An autonomous work-loop command that keeps implementing the current task list until a defined stopping point is reached. It checks project rules, budgets, and required pauses before each run.

Use it to continue an implementation slice, track completed tasks, and stop when the task, slice, phase, budget, or a governance gate requires it.

It removes the need to manually restart work between tasks while keeping progress within agreed limits and decision points.

drive — Command for Claude Code by jeet129: An autonomous work-loop command that keeps implementing the current…

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/jeet129/praxis/drive

Clone the repo

git clone --depth 1 https://github.com/jeet129/praxis

Made for: Claude Code.

Or install praxis, the plugin that ships this one along with the rest of its 105 skills, 12 commands, 17 agents, 6 hooks.

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.

README.md

[![agentmods](https://agentmods.dev/badge/commands/jeet129/praxis/drive.svg)](https://agentmods.dev/commands/jeet129/praxis/drive)

Your own site

<a href="https://agentmods.dev/commands/jeet129/praxis/drive"><img src="https://agentmods.dev/badge/commands/jeet129/praxis/drive.svg" alt="Measured on agentmods" height="20"></a>

Per session 35 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 912 The whole file, excluding the scripts and references it only reads on demand.

Security scan B 1 finding. 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.

Model	Per session	Once invoked
Fable 5	$0.00035	$0.00912
Opus 5	$0.00017	$0.00456
Sonnet 5	$0.00007	$0.00182
Haiku 4.5	$0.00003	$0.00091

Measured 5d ago against content hash 5f1a34745fdb, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

drive scanned grade B with 1 finding 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 5d 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- Marketplace/plugin install (Claude Code): `find ~/.claude/plugins -name praxis-drive.sh 2>/dev/null | head -1`

.claude/commands/drive.md · 43 lines

How it starts

The opening of the file, as written. The whole thing — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Start or resume autonomous drive.

Step 1 — Check the dial

Read governance/autonomy.yaml. Note stop_after (task | slice | phase | gate), run_budget, and stall.max_iterations_without_ledger_change. This is the OPTIONAL human boundary — the three non-negotiable stops (decision points, governance gates, budget/stall/exhaustion) fire regardless of this setting.

Step 2 — Locate or confirm the active ledger

Look for .project/working/slice-<id>-tasks.yaml for the current slice (per .project/working/active-workflow.md). If none exists, there's nothing to drive yet — route through /slice, which delegates to delivery-lead → lead-developer to produce the packet AND the task ledger (per references/loop-contracts.md §2) before drive can start. Do not fabricate a ledger yourself.

If a ledger exists, confirm it against the user: slice id, task count, how many are already done, and any existing stop_flags.

Step 3 — Explain the budget and stops once

Before iterating, tell the user (once, not every iteration):

the stop_after dial and what it means for this run
run_budget caps (max_slices_per_run, max_iterations_per_run, max_task_attempts, cost_ceiling_proxy)
the three non-negotiable stops that fire regardless of the dial
that slice-close summaries post to .project/telemetry/summaries/ even when the loop doesn't wait

Step 4 — Run it

Offer both paths; let the user pick (or infer from context — a long-running/background request implies unattended):

Unattended: instruct the user to run the runner script, which invokes the harness headlessly per governance/autonomy.yaml's harnesses block, enforcing iteration caps and stall detection outside the agent's own context. Locating it: the script ships inside the PLUGIN, not the project. Resolve the path first:
- Marketplace/plugin install (Claude Code): find ~/.claude/plugins -name praxis-drive.sh 2>/dev/null | head -1
- install.sh file install: ./scripts/praxis-drive.sh in the project
- Git clone used via --plugin-dir: <clone>/scripts/praxis-drive.sh Then run: bash <resolved-path> --project-dir . --dry-run first, and without --dry-run when satisfied. If the script is not found anywhere, the installed plugin predates drive mode — update the plugin. In-session drive (above) needs no script at all.
In-session: iterate the drive protocol yourself, continuously, within this session — each iteration is delivery-lead executing exactly one autonomous-drive pass (see agents/delivery-lead.md's Drive mode section) and reporting the outcome. After a pass returns, immediately begin the next one — do NOT report a completed task and wait for me. Nothing re-invokes delivery-lead in this path; you are the loop. Keep going through the slice drain (code review, security review, QA, closure) and into the next ledger, honoring the same stops as the unattended path. Stop cleanly ONLY when a non-negotiable stop or the dial's boundary is actually reached — a finished task under stop_after: gate is not one.

Read the full file on GitHub · 43 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.

5d ago First seen · 43 lines · 35 tokens per session scan B 5f1a34745fdb

Subscribe to this mod's changes

drive is a command published in the GitHub repository jeet129/praxis (7 stars, last pushed yesterday), licensed MIT. It adds 35 tokens to every session and 912 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Planning

drive

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade B, and why

Reads agent configuration directoriesmediumAgent snooping

How it starts

Step 1 — Check the dial

Step 2 — Locate or confirm the active ledger

Step 3 — Explain the budget and stops once

Step 4 — Run it

What this file has done since we first saw it

Other commands, from other repositories

checklist

clarify

specify

converge

implement

analyze