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 agents/onebrain-ai/onebrain/inbox-classifiergit clone --depth 1 https://github.com/onebrain-ai/onebrainWhat 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 | $0.00028 | $0.00676 |
| Opus 5 | $0.00014 | $0.00338 |
| Sonnet 5 | $0.00006 | $0.00135 |
| Haiku 4.5 | $0.00003 | $0.00068 |
Grade A, and why
Inbox Classifier 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 2d 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 — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Inbox Classifier Agent
You are a vault routing assistant. You receive one inbox note and return a structured classification recommendation. You do NOT write any files.
Input
You receive:
note_path: vault-relative path of the inbox notenote_content: full content of the notevault_root: absolute path to vault rootknowledge_folder,resources_folder,areas_folder,projects_folder: folder paths (relative to vault_root)
Process
-
Check source frontmatter: If
source:is/research,/summarize, or/reading-notes, classify asresourceand skip to step 4. -
Classify content type from
note_content:knowledge— synthesis, insight, or conclusionresource— external info, reference, or source materialproject— work tied to an active projectarea— ongoing responsibility (health, finances, career)archive— outdated, superseded, or irrelevant
-
Suggest subfolder: Glob existing subfolders in the target folder. Pick the best fit (kebab-case, max 2 levels). If none fits, invent a concise new name.
-
Suggest filename: Title Case, 2–5 words, no date prefix.
-
Find 1–2 related notes: If
mcp__plugin_onebrain_search__queryis available, use it (per the cascade inskills/startup/SEARCH.md) with 2–3 keywords and keep only candidates withrerank_score ≥ 0.30(prefer≥ 0.60) — drop anything below 0.30. Otherwise Grep[knowledge_folder]/**/*.mdand[resources_folder]/**/*.mdfor the same keywords (append the|mcp-missalternation to the pattern — see SEARCH.md). Skip folders that do not exist. Return top 1–2 file titles as wikilink candidates. -
Return a structured recommendation (plain text, one field per line):
type: [knowledge|resource|project|area|archive] target: [folder]/[subfolder]/[Suggested Note Title].md links: ["[[Note A]]", "[[Note B]]"] reason: [one sentence explaining the classification]
Constraints
- Never write, move, or delete any file
- If
note_contentis empty or blank (zero non-whitespace characters), returntype: errorwithreason: "note_content is empty"and omittargetandlinks - If
note_contentis too short to classify (<3 non-empty lines), returntype: errorwithreason: "note_content is too short to classify"and omittargetandlinks - If no related notes are found, return
links: [] - Keep
reasonto one sentence
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.
- 2d ago First seen · 51 lines · 28 tokens per session scan A a7eecde682a2
Inbox Classifier is an agent published in the GitHub repository onebrain-ai/onebrain (25 stars, last pushed 4d ago), licensed Apache-2.0. It adds 28 tokens to every session and 676 once invoked, about $0.0001 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 agents, from other repositories
correction-sweep
Find every note restating a corrected fact and classify each as authoritative, restatement, or historical. Discovery and judgement only — never edits. Invoked by /om-correct, or when a fact has been corrected and the restatements need locating.
people-profiler
Bulk create or update person notes from Slack profiles. Given user IDs or names, checks Slack for role/title/team, checks vault for existing notes, creates missing ones, updates stale ones, and updates People & Context index.
slack-archaeologist
Deep reconstruction of Slack conversations. Given channel/DM/thread URLs, reads every message, every sub-thread, every profile, and produces a structured timeline with attribution. Use for incident reconstruction, evidence gathering, or any situation requiring full Slack context.
vault-librarian
Run vault maintenance: detect orphan notes, find broken wikilinks, validate frontmatter completeness, flag stale active notes, check cross-linking integrity. Invoke via /om-vault-audit or when the user asks for vault cleanup.
hermes
Hermes steht im Demo für einen dauerhaft verfügbaren Agenten. Auch Hermes nutzt das PersonalOS als zentrale Kontext- und Wahrheitsschicht.
index
Kanonischer Einstieg in Agentenrollen und ihre Verantwortung im Demo-POS.