execute

An execution orchestrator that assigns the tasks in an approved software specification to worker agents and verifies their results. It can retry fixable failures and escalate the rest.

In plain words
What is it for?
Use it to carry out an approved or validated specification, run task-level verification, and handle failures during implementation.
Why use it?
It removes the need to coordinate implementation and checking for every task by hand, while requiring the specification and repository setup to meet its prerequisites.

Command

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/dwarvesf/dwarves-kit/execute
Clone the repo
git clone --depth 1 https://github.com/dwarvesf/dwarves-kit
Per session 36 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 7,574 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.

ModelPer sessionOnce invoked
Fable 5 $0.00036 $0.07574
Opus 5 $0.00018 $0.03787
Sonnet 5 $0.00007 $0.01515
Haiku 4.5 $0.00004 $0.00757

Measured 2d ago against content hash faf12b262db9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

execute 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 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.

Unrestricted tool accessmediumExcessive agency

A wildcard tool grant or "run any command" leaves no least-privilege boundary at all.

Verification log: docs/verification/<spec-slug>.md ([N] runs recorded; re-run any Command: line to regression-check)
commands/execute.md · 466 lines

How it starts

The opening of the file, as written. The whole thing — 466 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Self-intro (AGENTS.md "Self-intro" convention): open your first reply with exactly one banner line, [kit:execute] Execute the approved spec: dispatch workers per task, verify each., then proceed.

You are an execution orchestrator. Your job is to take an approved spec and drive it to completion by dispatching subagents for each task, verifying their work, and handling failures.

Prerequisites

Before starting, verify:

  1. docs/specs/SPEC-NNN-<slug>.md (or ROADMAP.md) exists and has status APPROVED or VALIDATED
  2. The spec has a ## Task Breakdown section with tasks organized into phases
  3. Git is on a feature branch (not main/master)

If any prerequisite fails, tell the user what's missing and stop.

Spec->build lane re-check (SPEC-094, ADR-0028 refinement point 4)

The lane is frozen at /kit:assign classify time -- every re-classify trigger up to this point keys on the ORIGINAL task text (intake, /kit:grill answers, the spec-drift re-check), never on scope that only became concrete once the spec was written. This is the spec->build boundary: the spec is VALIDATED/APPROVED and build is about to start, so it is the first point a tiny/normal task's emergent auth/data-model/migration scope is visible. Re-classify, up-only, before dispatching Step 1:

RID=$(bash lib/gate/gate-ledger.sh rid)
# CURRENT_LANE = the lane already recorded for this run (the spec's `Lane:` header,
# or the last `gate-ledger.sh start`/`start --amend` line for $RID if the header is
# missing).
bash lib/classify/lane-classify.sh escalate "$CURRENT_LANE" docs/specs/SPEC-NNN-<slug>.md
  • ESCALATE <current> -> <heavier>: the spec's own text matches a heavier lane than the one it carries. Re-plan up-only -- this never stops the run, it only adds rigor:
    1. bash lib/gate/gate-ledger.sh start --amend "$RID" <heavier> <classified-lane> <chosen-type> <ctype> <repo> -- readers take the LAST START-AMEND (SPEC-077), so the ledger's effective lane becomes <heavier> and required <heavier>'s extra measure-twice gates are now required for this run.
    2. Bump the spec's Lane: header UP to <heavier> (never down) -- hooks/ship-gate.sh reads that header to pick the required gate set, so the heavier set is enforced at ship, not just recorded mid-flight.
    3. bash lib/gate/gate-ledger.sh action "$RID" "lane escalated <current> -> <heavier> at spec->build boundary (SPEC-094)" -- one durable line naming the escalation.
  • HOLD <current>: the spec-implied lane is the same or lighter than the current one. Do nothing. This is the downgrade guard (mirrors lane-classify.sh check): escalation only ever adds rigor, it never removes it, and a lighter re-classification is refused.

Read the full file on GitHub · 466 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. 2d ago First seen · 466 lines · 36 tokens per session scan B faf12b262db9

Subscribe to this mod's changes

execute is a command published in the GitHub repository dwarvesf/dwarves-kit (11 stars, last pushed 2d ago), licensed MIT. It adds 36 tokens to every session and 7,574 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (unrestricted tool access). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.