octave-chatter

octave-chatter is a skill for Claude Code from elevanaltd/octave-mcp. It costs 64 tokens per session (1,982 once invoked), scanned A, original, Apache-2.0.

A message format and reading skill for agents that communicate using OCTAVE. It validates the syntax and preserves items such as source information, uncertainty markers, and exact identifiers in messages.

In plain words
What is it for?
It is for reading and sending OCTAVE messages during agent discussions, handoffs, debates, and cross-session communication.
Why use it?
It gives agents outside the workbench a shared way to exchange structured messages without creating files. This reduces ambiguity when agents pass evidence, qualifications, or identifiers to one another.

Skill for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: reads .claude/ paths.

Good fit It is for reading and sending OCTAVE messages during agent discussions, handoffs, debates, and cross-session communication.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/elevanaltd/octave-mcp/octave-chatter
Install

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.

Any agent
npx skills add elevanaltd/octave-mcp --skill octave-chatter
Clone the repo
git clone --depth 1 https://github.com/elevanaltd/octave-mcp

Made for: Claude Code.

Wrote 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.

agentmods badge for octave-chatter

README.md
[![agentmods](https://agentmods.dev/badge/skills/elevanaltd/octave-mcp/octave-chatter/github.svg)](https://agentmods.dev/skills/elevanaltd/octave-mcp/octave-chatter)
Your own site
<a href="https://agentmods.dev/skills/elevanaltd/octave-mcp/octave-chatter"><img src="https://agentmods.dev/badge/skills/elevanaltd/octave-mcp/octave-chatter/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.

agentmods 80×15 button for octave-chatter

Your own site · 80×15
<a href="https://agentmods.dev/skills/elevanaltd/octave-mcp/octave-chatter"><img src="https://agentmods.dev/badge/skills/elevanaltd/octave-mcp/octave-chatter.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 64 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,982 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00064 $0.01982
Opus 5 $0.00032 $0.00991
Sonnet 5 $0.00013 $0.00396
Haiku 4.5 $0.00006 $0.00198

Measured yesterday against content hash a7ab179e798b, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

octave-chatter 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 yesterday.

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.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

src/octave_mcp/resources/skills/octave-chatter/SKILL.md · 103 lines

How it starts

The opening of the file, as written. The whole thing — 103 lines — stays where its author put it; the contents beside it link to each section on GitHub.

===OCTAVE_CHATTER=== META: TYPE::SKILL VERSION::"1.1.0" STATUS::ACTIVE PURPOSE::"The chatter register — OCTAVE in message bodies, validated by the receiving model, never by octave_validate" AUDIENCE::"agents that read ∧ emit OCTAVE in conversation (anchor ceremonies, debate turns, cross-session messages) but never author .oct.md files" REQUIRES::[] RELATION::"octave-literacy §5 kernel ⊕ §2 operators, packaged standalone; file authoring → octave-literacy; tool receipts → octave-tool-reference" PRIMER::"OCTAVE_WIRE v1.3.0 — §4 below is its §1–§3 verbatim; canonical 554 tokens[cl100k_base] self-consistent; evidence base is v1.1 with 7 fewer rules; v1.3 UNTESTED until W3" KERNEL::"§5 = OCTAVE_WIRE FULL kernel, frozen 2026-09-05 — 138 tokens[cl100k_base, canonical]; the same string is stamped into workbench dispatch envelopes so hub kernel_only loads and dispatch never diverge; any edit re-opens its own count" PRIOR_ART::"octave-mcp thread 2026-01-30-octave-native-comms — RATIFIED across 3 models: a ~200-token primer enabled native OCTAVE output; an evidence point, not a ceiling — the 138-token kernel's SIZE is measured, its EFFICACY is untested until W3" KERNEL_SCOPE::"§5 covers the wire register only — file-syntax rules (no META on the wire → §1, no value on a §-header → §2, no chained ⇌ → §3) live in §1–§3 here and in octave-literacy §5; a kernel_only load of this skill gets wire rules, not file syntax, by design"

