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 instructions/aigorahub/elves/agents-mdgit clone --depth 1 https://github.com/aigorahub/elvesWhat 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 | $0.03173 | $0.03173 |
| Opus 5 | $0.01587 | $0.01587 |
| Sonnet 5 | $0.00635 | $0.00635 |
| Haiku 4.5 | $0.00317 | $0.00317 |
Grade A, and why
elves AGENTS.md scanned grade A with 0 findings 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 3d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Elves: Codex repository adapter
This file is a thin Codex adapter, not a second workflow fork. The compact
canonical workflow is SKILL.md — every contract below is a pointer into it or into one authoritative
references/ file, never a restatement. Differences here are invocation surface only;
workflow semantics, the safety kernel, landing policy, and acceptance identity are identical for
Claude Code and Codex (see references/host-parity.md).
The task-first user guide is published at https://aigorahub.github.io/elves/. This adapter and
SKILL.md remain authoritative when the guide is too short to cover an edge case.
Cobbler
Use Cobbler via $elves cobbler: <task>, $elves council: <task>, or natural language
("Ask the Cobbler…"). Do not invent top-level Codex slash commands; Claude Code managed aliases
(/cobbler, /setup-cobbler, …) are Claude-specific surfaces. Full Cobbler protocol: SKILL.md
## Cobbler.
Codex Goals are optional host continuation plumbing (references/codex-goals.md) — distinct
from Grok Build goal mode, the optional worker capability that is a
capability-proven enhancement with a recorded one-packet fallback; Grok models come only from the
authenticated live catalog (references/adaptive-worker-routing.md).
v2.24 run tools (host-neutral)
The v2.24 helpers — futile re-drive guard (redrive), learnings ledger (learnings),
observed-usage ledger (usage), salvage previews (salvage), and the continuity watchdog
manager (continuity) — are host-neutral CLI helpers with identical semantics on Claude
Code, Codex, Grok Build, and Oh My Pi: invoke them as
python3 "$ELVES_SKILL_ROOT/scripts/cobbler_agents.py" <verb> … from any host. They are
advisory instruments (never landing, merge, credential, or routing authority), there are no
per-host slash surfaces for them — do not invent top-level Codex or Grok commands — and the
workflow contracts that mandate them (SKILL.md worker-failure recovery, labor completeness,
Skill Memory) apply to all four hosts unchanged. See the guide's "v2.24 run tools" section.
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.
- 3d ago First seen · 188 lines · 3,173 tokens per session scan A 72a252d02571
elves AGENTS.md is an instructions file published in the GitHub repository aigorahub/elves (213 stars, last pushed 7d ago), licensed MIT. It adds 3,173 tokens to every session, about $0.0159 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
godmode AGENTS.md
Instructions for arbazkhan971/godmode, covering agents.md — godmode for ai coding agents, core workflow: the godmode loop, key principles, authoring discipline and how to use skills.
godmode GEMINI.md
Instructions for arbazkhan971/godmode, covering godmode for gemini cli, verify installation, tool mapping, how to use skills and subagents (7 built-in).
coco-workflow CLAUDE.md
Instructions for skullninja/coco-workflow, covering claude.md -- coco, project overview, architecture, key files and tracker (lib/tracker.sh).
pilot CLAUDE.md
Claude Code instructions for qf-studio/pilot, covering pilot: ai that ships your tickets, who is reading this file?, ⚠️ git & worktree discipline (all sessions), quick commands and plan a feature (navigator).
evolve-loop AGENTS.md
Instructions for mickeyyaya/evolve-loop, covering agents.md — cross-cli canonical instructions, what evolve-loop is, cross-cli invariants (the universal rules), 1. pipeline ordering is non-negotiable and 4. builder writes only inside its worktree.
evolve-loop GEMINI.md
Instructions for mickeyyaya/evolve-loop, covering project instructions (gemini cli), skill discovery, runtime adapter (tier-1-hybrid), tool name translation and invariants (apply to gemini context too).