-fewer-token architecture queries. Do not invoke; fall back to `gald3r search` or Grep until this capability actually ships. |
| g-skl-html-output | Render human-facing reports (status, review, backlog) as themed HTML using .gald3r/reports/templates/ + .gald3r/reports/themes/. Operations RENDER, CHOOSE_THEME, VALIDATE, EXPORT. Invoked by --html flag commands (T1318). Never used for coordination files (TASKS.md, BUGS.md, task specs) which are always markdown. |
| g-skl-ideas | Own and manage IDEA_BOARD.md — capture ideas instantly, review the board, promote to tasks, and run proactive codebase scans for improvement opportunities. |
| g-skl-ipb | >- Intelligence Preparation of the Battlespace adapted to landscape analysis: a systematic four-step process for understanding the operational environment and ranking likely and most-dangerous opposition courses of action, feeding g-ooda or g-mdmp. |
| g-skl-json-output | Emit gald3r command output (status, review, backlog) as structured JSON for scripting, CI gates, dashboards, and agent-to-agent handoff. Operations SERIALIZE, SCHEMA, VALIDATE, EXPORT. Invoked by --json flag commands (T1381). Mirrors g-skl-html-output. Coordination state files stay markdown. |
| g-skl-keep-it-simple | User-invoked terse mode toggle. Suppresses the active personality pack, response structure scaffolding, and footer metadata so the agent returns bare signal only. Intended for debugging, rapid lookups, and high-volume Q&A sessions where the standard ceremony adds noise. Deactivates at session boundary or on explicit toggle-off. |
| g-skl-knowledge-refresh | Audit vault freshness and structure. Check stale notes, rebuild compiled pages, detect broken links, missing pages, and contradictions worth review. |
| g-skl-learn | File-only continual learning — agents self-report insights to project and vault memory notes. No JSONL scanning, no schedulers, no SentinelOne risk. |
| g-skl-linking | > Unified linking file-mirror (D7 / T1610). Pulls the server-owned world_tree linking registry (parent/child/sibling edges + project_type/skills, keyed by project UUID — T1625) and writes the human-readable local mirror under .gald3r/linking/. Online the mirror reconciles against the registry; offline the local mirror is authoritative and reconciles on reconnect. Reconcile is non-destructive: conflicting edges open a review item, never overwrite. |
| g-skl-longship | >- Drive Longship, gald3r's agentic focus TUI (`gald3r-longship`). Suggest this whenever the user wants the TUI, a full-screen terminal agent, an interactive coding session instead of a scrolling REPL, or names longship — do not wait for them to name it. Complements g-skl-throne (desktop GUI) and the headless `gald3r chat` / `gald3r run` reference executors. |
| g-skl-marketing | > AI-powered marketing system for gald3r projects. Deploys specialized growth agents across SEO, GEO (AI search visibility), content, community, and launch channels. Designed for indie founders, solopreneurs, and small teams where distribution is the bottleneck — not building. |
| g-skl-mdmp | >- Military Decision-Making Process adapted to general planning: the seven-step heavyweight staff-planning process (mission analysis through COA development, wargaming, comparison, approval, and orders production) for large or ambiguous efforts. |
| g-skl-medic | > Tiered .gald3r/ health and intervention system. L1=triage (structural), L2=diagnosis (plan coherence), L3=surgery (cross-subsystem interface audit), L4=ecosystem (linked project negotiation). Replaces g-skl-medic. |
| g-skl-memory | > Capture structured insights, session summaries, and search cross-session agent memory via example_app MCP tools. Dual-mode: degrades to file-based g-skl-learn when backend is unavailable. Full semantic search requires Docker backend (adv tier). |
| g-skl-monitor | > Manage scheduled content monitors for YouTube playlists, docs sites, GitHub repos, and URLs via example_app MCP tools. Requires Docker backend (adv tier). Monitors trigger automatic re-ingestion when new content is detected. |
| g-skl-muninn | > NOT YET IMPLEMENTED (BUG-898, T837 Phase A) — target design doc only. No `gald3r muninn` verb, no `.gald3r/muninn/` db, no muninn plugin dir, and no `.mcp.json` entry exist in the current Go binary. Documents a PLANNED codebase knowledge graph (impact analysis, caller chains, dependencies, symbol search) — clean-room rewrite (T1147 epic, T1153-T1158) of the retired GitNexus integration. Do not invoke; fall back to `gald3r search` or Grep until this capability actually ships. |
