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 tboome33/obsidian-mcp-router --skill who-is-speakinggit clone --depth 1 https://github.com/tboome33/obsidian-mcp-routerWrote 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/tboome33/obsidian-mcp-router/who-is-speaking)<a href="https://agentmods.dev/skills/tboome33/obsidian-mcp-router/who-is-speaking"><img src="https://agentmods.dev/badge/skills/tboome33/obsidian-mcp-router/who-is-speaking/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/tboome33/obsidian-mcp-router/who-is-speaking"><img src="https://agentmods.dev/badge/skills/tboome33/obsidian-mcp-router/who-is-speaking.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 | $0.00237 | $0.01386 |
| Opus 5 | $0.00118 | $0.00693 |
| Sonnet 5 | $0.00047 | $0.00277 |
| Haiku 4.5 | $0.00024 | $0.00139 |
Grade A, and why
who-is-speaking 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 9d 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 — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
who-is-speaking
Identify the family member for the current session on a shared family vault, then lock + set mode.
Examples / Exemples
Moved here from the frontmatter: the Agent Skills specification caps description at 1024
characters, and worked examples are activation-time guidance, not discovery-time metadata — the
body is where they belong under progressive disclosure.
- EN: "I'm Karine, let's go" → identify as Karine, lock to
vault_tribu, setHybridmode. - FR: "c'est Roland, on enchaîne" → identifie comme Roland, lock
vault_tribu, modeHybrid. - EN: "who's speaking today?" → ask the user (don't guess), then identify.
Pre-conditions
This skill only makes sense when the current default vault is a shared family/multi-user vault that has the tribu-routing convention installed in its CLAUDE.md. Detect this by:
- Reading the vault's
CLAUDE.mdand looking for an H2 heading matchingFamily-member auto-routing(case-insensitive) - OR by recognizing the structure
wiki/People/<N>/with multiple member subfolders
If neither, this skill doesn't apply — push back with: "This skill is for shared family vaults using the tribu-routing convention. The current default vault (<name>) doesn't appear to use this pattern. If you want to install it, run /obsidian-router:conventions install tribu-routing."
Argument parsing from $ARGUMENTS
- bare member name or alias (Roland, roro, papa, karine, max, nico, …) → match against the vault
CLAUDE.mdtable - empty / "qui parle?" / "who is speaking?" → ASK the user explicitly: "Qui parle dans cette session ? (membres reconnus : Roland, Karine, Maxence, Nicolas, Amelie + leurs alias)". Wait for the answer before proceeding.
Always
- Read the vault's
CLAUDE.mdto get the canonical members + alias table. Never hardcode the list — it's per-vault. - Match the user's input (case-insensitive) against the table. Both canonical names and aliases are valid. If match found, store the canonical name (e.g.,
roro→Roland). - No match → DON'T guess. Tell the user the list of recognized members and ask them to clarify. Refuse to proceed.
- After successful match, in sequence:
- Call
mcp__obsidian-router__lock_vault({ vault: '<vault-name>' })(use the slug of the current default family vault, typicallytribu) - Call
mcp__obsidian-router__set_auto_enrich_mode({ mode: 'Hybrid' }) - Confirm to the user: "Identifié comme . Vault verrouillé sur
<vault-slug>, mode auto-enrich Hybrid. Les saves auto routent verswiki/People/<Canonical Name>/." (or EN equivalent if user is speaking EN)
- Call
- For the rest of the session, treat the identified member as the active speaker. When auto-enrichment proposes saves (triggers 1/2/3 of the auto-enrichment consigne), prefix paths with
wiki/People/<member>/. Items that are explicitly collective →wiki/Family/. - Medical sensitivity guard stays in force even in
Hybridmode. This skill switches the vault toHybrid, which auto-saves type-safe items without asking. But per thetribu-routingconvention, health/medical data (sleep, medications, diagnoses, symptoms) must NEVER be auto-saved without explicit user confirmation — the sensitivity filter overrides Hybrid's auto-save. When in doubt about whether an item is medical, ask before saving.
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.
- 9d ago First seen · 69 lines · 237 tokens per session scan A 1f530cfc5c02
who-is-speaking is a skill published in the GitHub repository tboome33/obsidian-mcp-router (1 stars, last pushed yesterday), licensed Apache-2.0. It adds 237 tokens to every session and 1,386 once invoked, about $0.0012 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 2026-09-03.
Other skills, from other repositories
capture
Captures ideas, notes, links, and thoughts quickly. Use when the user says 'note', 'idea', 'capture', 'save this', 'remember', or wants to save something for later.
librarian
Orchestrate vault knowledge workflows using the Librarian MCP server. Ingest solutions, search prior art, explore the knowledge graph, log daily learnings.
review-scheduler
(macOS only) Spaced repetition review scheduler for Obsidian. Based on the Ebbinghaus forgetting curve, auto-reminds you to review articles at optimal intervals via macOS launchd + screen unlock detection. Triggers: "/review-scheduler", "set up review reminders", "start review scheduler", "stop review scheduler".
daily-briefing
Proactive daily briefing that fires on a recurring schedule, pulls recent memory and workspace context, composes a structured summary (action items, progress, radar, next steps), and delivers it to all active channels. Enable with a time like "set up my daily briefing at 9am". Disable, reschedule, or check status at…
llm-wiki
Build and maintain an LLM-curated personal knowledge base — the "LLM Wiki" pattern from Andrej Karpathy's April 2026 gist. Use this skill whenever the user wants to ingest a source (paper, article, transcript, PDF, notes) into a persistent compounding knowledge base, ask a question against accumulated notes, lint or…
agentic-os-obsidian
Set up an agentic OS inside an Obsidian vault — a configurable command-center dashboard with 5 auto-installed, bundled plugins (Dataview, CustomJS, Shell-commands, Terminal, Homepage), Home + per-profile + Vault Overview pages, KPI cards, sparklines, heatmap, task rollup, and a button bar wired to user-defined Claude…