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 skills/febradc-github/turnstile/turnstile-brain-initnpx skills add febradc-github/turnstile --skill turnstile-brain-initgit clone --depth 1 https://github.com/febradc-github/turnstileWrote 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/febradc-github/turnstile/turnstile-brain-init)<a href="https://agentmods.dev/skills/febradc-github/turnstile/turnstile-brain-init"><img src="https://agentmods.dev/badge/skills/febradc-github/turnstile/turnstile-brain-init.svg" alt="Measured on agentmods" 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 | $0.00065 | $0.01159 |
| Opus 5 | $0.00032 | $0.00580 |
| Sonnet 5 | $0.00013 | $0.00232 |
| Haiku 4.5 | $0.00006 | $0.00116 |
Grade A, and why
turnstile-brain-init 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 4d 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 — 33 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Brain Init
Process
- Deliver the vault-size and staleness warning above and get the explicit opt-in confirmation. Declined: point at the lazy default (review keeps touched files' notes current) and stop.
- Verify the turnstile-brain MCP tools respond (any search_notes call). Unavailable: tell the user to reload the session so .mcp.json registers the server, and stop. No turnstile/ directory: create nothing; suggest running a turnstile command (or /turnstile:install-obsidian) first, and stop.
- Build the source file list:
- In a git repo:
git ls-files(respects .gitignore). Otherwise: a Glob sweep. - Keep only these extensions: .js .mjs .cjs .ts .tsx .jsx .py .rb .go .rs .java .kt .c .h .cpp .hpp .cs .php .swift .sh .ps1 .psm1 .sql .vue .svelte .lua .r .scala .ex .exs
- Always exclude: turnstile/, node_modules/, vendor/, dist/, build/, .git/, lockfiles, minified files (.min.), generated files.
- Nothing left after filtering: say so and stop.
- In a git repo:
- List existing code notes and their path aliases -- grep the
aliases:frontmatter line of each note in turnstile/code/ rather than reading whole notes (a missing code/ folder means no existing notes). Drop every file whose path matches an existing alias. Build one path -> slug map across ALL remaining files at once (lowercase; every run of characters outside a-z0-9 becomes one-, trimmed at both ends). Resolve collisions -- against existing notes AND between new files -- by appending -2, -3, ... in path-sorted order. - Identify stale notes: every code note whose aliased path no longer exists in the repo. Do not delete anything yet.
- Plan the batches: group by top-level directory, about 15 files per batch (split large directories, merge tiny ones).
- Report the plan -- N files in M batches, K skipped as already documented, S stale notes to remove, each stale note named -- and confirm once with the user (AskUserQuestion). After confirmation: delete the stale note files, then dispatch brain-curator once per batch with model: sonnet, at most 4 dispatches in flight. Each dispatch prompt carries: the batch's file paths with slugs, the full linkable-slugs list (every path -> slug in this run plus every existing code note), and the statement that this is bulk code documentation mode.
- After all batches return, dispatch brain-curator once more (default model) to stitch: for each top-level source directory, create or extend turnstile/architecture/AR-.md (root-level files: AR-root.md) with a ## Files section linking that directory's file notes in [[slug|path]] form (extend an existing AR note covering the area rather than creating a parallel one); run MOC upkeep for this run's file notes (shared top-level tag code, so moc-code is created or extended once 5+ file notes exist); fix every unresolved link this run introduced, including ones the batch dispatches left behind -- the stitch prompt explicitly authorizes this, overriding the curator's report-don't-fix default; include the slugs of the stale notes deleted in step 6 so the curator runs list_backlinks on each and removes dangling references; finish with list_changed_notes acknowledge: true.
- Report: notes written, AR notes created or extended, files skipped or failed (a failed batch's files are listed and picked up by a rerun -- they have no notes, so the skip check passes them through), and that /turnstile:obsidian-graph shows the result.
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.
- 4d ago First seen · 33 lines · 65 tokens per session scan A 79a687304a1a
turnstile-brain-init is a skill published in the GitHub repository febradc-github/turnstile (1 stars, last pushed 1mo ago), licensed MIT. It adds 65 tokens to every session and 1,159 once invoked, about $0.0003 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 2026-08-31.
Other skills, from other repositories
self-service-and-knowledge
Builds the help center, in-product guidance, and knowledge base that let customers resolve problems without contacting anyone — content, findability, maintenance, and deflection measurement. Use this to build or fix a help center, reduce support volume, write documentation for customers, improve findability, or decide…
agent-memory
3-tier markdown memory protocol (shared/agent/conversation) for cross-session knowledge. TRIGGER when: reading or writing agent memory files, choosing which memory tier an insight belongs in, or starting a task needing prior context. SKIP: vector recall (use semantic-memory-mcp); distilling conversations into…
memory
Unified memory CRUD: write/update/delete/query file-based markdown entries and sync to MCP semantic store. TRIGGER when: saving a new memory entry, updating or deleting an existing entry, querying stored memories, or deduplicating before write. SKIP: 3-tier agent memory (use agent-memory); in-task vector recall (use…
semantic-memory-mcp
Proactive use of semantic-memory MCP tools (semanticsearch, semanticrecall). TRIGGER when: an agent should search vector memory for prior insights mid-task. SKIP: file-based 3-tier memory (use agent-memory); storing a new memory (use semantic-memory-store).
distill
Knowledge distillation methodology: candidate extraction, confidence scoring, tier routing, conversation-scoped mode. TRIGGER when: running /learn or /distill, deciding if an insight is memory-worthy, or scoring a knowledge candidate. SKIP: ad-hoc memory reads/writes (use agent-memory); vector storage (use…
operate-hermes-agent
Install, update, migrate to, configure, run, secure, and troubleshoot Hermes Agent across its CLI, TUI, desktop, dashboard, profiles, tools, skills, memory, models, goals, voice, sessions, worktrees, and terminal backends.