| g-skl-ooda | >- Run the supplied situation through Colonel John Boyd's Observe-Orient- Decide-Act loop: raw facts, then the context/constraints/bias that shape them, then a reasoned decision, then a concrete next step plus the signal to re-loop. |
| g-skl-oracle | > Execute Oracle Database queries and operations via example_app MCP tools. Requires Docker backend (adv tier). Supports read-only queries (oracle_query) and full write operations (oracle_execute) including DDL and PL/SQL blocks. |
| g-skl-plan | Own and manage PLAN.md (master strategy) and features/ (individual Feature files) — create plans, stage features, validate scope, and keep the deliverable index current. |
| g-skl-platform-aider | Authoritative reference for Aider (terminal AI pair-programmer) customization in gald3r projects. Covers .aider.conf.yml, CONVENTIONS.md (native rules), built-in slash commands, architect/editor chat modes, auto-lint/test triggers, model roles, and gald3r install verification. |
| g-skl-platform-antigravity | Authoritative reference for Google Antigravity (agent-first IDE + CLI + SDK) customization in gald3r projects. Covers AGENTS.md/GEMINI.md instruction files, .agents/ rules/skills/hooks + .antigravity/mcp.json, Workflows (slash commands), dynamic subagents, lifecycle hooks, Scheduled Tasks, and gald3r install verification. Post-2.0-relaunch (I/O 2026); skill path + hook payload pin via install test. |
| g-skl-platform-astrbot | Authoritative reference for AstrBot customization in gald3r projects. Covers curated (BUG-137, human-assessed, not yet live-crawled) support signals for Hooks/Rules/Skills/Commands/MCP, the still-unverified folder hierarchy and instruction file, and how to upgrade this spec from curated to verified. |
| g-skl-platform-augment | Authoritative reference for Augment Code (Auggie CLI + VS Code + JetBrains) customization in gald3r projects. Covers .augment/ commands/agents/skills/hooks/rules + settings.json MCP, CLAUDE.md/AGENTS.md and .claude/.agents reuse, plugins/marketplace, and gald3r install verification. |
| g-skl-platform-claude | Authoritative reference for Claude Code (Anthropic) customization in gald3r projects. Covers .claude/ commands/agents/skills/hooks/rules + settings.json hooks/MCP, CLAUDE.md (imports @AGENTS.md, NOT AGENTS.md-native), plugins/Agent SDK/Routines/Channels, and gald3r install verification. |
| g-skl-platform-cline | Authoritative reference for Cline (VS Code / JetBrains extension + CLI + SDK) customization in gald3r projects. Covers .clinerules rules + workflows-as-slash-commands, .cline/skills SKILL.md, lifecycle hooks (macOS/Linux only), SDK subagents/teams, MCP marketplace, AGENTS.md (not CLAUDE.md), and gald3r install verification. |
| g-skl-platform-codex | Authoritative reference for OpenAI Codex (codex CLI / IDE / app) customization in gald3r projects. Covers config.toml, AGENTS.md instruction file, .agents/skills Agent Skills, .codex/agents TOML subagents, 10-event hooks (hooks.json / [hooks]), [mcp_servers.*] MCP, execpolicy/Memories rules, plugins/apps, and gald3r install verification. |
| g-skl-platform-continue | Authoritative reference for Continue.dev (open-source VS Code / JetBrains extension) customization in gald3r projects. Covers .continue/ rules/prompts/agents/skills/mcpServers, the corrected commands path (.continue/prompts, NOT .continue/commands), the absence of a native AGENTS.md read, the absent hook surface, and gald3r install verification. |
| g-skl-platform-copilot | Authoritative reference for GitHub Copilot customization in gald3r projects. Covers the .github/ tree (copilot-instructions.md, instructions/, prompts/, agents/, skills/, hooks/*.json), AGENTS.md + .claude/.agents skill reuse, Agentic Memory, surface fragmentation (VS Code / CLI / JetBrains / cloud agent), MCP, and gald3r install verification. |
| g-skl-platform-crawl | Crawl platform documentation and web docs with crawl4ai; write .platforms/ and vault markdown with frontmatter. For platform docs, knowledge refresh, web ingest. |
| g-skl-platform-cursor | Authoritative reference for Cursor IDE customization in gald3r projects. Covers .cursor/ folder layout, all supported primitives (rules/skills/agents/commands/hooks/MCP), AGENTS.md instruction file, .claude/.codex/.agents cross-tool reuse, parity tiers, and install verification. |
