Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/intellegix/intellegix-code-agent-toolkit/gba-build-developgit clone --depth 1 https://github.com/intellegix/intellegix-code-agent-toolkitWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.09427 |
| Opus 5 | $0.00000 | $0.04713 |
| Sonnet 5 | $0.00000 | $0.01885 |
| Haiku 4.5 | $0.00000 | $0.00943 |
Grade B, and why
gba-build-develop scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- MCP server registered in `~/.claude/settings.json` How it starts
The opening of the file, as written. The whole thing — 837 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/gba-build-develop — Self-Driving GBA Game Development Pipeline
Composes /gba-build develop, /gba-ai-full, /research-perplexity, /creative-research, and /orchestrator-multi into a single autonomous mega-pipeline that researches, plans, parallelizes implementation, AI-playtests, iterates on failures, discovers creative features, and ships production-ready GBA games.
Usage
/gba-build-develop "<game description>"
Arguments: $ARGUMENTS = A natural-language description of the GBA game to build (e.g., "dungeon crawler with 3 floors, enemies, and a boss fight").
7-Stage Pipeline
STAGE 0: INIT ─── Load skill, compile context, detect project state
│
STAGE 1: RESEARCH ─── /research-perplexity → architecture + module plan
│
STAGE 2: ORCHESTRATE ─── /orchestrator-multi patterns → parallel agent build
│
STAGE 3: AI PLAYTEST ─── Generalized /gba-ai-full → save state + assert
│
STAGE 4: RESEARCH ITERATION ─── Feed test results → fixes/improvements
│ ╰─► Loop to STAGE 2 if fixes needed
STAGE 5: CREATIVE DISCOVERY ─── /creative-research adapted for GBA
│ ╰─► Quick Wins loop to STAGE 2
STAGE 6: SHIP ─── Full playthrough, profiling, production audit, docs
Prerequisites
- devkitPro installed (
pacman -S gba-dev) - Butano (default engine) or libtonc —
BUTANO_PATHset if using Butano - mGBA running with Lua MCP server on port 61337
- git initialized in project directory
- loop_driver.py at
"C:\Users\AustinKidwell\ASR Dropbox\Austin Kidwell\04_ResearchAnalysis\automated claude\automated-loop\loop_driver.py" - MCP server registered in
~/.claude/settings.json
Frame Timing Constants (Research-Derived)
These constants are hardware-derived from GBA specs (280,896 cycles/frame at 16.78 MHz) and validated against mGBA Lua API behavior. Use these exact values — never guess.
| Constant | Frames | Use Case |
|---|---|---|
| FRAMES_BOOT_BIOS | 270 | After gba_flash_rom() with BIOS intro enabled |
| FRAMES_BOOT_SKIP | 10 | After gba_flash_rom() with BIOS skipped |
| FRAMES_BOOT_BUTANO | 15 | After gba_flash_rom() for Butano engine (heavier C++ init) |
| FRAMES_TITLE_TO_GAME | 60 | After pressing START — covers 32f fade-out + 5f load + 20f fade-in + 3f settle |
| FRAMES_INPUT_SETTLE | 3 | After press_button() before reading memory (key_poll + logic + 1 frame margin) |
| FRAMES_INPUT_VISUAL | 4 | After press_button() before screenshot (input + OAM DMA copy) |
| FRAMES_SCENE_TRANSITION | 75 | After triggering scene change (32f fade + 10f VRAM load + 32f fade + 1f settle) |
| FRAMES_FAST_TRANSITION | 40 | Scene change with same tileset / no fade |
| FRAMES_COMBAT_INIT | 90 | From encounter trigger to stable battle screen (RPG) |
| FRAMES_COMBAT_INIT_ACTION | 45 | From encounter trigger to stable combat (action game, simpler transition) |
| FRAMES_SAVE_COMPLETE | 3 | After triggering SRAM save (1f write + 2f state machine margin) |
| FRAMES_SOFTLOCK_TIMEOUT | 900 | 15 seconds of zero state change = confirmed softlock |
| FRAMES_SCREENSHOT_NORMAL | 60 | Screenshot interval during normal gameplay (1/sec) |
| FRAMES_SCREENSHOT_TRANSITION | 5 | Screenshot interval during transitions (high granularity) |
| FRAMES_SCREENSHOT_GAMEPLAY | 30 | Screenshot interval during active testing (~2/sec) |
| FRAMES_STATE_CHECK_FINE | 5 | Memory read interval for position/HP (catches movement) |
| FRAMES_STATE_CHECK_MODE | 1 | Game mode byte check interval (every frame during transitions) |
| FRAMES_STATE_CHECK_COARSE | 30 | Memory sentinel / corruption check interval |
| FRAMES_CHECKPOINT_INTERVAL | 300 | Save state safety checkpoint (~5 seconds) |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 837 lines · 0 tokens per session scan B ff89c984d813
gba-build-develop is a command published in the GitHub repository intellegix/intellegix-code-agent-toolkit (57 stars, last pushed 8d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 9,427 tokens. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
README
Git workflow and quality assurance commands for the claude-skills repository.
agentlas-cloud
Staff a task only from the signed-in owner's Agent Cloud agents.
dispatcher
Pick the next-best repo to work on across the portfolio — rank free repos, recommend one, claim its lease atomically, and route to the entry command.
tldr
Re-apply TLDR rules for this turn (verdict first, no filler).
me-figure-discussion
Write or review thermal-fluid figure discussion with observation, physical mechanism, comparison, uncertainty, and claim limits.
execute
Interactive workflow for workspace isolation, pane-delegated implementation, testing, review, and cleanup.