CLI Flags

This page lists the user-facing CLI flags from the current mistermorph --help and subcommand --help output.

Omitted from the main list:

• Cobra built-ins such as completion and help
• version, which currently has no command-specific flags

Global Flags

These flags are inherited by most commands:

• --config: Config file path.
• --log-add-source: Include source file:line in logs.
• --log-format: Logging format, text|json.
• --log-include-skill-contents: Include loaded SKILL.md contents in logs.
• --log-include-thoughts: Include model thoughts in logs.
• --log-include-tool-params: Include tool params in logs.
• --log-level: Logging level, debug|info|warn|error.
• --log-max-json-bytes: Max bytes of JSON params to log.
• --log-max-skill-content-chars: Max SKILL.md characters to log.
• --log-max-string-value-chars: Max characters per logged string value.
• --log-max-thought-chars: Max characters of thought to log.
• --log-redact-key: Extra param keys to redact in logs. Repeatable.

benchmark

This command accepts an optional profile-name positional argument. Without one, it benchmarks the default route plus every named LLM profile.

• --json: Output benchmark results as JSON.
• --timeout: Overall timeout for the selected benchmarks. 0 disables the timeout.

run

• --api-key: API key.
• --endpoint: Base URL for provider.
• --heartbeat: Run a single heartbeat check and ignore --task and stdin.
• --inspect-prompt: Dump prompt messages into ./dump.
• --inspect-request: Dump LLM request/response payloads into ./dump.
• --interactive: Allow Ctrl-C pause and interactive context injection.
• --llm-request-timeout: Per-LLM HTTP request timeout.
• --max-steps: Max tool-call steps.
• --max-token-budget: Max cumulative token budget.
• --model: Model name.
• --parse-retries: Max JSON parse retries.
• --provider: Provider name.
• --skill: Skill name or id to load. Repeatable.
• --skills-dir: Skills root directory. Repeatable.
• --skills-enabled: Enable configured skills loading.
• --task: Task to run. If empty, reads from stdin.
• --timeout: Overall timeout.
• --tool-repeat-limit: Force final output when the same successful tool call repeats too many times.

chat

• --api-key: API key override.
• --compact-mode: Compact display mode for interactive chat output.
• --endpoint: Provider base URL override.
• --llm-request-timeout: Per-LLM HTTP request timeout.
• --max-steps: Max tool-call steps.
• --max-token-budget: Max cumulative token budget.
• --model: Model override.
• --no-workspace: Start chat without a workspace attachment.
• --parse-retries: Max JSON parse retries.
• --profile: Named LLM profile override.
• --provider: Provider override.
• --skill: Skill name or id to load. Repeatable.
• --skills-dir: Skills root directory. Repeatable.
• --skills-enabled: Enable configured skills loading.
• --timeout: Overall timeout.
• --tool-repeat-limit: Force final when the same successful tool call repeats this many times.
• --verbose: Show info-level logs during chat.
• --workspace: Attach a workspace directory for this chat session.

console serve

• --allow-empty-password: Allow console to run without console.password or console.password_hash.
• --console-base-path: Console base path.
• --console-listen: Console server listen address.
• --console-session-ttl: Session TTL for console bearer token.
• --console-static-dir: SPA static directory.
• --inspect-prompt: Dump prompt messages into ./dump.
• --inspect-request: Dump LLM request/response payloads into ./dump.

telegram

• --inspect-prompt: Dump prompt messages into ./dump.
• --inspect-request: Dump LLM request/response payloads into ./dump.
• --telegram-addressing-confidence-threshold: Minimum addressing confidence to accept.
• --telegram-addressing-interject-threshold: Minimum interject score to accept.
• --telegram-allowed-chat-id: Allowed chat ids. Repeatable.
• --telegram-bot-token: Telegram bot token.
• --telegram-group-trigger-mode: Group trigger mode, strict|smart|talkative.
• --telegram-max-concurrency: Max number of chats processed concurrently.
• --telegram-poll-timeout: Long polling timeout for getUpdates.
• --telegram-task-timeout: Per-message agent timeout.

slack

• --inspect-prompt: Dump prompt messages into ./dump.
• --inspect-request: Dump LLM request/response payloads into ./dump.
• --slack-addressing-confidence-threshold: Minimum addressing confidence to accept.
• --slack-addressing-interject-threshold: Minimum interject score to accept.
• --slack-allowed-channel-id: Allowed Slack channel ids. Repeatable.
• --slack-allowed-team-id: Allowed Slack team ids. Repeatable.
• --slack-app-token: Slack app token for Socket Mode.
• --slack-bot-token: Slack bot token.
• --slack-group-trigger-mode: Group trigger mode, strict|smart|talkative.
• --slack-max-concurrency: Max number of Slack conversations processed concurrently.
• --slack-task-timeout: Per-message agent timeout.

line

• --inspect-prompt: Dump prompt messages into ./dump.
• --inspect-request: Dump LLM request/response payloads into ./dump.
• --line-addressing-confidence-threshold: Minimum addressing confidence to accept.
• --line-addressing-interject-threshold: Minimum interject score to accept.
• --line-allowed-group-id: Allowed LINE group ids. Repeatable.
• --line-base-url: LINE API base URL.
• --line-channel-access-token: LINE channel access token.
• --line-channel-secret: LINE channel secret for webhook signature verification.
• --line-group-trigger-mode: Group trigger mode, strict|smart|talkative.
• --line-max-concurrency: Max number of LINE conversations processed concurrently.
• --line-task-timeout: Per-message agent timeout.
• --line-webhook-listen: Listen address for the LINE webhook server.
• --line-webhook-path: HTTP path for the LINE webhook callback.

lark

• --inspect-prompt: Dump prompt messages into ./dump.
• --inspect-request: Dump LLM request/response payloads into ./dump.
• --lark-addressing-confidence-threshold: Minimum addressing confidence to accept.
• --lark-addressing-interject-threshold: Minimum interject score to accept.
• --lark-allowed-chat-id: Allowed Lark chat ids. Repeatable.
• --lark-app-id: Lark app id.
• --lark-app-secret: Lark app secret.
• --lark-base-url: Lark Open API base URL.
• --lark-encrypt-key: Lark event subscription encrypt key.
• --lark-group-trigger-mode: Group trigger mode, strict|smart|talkative.
• --lark-max-concurrency: Max number of Lark conversations processed concurrently.
• --lark-task-timeout: Per-message agent timeout.
• --lark-verification-token: Lark event subscription verification token.
• --lark-webhook-listen: Listen address for the Lark webhook server.
• --lark-webhook-path: HTTP path for the Lark webhook callback.

install

• -y, --yes: Skip confirmation prompts.

skills list

• --skills-dir: Skills root directory. Repeatable.

skills install

• --clean: Remove existing skill dir before copying.
• --dest: Destination directory.
• --dry-run: Print operations without writing files.
• --max-bytes: Max bytes to download for a remote SKILL.md.
• --skip-existing: Skip files that already exist in destination.
• --timeout: Timeout for downloading a remote SKILL.md.
• -y, --yes: Skip confirmation prompts.

tools

This command currently has no command-specific flags.