roundloop

roundloop is a skill for Claude Code, Codex from SoliEstre/EstreGenesis. It costs 114 tokens per session (1,242 once invoked), scanned A, original, Apache-2.0.

An operating procedure that repeats defined work rounds until the remaining tasks require a human decision. A round means checking the work queue, doing available work, and measuring the resulting board update.

In plain words
What is it for?
Running repeated queue-processing rounds, tracking in-flight and completed work, stopping after declared limits or dry rounds, and handing back unresolved decisions.
Why use it?
It prevents an automated loop from creating more tasks for itself or claiming progress merely because work was registered.

Skill for Claude CodeCodex

Part of the constellation plugin — 10 skills, 1 hook 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 skills/soliestre/estregenesis/roundloop
Any agent
npx skills add SoliEstre/EstreGenesis --skill roundloop
Clone the repo
git clone --depth 1 https://github.com/SoliEstre/EstreGenesis

Made for: Claude Code, Codex.

Or install constellation, the plugin that ships this one along with the rest of its 10 skills, 1 hook.

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.

agentmods badge for roundloop

README.md
[![agentmods](https://agentmods.dev/badge/skills/soliestre/estregenesis/roundloop.svg)](https://agentmods.dev/skills/soliestre/estregenesis/roundloop)
Your own site
<a href="https://agentmods.dev/skills/soliestre/estregenesis/roundloop"><img src="https://agentmods.dev/badge/skills/soliestre/estregenesis/roundloop.svg" alt="Measured on agentmods" height="20"></a>
Per session 114 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,242 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 $0.00114 $0.01242
Opus 5 $0.00057 $0.00621
Sonnet 5 $0.00023 $0.00248
Haiku 4.5 $0.00011 $0.00124

Measured 3d ago against content hash 10f04d971372, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

roundloop 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.

plugins/constellation/skills/roundloop/SKILL.md · 50 lines

How it starts

The opening of the file, as written. The whole thing — 50 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/roundloop — repeat rounds until the remainder is not self-actionable

Repeats /roundnext until the queue can no longer be advanced by this agent alone. Contract: Constellation.md §13.34. The product of a run is the list of things only a human can decide — not the round count, and not a longer backlog.

Read §13.34 before changing anything here. In particular: registering work is not doing work (§13.34.1), and a round is complete when its board write is measured, not when it is sent (§13.34.3).

0. Preflight — declare the envelope, then read the board

State the caps before the first round, in the opening line of the run: max rounds, the dry-round threshold K (default 2), and any wall-clock or token budget. A cap discovered mid-run is not a cap (§13.34.2-3). If the operator gave a bound ("3 rounds", "until the P2s are gone"), that bound replaces the default and is echoed back.

Then load board truth once (board_state_get, or the board's state.json): planned[] with blocked/blockReason, current[], open decisions[], recent done[].

Register the loop as in-flight work (§13.34.5) so the operator can see it exists and stop it: one current[] entry, id c-roundloop, whose detail line carries the round index and the dry count. Remove it when the loop stops. A loop nobody can see is a loop nobody can interrupt.

1. Loop body — per round

  1. Run one round exactly as /roundnext specifies (all five steps, in order). Do not abbreviate the round because it is inside a loop; the blocker re-examination and the new-work scan are what keep later rounds honest.
  2. Barrier (§13.34.3): wait for the board write to land, then measure it — worker report received, or a state condition confirmed by reading. Never compute the next round's ripeness from pre-write state. This is the difference between a loop and a race.
  3. Score the round against the productivity predicate (§13.34.1):
    • productive ⟸ work was started, or a blocker was cleared with evidence.
    • not productive ⟸ everything else, including a round that registered new planned items, re-sorted the queue, refreshed a blockReason, or wrote a gate. Those are useful; they are not progress. Keep a running ledger: round index, productive?, what made it so, dry-count.
  4. Check the guards in order and stop at the first that fires — dry rounds ≥ K · remainder not self-actionable · any cap reached · stall (same item started twice with no done[] delta) · context pressure · operator interrupt. Guards are checked at the round boundary, never mid-round, so the board is never left half-written.

Read the full file on GitHub · 50 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 · 50 lines · 114 tokens per session scan A 10f04d971372

Subscribe to this mod's changes

roundloop is a skill published in the GitHub repository SoliEstre/EstreGenesis (8 stars, last pushed 10d ago), licensed Apache-2.0. It adds 114 tokens to every session and 1,242 once invoked, about $0.0006 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 2026-08-31.

Related

Other skills, from other repositories

update-agent-context

This skill should be used to keep CLAUDE.md, AGENTS.md, and the skill files themselves compact, current, and internally consistent. It runs in three phases: Phase 1 performs a one-time structural refactor of CLAUDE.md using a Karpathy-inspired behavioral scaffold and derives AGENTS.md from it by stripping Claude…

mostlyharmless-ai/watercooler · 203 tokens

watercooler-onboarding

Bootstrap Watercooler memory for a repository by inspecting local code, docs, CI, git history, and existing Watercooler threads, then writing a small set of durable, provenance-backed seed threads that future agents can query and extend. Use when entering a repo for the first time, seeding a repo with Watercooler…

mostlyharmless-ai/watercooler · 79 tokens

ppgp

Portable Persistent Goal Protocol for long-running coding-agent work. Use when starting, resuming, handing off, distilling, or closing a substantial software goal across long sessions, context compaction, agent replacement, or other Agent Skills-compatible coding-agent environments.

Fatboy-coder/ppgp · 53 tokens

search-threads

Search threads with filters. Supports filters like role:planner, type:Decision, after:2024-01, thread:topic-name, status:OPEN.

mostlyharmless-ai/watercooler · 35 tokens

watercooler-health

Check watercooler system health — MCP server, baseline graph (T1), git auth, GitHub rate limit, and daemons. Use when syncs break or anything in the watercooler stack behaves unexpectedly.

mostlyharmless-ai/watercooler · 49 tokens

recall

Recall project context or answer questions about history and decisions. Use before starting work, when investigating unfamiliar code, or asking "What was decided about X?" / "Why did we choose Y?".

mostlyharmless-ai/watercooler · 41 tokens