Octo Shared logo

Octo Shared

Organization
Mininglamp-OSS
octo-shared

Shared knowledge for using the octo CLI — authentication, unified gateway config, output envelopes, universal flags, error handling, and common patterns. Load before invoking any octo domain skill.

Overview

PublisherMininglamp-OSS
Repositoryocto-cli
Skill nameocto-shared
Stars
918
Forks
130
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by Mininglamp-OSS on GitHub. Read the source before you install it.

Installation

Install the Octo Shared AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/Mininglamp-OSS/octo-cli/tree/main/skills/octo-shared
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Mininglamp-OSS/octo-cli.git /tmp/octo-cli
mkdir -p .claude/skills
cp -r /tmp/octo-cli/skills/octo-shared .claude/skills/octo-shared
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Octo Shared in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Octo Shared on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Octo Shared is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

octo-shared — CLI fundamentals for AI Agents

octo-cli is a thin REST client that exposes the Octo ecosystem (matters, messaging, groups, threads, files, bot, events, docs, html) as a single binary. Every service command is auto-generated from an embedded OpenAPI registry; output is a JSON envelope designed to be parsed by agents.

The matter domain is temporarily withheld while its backend API stabilizes — octo-cli matter ... is not registered and the octo-matter skill is not listed. Do not emit matter commands until it is re-enabled. The examples below use other domains.

Bots cannot delete documents, even documents they created or own/administer. Do not invoke docs delete, html rm, or equivalent raw api DELETE calls to delete a document. Ask a human with document admin permission to delete it in Octo instead. Do not switch credentials/routes or clear the document's contents as a workaround. This rule concerns whole documents, not normal editing or otherwise-permitted comment, version, asset, or sheet row/column removal.

1. Authentication

Bots authenticate with a bearer token. There is no interactive user login — but besides the two bot tokens (app_*, bf_*) there is a third kind, a user API key (uk_*), which carries a real person's identity and is used mainly for message search. Two ways to supply any of them:

Stored profile (recommended). A human (or provisioning step) logs the token in once; it is encrypted at rest under ~/.octo-cli, and the raw token never appears in any command line, shell history, or transcript afterward:

bash
# Operator setup (token read from a hidden prompt, or --with-token < file):
octo-cli auth login --bot-id cli_xxxxxxxx          # robot id you got when creating the bot
echo "$TOKEN" | octo-cli auth login --bot-id cli_xxxxxxxx --with-token   # non-interactive

# A user API key (uk_) is stored the same way — encrypted profile, bot_kind shows
# user_key. Use a friendly --profile name since it has no robot id:
echo "$UK_TOKEN" | octo-cli auth login --profile alice-search --with-token

Then, at runtime, select which bot to act as — the agent passes its own robot id, which it knows:

bash
octo-cli --bot-id cli_xxxxxxxx matter list         # or env OCTO_BOT_ID=cli_xxxxxxxx
octo-cli --profile myname matter list              # or by the friendly profile name

With exactly one stored profile, the selector is optional. With two or more, you must pass --bot-id or --profile — omitting it is a hard error (the CLI never guesses which identity to use).

Env token (fallback). When no profile is stored, the raw token is read from OCTO_BOT_TOKEN:

bash
export OCTO_BOT_TOKEN=app_xxxxxxxxxxxxxxxxxxxx      # App Bot (DM-only)
export OCTO_BOT_TOKEN=bf_xxxxxxxxxxxxxxxxxxxxx       # User Bot (full access)
export OCTO_BOT_TOKEN=uk_xxxxxxxxxxxxxxxxxxxxx       # User API key (real person; message search)

OCTO_BOT_ID is a selector (a robot id), not a secret; OCTO_BOT_TOKEN is the secret. If OCTO_BOT_ID names no stored profile the command fails — it never falls back to silently using OCTO_BOT_TOKEN under that id.

Token prefix determines capability — the CLI does NOT enforce this locally (the backend rejects unsupported operations with FORBIDDEN), with one exception: an app_* token running message search is rejected locally with a validation error before any request.

