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 TadFuji/kirok-mcp --skill kirokgit clone --depth 1 https://github.com/TadFuji/kirok-mcpWrote 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/tadfuji/kirok-mcp/kirok)<a href="https://agentmods.dev/skills/tadfuji/kirok-mcp/kirok"><img src="https://agentmods.dev/badge/skills/tadfuji/kirok-mcp/kirok/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/tadfuji/kirok-mcp/kirok"><img src="https://agentmods.dev/badge/skills/tadfuji/kirok-mcp/kirok.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00049 | $0.00958 |
| Opus 5.5 | $0.00020 | $0.00383 |
| Sonnet 5.5 | $0.00010 | $0.00192 |
| Haiku 4.5 | $0.00005 | $0.00096 |
Grade A, and why
kirok 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 — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Kirok: Fully Autonomous Memory Protocol
Kirok (記録) gives you persistent, cross-session memory through its MCP tools — a "second brain" you manage proactively: record durable facts and preferences without waiting to be asked, and check memory before tasks that may depend on past context.
Note: Tools are prefixed with
KIROK_(e.g.,KIROK_retain,KIROK_recall).
🛑 Core Operational Directives
- Pre-flight Recall (The "Look Before You Leap" Rule)
When a task may touch past decisions, user preferences, or known bugs — debugging, writing a new script, planning an architecture — run
KIROK_recallfirst. - Post-Task Retain (The "Record the Lesson" Rule)
When a bug is fixed, a preference is stated, or a milestone is reached, record it with
KIROK_retain. If the fact contains credentials, personal data, or anything the user has not agreed to store, check with them before writing it. - Periodic Reflection (The "Wisdom" Rule)
If a series of complex tasks has been completed, consider using
KIROK_reflectto synthesize underlying patterns into mental models.
🏦 Bank Taxonomy & Fragmentation Prevention
To prevent memory fragmentation (which destroys recall accuracy), keep banks to a small, stable set with broad purposes — do not create arbitrary one-off banks. If your host environment already defines a bank scheme (for example via a project's CLAUDE.md, an existing configuration, or a set of banks already in use), follow that scheme. Otherwise, use these general-purpose banks as a sensible default:
| Bank ID | Purpose | Example Triggers |
|---|---|---|
user-prefs |
User personal preferences and rules | Coding style, language policy (e.g., Japanese for UI), workflow habits |
architecture |
System design and technical specs | Tool versions, specific frameworks in use, core logic choices |
troubleshooting |
Error logs, root causes, and fixes | "We found why X fails, we must use workaround Y." |
milestones |
Project achievements and work history | "Successfully migrated to Next.js on 2026-04-10" |
scratch |
Temporary or volatile memory | Unfinished ideas, pending tasks that don't need permanent record |
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 Changed · -14 lines 3bcd3d3a9611
- 1mo ago Changed · -9 lines f1c174908f6b
- 1mo ago First seen · 87 lines · 49 tokens per session scan A 3470fe6759b2
kirok is a skill published in the GitHub repository TadFuji/kirok-mcp (1 stars, last pushed 1mo ago), licensed MIT. It adds 49 tokens to every session and 958 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
lc-curate-context
Decide which files a task actually needs, record that as a reusable llm-context rule, verify it against the codebase - including the files your selection references but leaves out - and pack it for your own context, a chat, or a sub-agent you dispatch. Load when choosing what code to put in front of a model, packing…
compact-now
Operator-triggered proactive compaction — Chrono externalizes load-bearing state (active decisions, open tasks, next action) to a snapshot + a durable Vault learning note before invoking Claude Code's native /compact, then resumes from the snapshot. Use when the operator says "/compact-now" / "compact now" or when…
context-reconstruct
Re-anchor on the durable record (current-track, live owner thread, pending-questions, relay, buildlog) before acting on anything that depends on earlier context. Read, do not recall.
learning-window
Persist collection evidence and plan unread room windows separately from learned facts.
session-recap
Reconstruct what happened in a past core session — from a high-level summary down to verbatim owner quotes — by reading the raw session transcripts (complete, crash-proof, unbiased), not the curated relay/handoff notes.
cortex-consolidate
Run memory maintenance — decay old memories, compress stale content, consolidate episodic memories into semantic knowledge, and run sleep-like replay. Use when the user says 'clean up memories', 'consolidate', 'run maintenance', 'compress old memories', 'memory cleanup', or periodically to keep the memory system…