| g-skl-platform-deepcode | Authoritative reference for Deep Code CLI (lessweb/deepcode-cli, a third-party community terminal AI coding assistant for the deepseek-v4 model family) customization in gald3r projects. Covers AGENTS.md, .agents/skills + .deepcode/skills Agent Skills, MCP inside .deepcode/settings.json, the fixed built-in slash-command set (no user commands), the absent hook/subagent surface, and gald3r install verification. |
| g-skl-platform-gald3r | Authoritative reference for gald3r ITSELF as a platform — a user driving `gald3r run`/`gald3r chat` directly as their harness, with no external IDE/agent tool in front of it. Covers the 38th-roster-entry decision record (T847): platform id `gald3r`, the open overlay-location question (root dirs vs `.gald3r/platform/`, pending Friday owner ratification), which components the runtime can honor first (hooks + commands, per owner sequencing), and the relationship to task714 (self-coordinator provider) and task821 (recursive agent primitive, the agents execution layer). |
| g-skl-platform-goose | Authoritative reference for Goose (Block) AI agent customization in gald3r projects. Covers .goosehints + AGENTS.md instructions, ~/.config/goose/config.yaml MCP extensions, Recipes/Subrecipes as slash commands, native subagents, Agent Skills (SKILL.md, shared via ~/.claude/skills/), lifecycle hooks (hooks.json), and gald3r install verification. |
| g-skl-platform-hermes | >- Authoritative reference for Hermes Agent (Nous Research — self-improving CLI + Telegram/Discord/Slack gateway) customization in gald3r projects. Covers ~/.hermes/ config (config.yaml/.env), AGENTS.md-native instruction loading (auto-injected with SOUL.md/CLAUDE.md/.cursorrules), agentskills.io SKILL.md skills (the gald3r distribution opportunity), delegate_task subagents, MCP (mcp_servers), the native config.yaml hooks: surface (17+ events, pre_tool_call blocking), and gald3r install verification. |
| g-skl-platform-junie | Authoritative reference for JetBrains Junie (IDE plugin + Junie CLI) customization in gald3r projects. Covers .junie/ commands/agents/skills + AGENTS.md guidelines + mcp.json + EAP SessionStart hooks, the extension bundle, and gald3r install verification. |
| g-skl-platform-kimi | Authoritative reference for Kimi Code CLI (Moonshot AI, MoonshotAI/kimi-code, renamed from kimi-cli) customization in gald3r projects. Covers AGENTS.md, .kimi-code/skills + .agents/skills Agent Skills, YAML sub-agent specs, config.toml [[hooks]] lifecycle hooks, native MCP, and the post-rebrand config-directory rename (.kimi/ -> .kimi-code/). |
| g-skl-platform-kiro | Authoritative reference for Kiro (Amazon's agentic IDE + Kiro CLI, the Q Developer CLI rebrand) customization in gald3r projects. Covers .kiro/ steering/prompts/agents/skills + settings/mcp.json, Agent Hooks (IDE file-event + CLI lifecycle), AGENTS.md instruction file, spec-driven workflow, and gald3r install verification. |
| g-skl-platform-kiro-cli | Authoritative reference for Kiro CLI (Amazon's terminal agent, the Q Developer CLI rebrand) customization in gald3r projects. Covers .kiro/steering/ + AGENTS.md, Agent Skills (SKILL.md), JSON custom agents + subagents, lifecycle hooks, slash commands, MCP, and gald3r install verification. Distinct from Kiro IDE (g-skl-platform-kiro). |
| g-skl-platform-mimo | Authoritative reference for Xiaomi MiMo-Code (terminal-native AI coding agent, a fork of OpenCode with persistent cross-session memory) customization in gald3r projects. Covers AGENTS.md + CLAUDE.md dual instruction files, MEMORY.md persistent memory, .mimocode/agents custom agents, MCP in mimocode.json, Compose-mode/Distill skills equivalents, and the partial/inherited hook surface pending verification. |
| g-skl-platform-mistral | Authoritative reference for Mistral Vibe CLI (mistral-vibe) coding agent customization in gald3r projects. Covers the .vibe/ config tree (TOML), AGENTS.md instructions, native Agent-Skills SKILL.md, native custom agents + subagents, MCP servers, experimental hooks, and honest capability boundaries. |
| g-skl-platform-monitor | Cross-platform health and freshness monitor for the registry-driven gald3r platform roster (PLATFORM_REGISTRY.yaml — single source of truth, T516). Checks per-platform capability gaps against the Cursor reference, scans official docs for breaking changes, validates platform-specific config, and generates the PLATFORM_STATUS / PLATFORM_CAPABILITY_MATRIX living indexes. Owned by g-agnt-platformer. |