PrefixTypeDM msgGroup readGroup writeThreadVoiceSearch
app_*App Botyesyesnononono
bf_*User Botyesyesyesyesyesyes
uk_*User API key—————yes

uk_* carries a real person's identity and is meaningful mainly for message search (routed to /v1/user/*); for other domains use a bot token. bf_* can also search on behalf of a person with --on-behalf-of <uid>. identity.bot_kind reflects the kind: app_bot, user_bot, or user_key.

Outside OCTO_CREDENTIAL_MODE=task, inspect octo-cli auth status (or octo-cli config show) to confirm the active identity. Daemon task mode does not expose profile diagnostics; use the identity echoed by each success envelope and report authentication failures to the daemon instead (see §3).

2. Unified gateway configuration

All Octo domains use one gateway. Override it only for test or self-hosted deployments:

bash
# Production defaults to https://im.deepminer.com.cn. Override for test or
# self-hosted deployments:
export OCTO_API_BASE_URL=https://im-test.deepminer.com.cn

export OCTO_SPACE_ID=space_xxx                     # only for platform-scoped bots
export OCTO_FORMAT=json                            # default output format

Routing: all services go through OCTO_API_BASE_URL. The --service flag on octo-cli api is for documentation only — all traffic routes to the same gateway.

3. Output: the JSON envelope

Every successful invocation prints a single JSON object to stdout:

json
{
  "ok": true,
  "identity": { "type": "bot", "profile": "prod", "robot_id": "cli_xxx", "bot_kind": "app_bot", "source": "profile:prod" },
  "data": { ... or [...] },
  "_pagination": { "has_more": true, "next_cursor": "..." },
  "_rate_limit": { "remaining": 99, "reset": 1730000000 }
}

identity echoes the bot the command actually ran as — check it to catch acting as the wrong identity. It is always an object: a stored profile fills in profile / robot_id / source: "profile:<name>"; a raw OCTO_BOT_TOKEN yields { "type": "bot", "bot_kind": ..., "source": "env:OCTO_BOT_TOKEN" } (no profile/robot_id); a command that resolves no credential (e.g. version) yields the minimal { "type": "bot" }.

Every failure prints an error envelope to stderr and exits non-zero:

json
{
  "ok": false,
  "error": {
    "type": "validation",
    "code": "VALIDATION_ERROR",
    "message": "title is required",
    "hint": "check params with `octo-cli schema <op>`",
    "detail": { ...original backend payload... }
  }
}

Parse ok first. On failure, branch on error.type (a small fixed taxonomy) or error.code (a string, may come straight from the backend).

error.code: bot_delete_forbidden (error.type: permission, HTTP 403) is a terminal policy denial, not missing membership or a transient error. Do not retry, request a higher bot role, or change uid, owner, Space, credentials, or routes; ask a human document admin to perform the deletion. Do not claim success, even if the document was already deleted.

Backends differ in their raw error shape. The CLI normalizes both:

  • matters (structured): {error:{code, message, details}} → passes through into detail unchanged.
  • dmworkim (flat): {msg, status} → mapped to code/message via HTTP status.

4. Universal flags

These flags work on every command (they are root-level persistent flags):

FlagPurpose
--formatjson (default) · table · csv · ndjson
--jq, -qApply a jq expression to the success envelope before formatting
--dry-runPrint the resolved request instead of sending it
--verboseLog request/response trace to stderr
--timeoutPer-request deadline, e.g. 30s, 2m
--no-retryDisable the default retry-on-transient policy
--spaceOverride OCTO_SPACE_ID for this invocation
--bot-idSelect/assert the stored credential by robot id (env OCTO_BOT_ID)
--profileSelect the stored credential by profile name

Paginated operations additionally expose:

FlagPurpose
--page-allWalk pages until has_more=false, emit one merged array
--page-limitHard cap on pages fetched with --page-all (default 10)

5. Error taxonomy and exit codes

error.typeExitTypical error.code
auth_error3UNAUTHORIZED, AUTH_UNAVAILABLE
validation2VALIDATION_ERROR, PAYLOAD_TOO_LARGE
config2missing env vars
permission1FORBIDDEN, SPACE_FORBIDDEN
rate_limited1RATE_LIMITED
network1NETWORK_ERROR, UPSTREAM_UNAVAILABLE
api_error1MATTER_NOT_FOUND, NOT_FOUND, INTERNAL_ERROR
internal1CLI-side bug

Agents should switch on error.code first (specific, deterministic), then error.type (broad), then exit_code (coarse).

