run

An autonomous development loop that works through ready features by dispatching developer and reviewer roles, applying changes, and running quality checks. It stops when it needs a person or encounters a defined failure.

In plain words
What is it for?
Use it when explicitly asking Cladding to make autonomous progress on planned features, with options for iteration limits, time limits, retries, project directory, and JSON output.
Why use it?
It automates repeated feature implementation and checking while placing limits around retries, time, budget, and human approval.

Skill for Claude CodeCodex

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 skills/qwerfunch/cladding/run
Any agent
npx skills add qwerfunch/cladding --skill run
Clone the repo
git clone --depth 1 https://github.com/qwerfunch/cladding

Made for: Claude Code, Codex.

Per session 78 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 736 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 $0.00078 $0.00736
Opus 5 $0.00039 $0.00368
Sonnet 5 $0.00016 $0.00147
Haiku 4.5 $0.00008 $0.00074

Measured 2d ago against content hash 9f95ff17d70c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

run 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 2d 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.

plugins/antigravity/skills/run/SKILL.md · 35 lines

What it actually says

Cladding run (formerly drive)

Run clad run from the project root. The autonomous loop:

  1. Pre-flight adapter.healthCheck() — fails fast on missing credentials or unreachable host.
  2. For each ready feature (status planned, depends_on satisfied):
    • Specialist dispatch authors the implementation.
    • Apply mutations to the working tree.
    • L1 gates: Type / Lint / Arch.
    • Reviewer dispatch — HUMAN_REQUIRED halt if reviewer identity equals specialist (anti-self-cert barrier).
    • UAT requires a human-pass evidence entry; missing → HUMAN_REQUIRED halt.
  3. Halt class is one of the 13 enumerated reasons (ALL_FEATURES_DONE, MAX_ITERATIONS, WALL_CLOCK, BUDGET_EXCEEDED, BLOCKED_FEATURE, RETRY_THRESHOLD, GATE_NO_PROGRESS, HUMAN_REQUIRED, TRANSPORT_AUTH_FAILED, TRANSPORT_RATE_LIMITED, TRANSPORT_NETWORK, LLM_UNAVAILABLE, UNCAUGHT_ERROR).

Budget flags: --max-iterations, --max-wall-clock-ms, --max-retries. --cwd <path> targets a project directory other than the current one. --json emits the raw Iron Core result; default is the plain Soft Shell summary.

clad run
clad run --cwd /path/to/project
clad run --max-iterations 10
clad run --json

Heads-up — run needs a real LLM, and is for unattended/headless use only. The host AI (Claude Code, Cursor, …) drives work naturally in-session; clad run is the entry point for the opposite case — autonomous, no-human-in-the-loop progress (CI/cron/SDK). Two requirements:

  • A real dispatch must be available: either run inside clad serve (MCP sampling) or use SDK mode (agent.mode = sdk + an API key). With neither, the loop falls back to the Mock transport and produces empty module stubs, not real implementations — yet still reports a normal halt. Treat a standalone clad run with no MCP server and no SDK key as not doing real work; verify with clad doctor afterward (a deterministic/Mock run is a red flag, not success).
  • run modifies the working tree.

Known gap (tracked): standalone run on the Mock fallback should hard-fail with LLM_UNAVAILABLE rather than silently stubbing. That change reconciles the adapter healthCheck parity contract (F-049 AC-089, which currently treats the Mock fallback as "ready") and is a deliberate follow-up, not yet shipped.

After a run session, run clad doctor over the same --cwd to confirm the LLM dispatcher behaved — any sentinel_miss events surface as a health summary so you can tell whether the loop ran with full LLM refinement or fell back to deterministic per-artifact.

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. 2d ago First seen · 35 lines · 78 tokens per session scan A 9f95ff17d70c

Subscribe to this mod's changes

run is a skill published in the GitHub repository qwerfunch/cladding (14 stars, last pushed 4d ago), licensed MIT. It adds 78 tokens to every session and 736 once invoked, about $0.0004 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.

Related

Other skills, from other repositories

map-plan

ARCHITECT phase - decompose complex tasks into atomic subtasks with research, spec, and branch-scoped plan artifacts under .map.

azalio/map-framework · 31 tokens

map-review

Interactive 4-section code review using monitor, predictor, and evaluator agents plus the user and maintainer role reviewers on current changes. Use when reviewing a diff, PR, or staged work before merge. Do NOT use to plan or implement; use map-plan or map-efficient.

azalio/map-framework · 58 tokens

map-debug

Structured MAP debugging via task-decomposer, actor, and monitor agents. Use when reproducing a bug, isolating a regression, or diagnosing an error with specialized agents — including failing or flaky tests (pytest AssertionError), crashes and segmentation faults, memory-corruption or memory errors in native/C…

azalio/map-framework · 213 tokens

map-learn

Capture reusable lessons after a completed MAP workflow. Use when a MAP run has finished and you want rules written to .claude/rules/learned/ from a workflow summary or handoff. Do NOT use during active implementation.

azalio/map-framework · 51 tokens

map-efficient

State-machine MAP execution workflow for Codex. Use when implementing an approved MAP plan end to end, resuming from branch MAP taskplan or stepstate.json artifacts, or running non-trivial multi-subtask work. Use map-fast for tiny one-shot edits.

azalio/map-framework · 55 tokens

map-task

Execute a single subtask from an existing MAP plan via Actor and Monitor. Use when map-plan has decomposed work and you want fine-grained control over one subtask. Do NOT use without an existing plan; run map-plan first.

azalio/map-framework · 51 tokens