| g-skl-platform-openclaw | Authoritative reference for OpenClaw (self-hosted AI agent gateway — Discord/Telegram/Slack/WhatsApp) customization in gald3r projects. Covers the AGENTS.md + SOUL.md instruction-file convention, folder-per-skill SKILL.md (which double as slash commands), TypeScript HOOK.md hooks, sessions_spawn sub-agents, native MCP client+server, and gald3r install verification. |
| g-skl-platform-opencode | Authoritative reference for OpenCode (sst/opencode) customization in gald3r projects. Covers .opencode/ folder layout, opencode.json config, native skills discovery (.opencode/skills/ + .claude/skills/ + .agents/skills/), AGENTS.md/CLAUDE.md instructions, JS/TS plugin hooks, MCP, and gald3r install verification. |
| g-skl-platform-openhands | Authoritative reference for OpenHands (All Hands AI, formerly OpenDevin) AI agent customization in gald3r projects. Covers AGENTS.md context, .agents/skills + .agents/agents File-Based Agents, .openhands/hooks.json lifecycle hooks, config.toml MCP, Plugins bundle, and gald3r install verification. |
| g-skl-platform-pi | Authoritative reference for Pi (badlogic/pi-mono terminal coding-agent harness) customization in gald3r projects. Covers hierarchical AGENTS.md/CLAUDE.md instructions, native Agent Skills (SKILL.md), native prompt-template slash commands, TypeScript-extension lifecycle hooks (pi.on events), the absent MCP surface, and gald3r install verification. |
| g-skl-platform-qwen | Authoritative reference for Qwen Code (Alibaba CLI coding agent, a Gemini CLI fork) customization in gald3r projects. Covers .qwen/settings.json hooks+MCP, QWEN.md context (NOT AGENTS.md by default), custom commands, subagents, Agent Skills (SKILL.md), and gald3r install verification. |
| g-skl-platform-replit | Authoritative reference for Replit Agent (cloud IDE) customization in gald3r projects. Covers replit.md Agent instructions, Agent Skills in /.agents/skills, Plan/Build modes + effort tiers, MCP (client + hosted server), and cloud-IDE constraints (no hooks, no commands). |
| g-skl-platform-roo | Authoritative reference for Roo Code (VS Code extension; formerly Roo Cline) customization in gald3r projects. Covers .roo/rules/, custom modes (.roomodes), slash commands, Agent Skills (.roo/skills + .agents/skills), MCP, AGENTS.md, and gald3r install verification. NOTE — Roo Code was discontinued 2026-05-15. |
| g-skl-platform-subq | Authoritative reference for SubQ (Subquadratic Inc.) and "SubQ Code" in gald3r projects. SubQ is a long-context LLM (OpenAI-compatible Chat Completions API); SubQ Code is a PLUGIN / long-context layer that plugs into Claude Code, Codex, and Cursor — it exposes NO native commands/rules/agents/skills/hooks/MCP of its own. gald3r installs target the HOST tool, not SubQ. |
| g-skl-platform-void | Authoritative reference for Void (VS Code fork) customization in gald3r projects. Covers curated (BUG-137, human-assessed, not yet live-crawled) capability signals — Hooks/Skills/Commands curated not-supported, Rules via the legacy .cursorrules file, and MCP curated supported — plus the still-unverified folder hierarchy and instruction file. |
| g-skl-platform-warp | Authoritative reference for Warp (AI terminal + Oz agent platform) customization in gald3r projects. Covers AGENTS.md/WARP.md project rules, Agent Skills (SKILL.md) cross-vendor discovery, Agent Profiles + Oz subagent orchestration, Warp Drive Workflows + MCP, the absent hook surface, and gald3r install verification. |
| g-skl-platform-windsurf | Authoritative reference for Devin Desktop (formerly Windsurf — Cognition renamed the product 2026-06-02 after acquiring it; Cascade IDE) customization in gald3r projects. Covers .windsurf/ rules/workflows/skills/hooks + ~/.codeium/windsurf MCP, AGENTS.md/.windsurfrules instruction files, .claude/.agents skill reuse, and gald3r install verification. |
| g-skl-platform-zcode | Authoritative reference for ZCode (Z.ai / Zhipu, GLM-5.2 Agentic Development Environment) customization in gald3r projects. Covers the two-scope AGENTS.md convention (global-then-workspace, no imports/hierarchy), native Agent Skills (SKILL.md), native slash commands, Beta global-only subagents, native MCP, the absent hooks surface, and gald3r install verification. |
