mad-max

An unattended SoloFlow mode that processes every ready backlog task through execution, verification, code review, and test writing. A backlog is the list of pending work, and a branch is an isolated line of Git changes.

In plain words
What is it for?
Use it to drain all ready tasks, or a selected set or idea, when you want the workflow to run in the background with visual verification and end-of-sprint verification disabled.
Why use it?
It lets a run continue without interactive prompts, while recording tasks that are stuck or need a person and leaving the branch open for review.

Command

Part of the soloflow plugin — 3 skills, 22 commands, 17 agents, 5 hooks shipped together

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.

agentmods
npx agentmods add commands/kesteva/soloflow/mad-max
Clone the repo
git clone --depth 1 https://github.com/kesteva/soloflow

Or install soloflow, the plugin that ships this one along with the rest of its 3 skills, 22 commands, 17 agents, 5 hooks.

Per session 39 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,748 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00039 $0.03748
Opus 5 $0.00019 $0.01874
Sonnet 5 $0.00008 $0.00750
Haiku 4.5 $0.00004 $0.00375

Measured 3d ago against content hash 9ce09b462d47, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

mad-max 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 3d 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.

commands/mad-max.md · 225 lines

How it starts

The opening of the file, as written. The whole thing — 225 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/soloflow:mad-max

Unattended variant of /soloflow:sprint. Drains every ready task in the backlog through the full per-task quality loop (executor → verifier → code-reviewer → test-writer) and leaves the run branch open for human review. Visual verification is unconditionally off and tasks run in parallel by default (floor of 3); end-of-sprint verification is skipped. No interactive prompts during the run.

Use this when you want to start a run and walk away. Use /soloflow:sprint when you want to review scope, deferred items, and merge choice interactively.

Arguments: $ARGUMENTS (optional — specific task IDs to include, or an IDEA-NNN to filter; if empty, include all ready tasks — this is drain mode, not a capped sprint).


Model + limits resolution

Run once at the start, cache the result:

node "${CLAUDE_PLUGIN_ROOT}/scripts/config/resolve.js" --all

Mad-max spawns the same agents as /soloflow:sprint. Keys consumed:

  • models.sprint_initiator / models.executor / models.test_writer / models.sprint_closer (fallback: sonnet)
  • models.verifier / models.code_reviewer / models.sprint_verifier (fallback: opus)
  • limits.executor_retry_max / limits.checkpoint_interval / limits.context_limit_respawn_max (fallbacks: 3)
  • limits.max_parallel_tasks (fallback: 3) — controls how many task pipelines run concurrently per batch. 1 disables parallel mode.

Mad-max intentionally ignores limits.max_sprint_tasks — it drains every ready task, not a capped sprint. Reuse the cached values across respawns.

Step 0.5: Hard-stop guardrails

Run these three checks in the orchestrator before spawning anything. Any failure prints a single-line reason and stops — mad-max never prompts.

  1. Not initialized. If .soloflow/ does not exist, print: mad-max: SoloFlow not initialized. Run /soloflow:init first. and stop.

  2. Active sprint. Read .soloflow/checkpoint.md (if present) and glob .soloflow/active/sprints/*/sprint.json. If either indicates an active sprint (checkpoint says mid-execution, or any per-sprint sprint.json has sprint.status: "active"), print: mad-max: active sprint {SPRINT-NNN} detected. Run /soloflow:sprint to resume or close it first. and stop.

Read the full file on GitHub · 225 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. 3d ago First seen · 225 lines · 39 tokens per session scan A 9ce09b462d47

Subscribe to this mod's changes

mad-max is a command published in the GitHub repository kesteva/soloflow (40 stars, last pushed 3mo ago), licensed MIT. It adds 39 tokens to every session and 3,748 once invoked, about $0.0002 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.