The hint field is a one-line next action meant for an agent: follow it literally where it applies. E.g. MATTER_NOT_FOUND → "verify ID with octo-cli matters list".

6. Input patterns

Promoted flags vs --data

Simple top-level body fields auto-promote to typed flags (strings, integers, booleans, []string). For objects, arrays-of-objects, or when sending a large payload, use --data:

bash
octo-cli thread create group-abc --name "design review"
octo-cli message send --data '{"channel_id":"chat-1","channel_type":1,"payload":{"type":1,"content":"hi"}}'
octo-cli message send --data @body.json
octo-cli some-cmd --data @-            # read JSON from stdin

Explicit flags override fields set in --data. The --data escape hatch exists on every non-multipart command.

Piping with --jq

bash
octo-cli group list --jq '.data[].id' | xargs -I{} octo-cli group get {}

Paginating

bash
octo-cli docs search --keyword "spec" --page-all --page-limit 20

--page-all applies to any list operation that reports a cursor in _pagination. The merged output drops _pagination — you get a flat data array.

Dry-run for agent self-verification

bash
octo-cli message send --data '{"channel_id":"chat-1","channel_type":1,"payload":{"type":1,"content":"foo"}}' --dry-run

Prints the exact HTTP request body and URL, emits no side effect.

7. Discovering the API

The registry is embedded in the binary — no network needed:

bash
octo-cli schema --list                # all services + operation IDs
octo-cli schema --list message        # operations in one domain
octo-cli schema message.send          # full request/response schema
octo-cli config show                  # resolved config (token masked)
octo-cli auth status                  # active bot identity (whoami)
octo-cli auth list                    # stored profiles (no tokens)

When an operation isn't auto-registered yet or you need low-level control:

bash
octo-cli api GET  /api/v1/messages --params '{"chat_id":"chat-1"}'
octo-cli api POST /api/v1/messages --data @body.json

8. Domain skills

Once these fundamentals are understood, load the skill for the domain you need:

  • octo-matter — matters (todos/tasks), assignees, channels, timeline, AI extract — temporarily withheld (backend API stabilizing; not currently loadable)
  • octo-messaging — message send/edit/sync/read-receipt, groups, threads, events
  • octo-files — file upload/download, presigned credentials, bot housekeeping
  • octo-docs — docs domain (CRDT/Yjs): documents, spreadsheets, whiteboard scenes, members/sharing, comments, versions, attachments
  • octo-html — HTML docs domain (octo-doc, a DIFFERENT backend from octo-docs): self-contained interactive HTML documents, share codes, media assets, comments, agent element read/replace
  • octo-summary — create owner-only summaries from explicit sources, then discover, read, and cite summaries visible to the personal Agent's human owner — temporarily withheld (create backend at Mininglamp-OSS/octo-smart-summary#181 not yet merged/deployed/enabled; not currently loadable)

Frequently asked questions

What does the Octo Shared AI skill do?

Shared knowledge for using the octo CLI — authentication, unified gateway config, output envelopes, universal flags, error handling, and common patterns. Load before invoking any octo domain skill.

Why use Octo Shared on TypingMind?

Because you install it once and use it with any model. Octo Shared is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Octo Shared in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Mininglamp-OSS/octo-cli/tree/main/skills/octo-shared. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Octo Shared?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Octo Shared?

As many as you like. As long as a model supports skills, you can use Octo Shared with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Octo Shared AI skill free?

Yes. It is published on GitHub by Mininglamp-OSS under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