| g-skl-platform-zed | Authoritative reference for Zed (agent panel + ACP host) customization in gald3r projects. Covers the .rules-first project-instructions precedence list, native AGENTS.md (project + personal ~/.config/zed/AGENTS.md), native Agent Skills (.agents/skills/, shared with Codex/Amp/Deep Code), native MCP (context_servers), External Agents hosted via the Agent Client Protocol (agent_servers), the absent hooks/commands surfaces, and gald3r install verification. |
| g-skl-plugins | Authoritative reference for the gald3r plugin system — install, remove (uninstall), update, list, and per-project enable/disable git-cloned, SKILL.md-based third-party plugins via `gald3r plugin <verb>`. Documents the live `PluginManager` mechanism (git clone/pull/rmtree, GLOBAL install, no manifest, no ledger) plus the per-project ON/OFF association overlay (T287), and honestly flags the two operations (scaffold/NEW and CHECK_COMPAT) that are not implemented. Single source of truth for everything plugin-related. |
| g-skl-policy | Policy-as-code guardrail — loads an org policy bundle (file offline, world_tree online) and exposes a CHECK op used by the pre-tool-call and pre-commit hooks to deterministically block/allow. Team/Org tier only; free/retail installs run an empty/default bundle. |
| g-skl-prds | Own and manage all PRD data — PRDS.md index, prds/ individual files, governance lifecycle (draft→review→approved→in-implementation→released→archived), revision chain, and freeze enforcement. Parallel artifact to Features for compliance, audit, and external sign-off. |
| g-skl-project | Own and manage PROJECT.md (mission, goals, project linking). Single source of truth for project identity. Constraints operations delegated to g-skl-constraints. |
| g-skl-project-agent | >- Talk to THIS repo as its always-on project agent (`gald3r project-agent status|ask|chat`). Suggest this whenever the user asks the project a question, wants a board/search answer without hand-editing `.gald3r/`, or wants the project-as-agent seat instead of a second Cursor folder. Complements g-skl-valk (live other-repo) and g-skl-wpac-* (file transport). |
| g-skl-qa | Track bugs, QA, and fixes in .gald3r/. For bug reports, issues, quality, fixes, QA workflows. |
| g-skl-recon-docs | Documentation URL ingestion with periodic revisit. Crawls URLs into research/platforms/, tracks staleness per _index.yaml, surfaces stale count at session start. Depends on g-skl-crawl (crawl4ai). |
| g-skl-recon-file | Capture a local file (PDF, DOCX, XLSX, TXT, MD, CSV, RTF) into the vault knowledge store. Extracts text content, generates Obsidian-compatible vault notes in research/files/. |
| g-skl-recon-repo | Capture a GitHub repository into the vault as a structured summary note. Produces research/repos/ notes and _index.yaml. Optional --deep flag triggers full g-skl-res-deep analysis pass. |
| g-skl-recon-url | One-time URL ingestion into vault research/articles/. No revisit scheduling. Dedup by source URL in _index.yaml. Depends on g-skl-crawl (crawl4ai). |
| g-skl-recon-yt | YouTube transcript ingestion into the vault. Uses yt-dlp to fetch transcripts locally — no Docker, no MCP, no screen captures. Stores in research/videos/ with analysis_depth=transcript_only for future vision upgrade. |
| g-skl-release | >- Own and manage all release data — RELEASES.md index and releases/ individual files. Operations: CREATE new release, ASSIGN tasks to a release, STATUS summary, PUBLISH ROADMAP.md, ACCELERATE target dates with cascading shift to subsequent planned releases, SYNC reconcile CHANGELOG entries against release records (C-023). |
| g-skl-res-apply | > Convert a reverse-spec FEATURES.md (produced by g-skl-res-deep) into gald3r artifacts: project goals, PRDs, subsystem specs (merge or create), and tasks. Vault-aware — reads from {vault}/research/CRR_FunctionalSpecs/{slug}/FEATURES.md when a shared vault is configured, else falls back to local research/CRR_FunctionalSpecs/{slug}/FEATURES.md. This is the "Execute" layer in the Discover → Curate → Execute harvest pipeline. |
| g-skl-res-deep | Analyze any external repository and produce a whole-system, consumer-neutral functional specification (FEATURES.md + supporting notes). Centralizes output in the shared vault at {vault}/research/CRR_FunctionalSpecs/{slug}/ so a run from any project/workspace lands in one place reusable across projects; falls back to local research/CRR_FunctionalSpecs/{slug}/ only when no shared vault is configured. Documents the ENTIRE source system regardless of relevance to any one consumer project. Performs cross-project dedup via _recon_index.yaml. Agents are reporters — humans are editors. No .gald3r/ writes until human approves APPLY. |