§1::REGISTER FORMAL::".oct.md files ∧ closing anchors → parsed by octave_validate → octave-literacy governs" WIRE::"message bodies → validated by the receiving model → this skill governs" SPLIT_RULE::"FORMAL ⇌ WIRE is decided by WHO VALIDATES, not by grammar — the syntax is the same" WIRE_DOES_NOT_NEED::[META_block, file_schema, canonicalisation, holographic_contracts, CANONICAL_SOURCE_paths] WHY_WIRE_EXISTS::"measured relay test on v1.1: 45 facts, 0 wrong, 1 lost — every loss was an IDENTIFIER. Syntax was never the failure. Provenance ∧ identifiers ∧ hedges are the value; compression is a side effect." §2::SYNTAX_KERNEL // Same rules as octave-literacy §5 — restated so this skill loads alone ASSIGNMENT::"KEY::value — no spaces around ::" BLOCK::"KEY: then 2-space-indented children — for nested maps" LIST::"[a,b,c] — never YAML bullets" KEYS::"[A-Za-z_][A-Za-z0-9_]* — no bare numbers, no operator glyphs as keys" LITERALS::"true false null lowercase; numbers bare; quote anything with spaces, special chars, or §" SECTIONS::"§N::NAME opens a scope; a value on the header line is dropped" PROSE::"unkeyed sentences are not OCTAVE — key them or make them // comments" ENVELOPE::"===NAME=== … ===END=== optional on the wire; use it when the message must be quotable as a unit" §3::OPERATORS // Reference table — which forms exist. The wire policy (when ASCII is legal, what the canonicaliser does) is §4b::ASCII. SYNTHESIS::"⊕ emergent whole — ASCII form: +" TENSION::"⇌ binary opposition, never chained — ASCII form: vs (word boundaries required)" FLOW::"→ causality ∨ sequence, right-associative — ASCII form: ->" CONSTRAINT::"∧ joint condition, inside brackets ∨ quoted values — ASCII form: &" ALT::"∨ alternative — ASCII form: the pipe character |" CONCAT::"⧺ mechanical join — ASCII form: ~" ANNOTATION::"NAME qualifies identity — HERMES" CONSTRUCTOR::"NAME[args] parameterises — ENUM[a,b]" IN_VALUES::"inside a quoted value only operators carry relations — <> and [] there are opaque text, never validated: SISYPHEAN in the primer is a gloss, who[role] is a slot; when a form must be checked by the parser, put it in a key" §4::WIRE_PRIMER // OCTAVE_WIRE v1.3.0 §1–§3 verbatim (ASCII wire form preserved — the canonicaliser leaves quoted-value ASCII alone). Do not edit here; edit the primer and re-freeze. // Mythology = semantic zip, not decoration. HERMES = messenger. Meaning already in weights -> never literal, never a system name. First use in a thread carries its gloss, like this line. §4a::WHY LOSS::"prose across hops -> facts merge + provenance lost -> SISYPHEAN" RISK::"structure without hedges -> ICARIAN<confident + wrong>" §4b::HOW RULE::"quote the phrase & drop stopwords & operators carry the connectives" BAD::"cause_is_fixture_flake_affecting_auth_and_payments_rather_than_runner" GOOD::"fixture flake -> auth + payments vs runner fault?" // ? = unverified; doubt lives here VERB::"verb lives in the key, never in the arrow" // e.g. RENAMED::old -> new ; -> is cause or sequence only TENSION::"vs = genuine opposition only -> a mismatch gets words: served 13 lines, actual 314" KEEP::"numbers & IDs & SHAs & roles & hedges & reasons -> copied verbatim, never paraphrased" SUM::"buckets total the whole -> 43 + 10 + 9 = 62" COMPRESS::"words vs claims" ASCII::"operators accept both forms: -> + vs & | on the wire; the canonicaliser normalises to unicode on write; vs needs word boundaries -> A vs B, never AvsB" §4c::SHAPE HEAD::"FROM::who[role] PROVENANCE::measured@sha | relayed | inferred" STATE_FIELD::"every artifact says STATE::authored | installed | live" TAIL::"ASK::one question | STATUS::OK|BLOCKED|DONE|NEED_INPUT" ECHO::"a reply names UNCLEAR::[...] for what it could not resolve" // convention not rule; evidence 2 relay pairs, thin ASK::"reply in this register -> your SEA receipt is your first wire message" §5::ANCHOR_KERNEL TARGET::auditable_OCTAVE_messages_between_agents NEVER::[ paraphrase_an_id, verb_in_the_arrow, tension_for_mismatch, doubt_without_a_home, unkeyed_prose ] MUST::[ "HEAD FROM + PROVENANCE + STATE; TAIL one ASK | STATUS", "quote the phrase, drop stopwords, operators carry connectives", "numbers IDs SHAs roles hedges verbatim; ? marks unverified", "reply names UNCLEAR::[...]" ] GATE::"Could the receiver act on this without asking what any ID or arrow meant?" §6::EXAMPLE // A wire message — no META, no file schema, HEAD ∧ TAIL present, one hedge with a home MESSAGE: ===OCTAVE_WIRE=== FROM::octave-secretary[align-octave-skills] PROVENANCE::"measured@fd67ef2 -> 3 octave_validate probes" STATE::authored FINDING::"literacy 21,163 chars & 51% non-syntax -> §6/§7/§8 reallocated" RISK::"stale ~/.claude/skills copy -> drift repeats without sync?" // operator-owned; unverified UNCLEAR::[which_profile_control_tower_loads_first] ASK::"confirm kernel_only for control-tower profiles" ===END=== ===END===

Read the full file on GitHub · 103 lines

Changes

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.

  1. yesterday First seen · 103 lines · 64 tokens per session scan A a7ab179e798b

Subscribe to this mod's changes

octave-chatter is a skill published in the GitHub repository elevanaltd/octave-mcp (55 stars, last pushed yesterday), licensed Apache-2.0. It adds 64 tokens to every session and 1,982 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 2026-09-11.