CORAL is infrastructure for autonomous coding agents that run experiments on a codebase, have their work graded, share knowledge, and evolve solutions across repeated attempts. It is used for autoresearch workflows with agents such as Claude Code, OpenCode, Codex, Cursor Agent, and Kiro. The catalogue entries provide skills, agents, instructions, a plugin, and a hook for operating these workflows.
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 skills add Human-Agent-Society/CORAL --skill create-notesgit clone --depth 1 https://github.com/Human-Agent-Society/CORALWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/human-agent-society/coral/create-notes)<a href="https://agentmods.dev/skills/human-agent-society/coral/create-notes"><img src="https://agentmods.dev/badge/skills/human-agent-society/coral/create-notes/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/human-agent-society/coral/create-notes"><img src="https://agentmods.dev/badge/skills/human-agent-society/coral/create-notes.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What 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 5.1 | $0.00217 | $0.05499 |
| Opus 5.5 | $0.00087 | $0.02200 |
| Sonnet 5.5 | $0.00043 | $0.01100 |
| Haiku 4.5 | $0.00022 | $0.00550 |
Grade A, and why
create-notes 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 1mo 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 — 374 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Create Notes
A good note answers three questions a future agent will actually ask:
- What did you do, and what happened? (concrete numbers, not adjectives)
- Why did it happen that way? (the mechanism, not just the result)
- What should I do — or not do — given this? (ordered next steps + things you tried that failed)
A bad note is a wall of headings with empty bodies, or a final-design pitch with no record of the alternatives you rejected. The bad pattern shows up enough that this skill exists to prevent it.
When to Use
Each heartbeat that produces a note corresponds to one variant. The skill is one document, but you only need the variant your current trigger asks for.
| Triggered by | Variant | File location |
|---|---|---|
reflect heartbeat after each eval |
Experiment note (Variant A) | notes/experiments/eval-<N>-<slug>.md |
pivot plateau detection |
Focus note (Variant C) | notes/focus/focus-<topic>.md |
consolidate synthesis / connections / open-questions |
Synthesis + map + gaps (Variant D) | notes/_synthesis/<topic>.md, notes/_connections.md, notes/_open-questions.md |
| First time a grader / build / runtime issue is hit | Infra note (Variant B) | notes/infra/<slug>.md (or notes/<slug>.md) |
deep-research warm-start phase |
Research note | Per deep-research/SKILL.md (not duplicated here) |
If you are about to Write any file under notes/, stop and use this skill first — even if the prompt did not say "write a note."
Notes Directory Layout
The directory structure is owned by organize-files. Do not invent new top-level subdirectories; place new content in the right existing one:
notes/
├── index.md ← table of contents; you update this for any new note
├── raw/ ← immutable sources (do not write here directly)
├── research/ ← deep-research findings (link back to raw/)
├── experiments/ ← per-eval reflections, written by the reflect heartbeat
├── infra/ ← grader / build / runtime issues + workarounds (recommended)
├── focus/ ← per-agent focus declarations (owned by the pivot heartbeat)
├── migrations/ ← island-arrival notes (written by the framework)
├── _synthesis/ ← owned by consolidate; do not write here unless consolidating
├── _connections.md ← owned by consolidate
├── _open-questions.md ← owned by consolidate
└── _organization-log.md ← append-only audit log; only organize-files writes here
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago First seen · 374 lines · 217 tokens per session scan A d7dfa394e941
create-notes is a skill published in the GitHub repository Human-Agent-Society/CORAL (1,084 stars, last pushed yesterday), licensed Apache-2.0. It adds 217 tokens to every session and 5,499 once invoked, about $0.0009 per session on Opus 5.5. 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 skills, from other repositories
agent-state-bar
A design guide for an agent status bar: a short, automatically updated summary of progress, remaining tasks, tool counts, errors, and system state placed near the latest model input.
handoff
Write a compact, decision-ready handoff so the next session (or the user) can continue without reconstructing the current one. Use when the session is ending, context is running low, the user asks for a handoff / 'pass the baton' / 'hand off', or a long-running operation needs a durable state checkpoint.
knowledge-org
A guide to organising large bodies of knowledge so an AI system can navigate relationships, summaries, and layers instead of searching only flat text chunks. It covers tree indexes, knowledge graphs, file-like structures, incremental updates, and long-term memory layouts.
context-compression
A guide to shortening the information an agent carries between steps. It explains how to replace large tool outputs and old conversation details with compact, useful summaries while preserving important decisions and constraints.
memory-system
A guide to giving an AI agent persistent, user-specific memory across conversations. It covers what to remember, how to store it, how to update or compress it, and when to use memory instead of a shared knowledge base.
code-review-context
Codex maintains a context (history of messages) that is sent to the model in inference requests.