| g-skl-res-review | Analyze external sources (GitHub repos, URLs) for adoptable patterns and improvements. Vault-aware — reads from {vault}/research/CRR_FunctionalSpecs/ when a shared vault is configured, else falls back to local research/CRR_FunctionalSpecs/. Uses _recon_index.yaml for cross-project dedup. Produces structured harvest reports and optional IDEA_BOARD suggestions. Zero-change-without-approval. |
| g-skl-review | Code review for security, quality, performance, reusability; structured report with severity and actions. |
| g-skl-security-scan | SAST-style static analysis for hardcoded secrets, injection patterns, insecure deserialization, path traversal, and other critical vulnerabilities. Severity-ranked findings with line references and remediation guidance. Includes two-phase threat-model + diff-revalidation mode (T1167) for use as the post-implementation gate in g-go-review. |
| g-skl-sesh | Session-only goal control (Sunstone) -- the CURRENT session's own ordered heading list, consulted on drift or "where were we". Not project goals, not milestones, not the task board. |
| g-skl-setup | Initialize gald3r in a project — folder structure and template files for task management. |
| g-skl-ship | > Semantic versioning and release management for gald3r projects. Promotes CHANGELOG [Unreleased] to a versioned release, bumps VERSION, updates README badge, creates git tag, and optionally publishes a GitHub release. Also handles incremental CHANGELOG entries during development. |
| g-skl-status | Show project status — session context, active tasks, phase progress, goals, ideas. |
| g-skl-stz | >- Sun Tzu strategy advisor — counsels the supplied situation through Art of War doctrines (win before fighting, know both sides, terrain and timing, economy of force, shaping, strike the weak), naming the principle behind each recommendation. |
| g-skl-subsystem-graph | Generate and update .gald3r/SUBSYSTEM_GRAPH.md from subsystem spec file dependencies. Auto-triggered when subsystems are created or updated with dependency changes. Also callable via @g-dependency-graph --subsystems or @g-subsystem-graph. |
| g-skl-subsystems | Own and manage SUBSYSTEMS.md (registry + mermaid graph) and subsystems/ spec files. Tracks what subsystems exist, their boundaries, dependencies, and activity logs. |
| g-skl-swot-review | >- Automated SWOT analysis for the current project phase. Reviews progress, architectural compliance, code quality, goal alignment, and technical debt. Runs weekly via heartbeat or on-demand. |
| g-skl-tasks | Own and manage all task data — TASKS.md index, tasks/ individual files, status transitions, sync validation, complexity scoring, and sprint planning. Single source of truth for everything task-related. |
| g-skl-territory | Lease, renew, release, and check status of subsystem/path territories with TTLs — the coordination primitive multiple coordinators use to safely partition a project without stepping on each other's work. |
| g-skl-test | Create, maintain, and run multi-level test plans (fast, comprehensive, regression) for gald3r systems and features. Use when creating test plans, running tests, checking test coverage gaps, doing code review, verification, or when preparing a release. Enforces C-013/C-014/C-015 constraints. Triggered by @g-test command and g-agnt-test agent. |
| g-skl-theme-editor | Create and edit gald3r HTML themes against .gald3r/reports/themes/theme-schema.json. Visual editor (live preview, per-token color pickers, import/export :root blocks) ships in example_desktop; this skill is the spec + a file-first fallback that works without the app. Invoked by g-theme-edit. |
| g-skl-thinker | >- Deep-thinker reasoning mode invoked before answering a non-trivial question: lay out assumptions, weigh at least two genuinely different angles, stress-test the first instinct, then land on a clear, committed verdict. |
| g-skl-throne | >- Drive Gald3r Throne, the desktop workspace control plane (codename Hlidskjalf). Suggest this whenever the user is looking at the desktop control plane, wants the GUI, a visual task/bug board, or a local-first desktop app for .gald3r/ — do not wait for them to name Throne. Complements g-skl-longship (TUI) and g-skl-valk (live connector). |
| g-skl-todos | Own and manage TO-DOs — a commitment tier between an idea (may never be used) and a task (actively worked). List/show for everyone; add/promote are USER-WRITE-ONLY. |
| g-skl-toon-output | Emit gald3r output as TOON (Token-Oriented Object Notation) — a compact, lossless, LLM-friendly format that states record keys once (tabular arrays) to cut tokens vs markdown/JSON. Operations ENCODE, DECODE, VALIDATE, EXPORT. Invoked by --toon flag commands (T1382). Coordination state files stay markdown. |
| g-skl-valk | >- Drive the resident Valkyrie world_tree connector (`gald3r valk`). Suggest this whenever the user needs a live question to another project, mid-swarm human reach-back, connector status, or world_tree login — do not wait for them to name Valkyrie. Complements g-skl-wpac-* (file transport) and g-skl-project-agent (this repo). |
| g-skl-vault | Own and manage the file-first vault plus repo mirror metadata. Obsidian-compatible notes, wiki compilation, path resolution, reindexing, linting, and GitHub repo summaries. |
| g-skl-verify-ladder | >- Multi-level verification gates for autonomous task completion. Configurable levels from minimal (lint) to thorough (tests + acceptance + hallucination guard). |
| g-skl-wishlist-mine | Mine a free-form, human-prose intent/wishlist document into formal gald3r task specs — READ-ONLY against the prose doc, dedup against existing tasks, curation discipline (epics for vision, backlog-candidates for unsure). Productizes the DELIVERABLES.md mining pattern. |
| g-skl-workspace | Workspace-Control Mode skill. Reads .gald3r/linking/workspace_manifest.yaml as the canonical registry and provides STATUS, VALIDATE, MEMBER LIST, SPAWN, ADOPT, EXPORT/SYNC dry-run planning, and member `.gald3r/` marker bootstrap/remediate/validate operations for manifest-declared repositories. |
| g-skl-wpac-adopt | Register another project as a child of the current project. Creates or updates linking/link_topology.md on both sides when the target is locally accessible. |
| g-skl-wpac-ask | Child project requests parent action — writes parent INBOX.md and marks local task as blocked with cross-project metadata. |
| g-skl-wpac-claim | Register another project as the parent of the current project. Creates or updates linking/link_topology.md on both sides when the parent is locally accessible. |
| g-skl-wpac-move | > Transfer files or folders from this project to another project in the topology (parent, child, or sibling) with provenance tracking, safety gates, and vault log entries in both projects. Use when migrating subsystems between repos, promoting assets to the canonical template, or extracting code that grew in one project but belongs in another. |
| g-skl-wpac-notify | Send lightweight [INFO] freeform notifications to one or more project INBOXes — no task created, no approval needed. FYI-only cross-project messaging. |
| g-skl-wpac-order | As a parent project, push a task to one or more child projects with configurable cascade depth (1–3). Creates tasks in child .gald3r/ folders and an INBOX notification. |
| g-skl-wpac-read | Review and action all incoming cross-project coordination items — conflicts (block planning), requests from children, broadcasts from parents, and peer syncs from siblings. |
| g-skl-wpac-send-to | > Send files, features, specs, ideas, or code from the current project to any related project in the topology (parent, sibling, or child). Handles file copying, INBOX notification, vault provenance logging, and optional source cleanup. Lighter-weight than g-skl-wpac-move (which requires topology pre-registration); this skill also works when the destination is a recently-spawned project. |
| g-skl-wpac-spawn | > Spawn a new gald3r project from the current project. Creates the new project folder in the same ecosystem root, installs gald3r (matching the current project's install type — symlinks or fresh template), seeds it with any passed description/features/code, runs gald3r-setup, and immediately links both projects via WPAC topology (--parent | --sibling | --child). |
| g-skl-wpac-sync | Initiate or respond to sibling contract sync — advisory only, non-blocking. Both sides update local peer topology copies. |
| g-skl-yt-video-analysis | MCP or full-pipeline video analysis — vault notes must match Obsidian standard. For local yt-dlp transcripts only, use g-skl-ingest-youtube. |