New: 97,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Loop Worker

by Sidiora-Labs in Sidiora-Labs/centra-gideon-agent

Per-cycle protocol for an autonomous goal loop worker — the unified autonomous goal engine that runs one self-directed cycle per turn until its goal is met, then self-retires.

Apache-2.0Auto-check: warnings

Install Loop Worker

The automated check flagged lines worth reading first. See the safety section below.

skills CLI
$ npx skills add Sidiora-Labs/centra-gideon-agent --skill loop-worker -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install Sidiora-Labs/centra-gideon-agent loop-worker --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/Sidiora-Labs/centra-gideon-agent.git skills-src && mkdir -p .claude/skills && cp -r skills-src/runtime/gideon/extensions/skills/bundled/loop-worker .claude/skills/loop-worker && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
loop-worker
GitHub stars
130
Token cost
~2k tokens
SKILL.md length
987 words
Files
1
Skills in repo
17
Repo updated
First seen
Licence
Apache-2.0

At a glance

Per-cycle protocol for an autonomous goal loop worker — the unified autonomous goal engine that runs one self-directed cycle per turn until its goal is met, then self-retires.

  • Works in 7 steps: Gate + STOP. Read status.json. If status… → Brief + guidance. Read brief.md (goal… → Orient from compact signals. Skim the… → …
  • SKILL.md covers The loop directory (your file…, Goal type drives the deliverable, Per-cycle protocol (strict… and Finding schema, plus 4 more sections
  • Calls git

What it does

Loop Worker is an agent skill from Sidiora-Labs/centra-gideon-agent. Per-cycle protocol for an autonomous goal loop worker — the unified autonomous goal engine that runs one self-directed cycle per turn until its goal is met, then self-retires. Drives one cycle against a loop's file interface (status.json gate, brief.md goal, guidance.txt nudges, findings/cycleNNN.json, FINDINGS.md, the type-appropriate deliverable, STOP sentinel). The worker produces work and reports evidence; a deterministic check or a separate judge — never the worker — decides done-ness.

Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: The companion AI agent that learns, adapts and gets the work done no matter the task. The licence is Apache-2.0.

Example prompts

  • “/loop-worker”

Workflow steps

7 steps, taken from the first numbered list in SKILL.md.

  1. Gate + STOP. Read status.json. If status is not running, **stop and
  2. Brief + guidance. Read brief.md (goal type, goal, sub-goals, scope,
  3. Orient from compact signals. Skim the one-line summary/key_insight of
  4. One atomic step. Pursue the single highest-value open lead toward the
  5. Record the finding. Write findings/cycle_NNN.json (next sequential N)
  6. Update the deliverable (only if the goal type has one — see above). Fold
  7. End the turn. The next cycle fires automatically after the idle interval.

What it can do on your machine

Read from SKILL.md and the folder at commit 56a2de3. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • git

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Loop Worker loads about 2k tokens when it runs. Until then it costs about 127 tokens; SKILL.md has 987 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~127
When it runs · the whole SKILL.md, loaded when a task matches
~2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: warnings

The automated check found patterns that need a careful read before installing.

  • WarningMentions a credentials file (SSH keys, cloud or package-manager tokens)SKILL.md:143
    redential files as text** (`~/.aws/*`, `~/.ssh/*`, `.env`,
  • WarningMentions a credentials file (SSH keys, cloud or package-manager tokens)SKILL.md:144
    `~/.netrc`, cookie/auth files); never echo secrets into a finding. If an

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from Sidiora-Labs/centra-gideon-agent at commit 56a2de3, republished under its Apache-2.0 licence (© Sidiora-Labs). 987 words, ~2,037 tokens.

Download SKILL.mdSave it as .claude/skills/loop-worker/SKILL.md (or your agent's skills folder).
name
loop-worker
description
Per-cycle protocol for an autonomous goal loop worker — the unified autonomous goal engine that runs one self-directed cycle per turn until its goal is met, then self-retires. Drives one cycle against a loop's file interface (status.json gate, brief.md goal, guidance.txt nudges, findings/cycle_NNN.json, FINDINGS.md, the type-appropriate deliverable, STOP sentinel). The worker produces work and reports evidence; a deterministic check or a separate judge — never the worker — decides done-ness.
triggers
goal loop, autonomous, autonomy loop, self-nudge, goal-driven loop, run until done, keep going, continuous improvement, north star, definition of done, next…
tags
skill, gideon, autonomous, loop

Autonomous Goal Loop Worker

You are the worker for an autonomous goal loop: a goal-driven session that runs one self-directed cycle per turn until its goal is met, then retires itself. A supervisor (the loop watchdog) arms you each cycle via a nudge and decides lifecycle deterministically — completion, stagnation, stalls, trust expiry. You decide direction: the single highest-value next step toward the goal.

You produce work and report evidence. You do not certify whether the goal is done — a deterministic check the supervisor runs, or a separate judge subagent that never touched your work, decides that. Never write a "passed" / "done" self-verdict.

The nudge is only a trigger. This protocol is the work.

The loop directory (your file interface)

Everything you read and write lives in the loop dir named in the nudge:

FileWho writesMeaning
status.jsonsupervisorThe cycle gate. {status, loop_id, ts}.
brief.mdsupervisorGoal type, goal, sub-goals, scope, attendedness, definition of done / verification check. Written once at launch.
guidance.txtuser (via nudge)A mid-flight steer. Present only when the user sent one.
questions.jsonyou (attended only)One clarification question, when the brief permits it.
findings/cycle_NNN.jsonyouOne structured finding per cycle.
FINDINGS.mdyouYour working log — cumulative findings + a ## State handoff note.
verdicts/cycle_NNN.jsonthe judgeThe third-party verdict. You never read or write this.
the deliverableyouThe document the goal asks for — see goal type below.
STOPsupervisor/userSentinel — if present, the loop is ending. Halt immediately.

Goal type drives the deliverable

brief.md states the goal type. It decides what you produce:

  • open_ended (research, analysis, "write N documents"): maintain REPORT.md — the polished, well-structured document the goal asks for, created cycle 1 and integrated every cycle.
  • verifiable (get CI green, migrate a pattern, hit 0 lint warnings): there is no document deliverable. The code / the passing check is the output. Make real progress toward the check; the supervisor runs the check itself each cycle.
  • monitor (watch a queue, triage new incidents): maintain MONITOR_LOG.md — a running log of what you saw and what you acted on. Never self-completes.

FINDINGS.md is always your log; the deliverable (when there is one) is the output.

Per-cycle protocol (strict order)

  1. Gate + STOP. Read status.json. If status is not running, stop and end the turn now. Likewise if the STOP sentinel exists: halt immediately.
  2. Brief + guidance. Read brief.md (goal type, goal, sub-goals, scope, attendedness, DoD/check). If guidance.txt exists, incorporate it and delete it (it is consumed).
  3. Orient from compact signals. Skim the one-line summary/key_insight of the most recent findings/cycle_*.json and the ## State section of FINDINGS.md. Do not re-read every prior finding — work from the summaries so context stays lean.
  4. One atomic step. Pursue the single highest-value open lead toward the goal: an unanswered sub-goal, a follow-up a prior finding surfaced, or shoring up weak evidence. One step — not the whole goal. Keep it to a small handful of tool calls. For monitor goals, the step is: poll the source, act on anything new, and record it.
  5. Record the finding. Write findings/cycle_NNN.json (next sequential N) and append a concise entry to FINDINGS.md, keeping a short ## State section current so the next cycle can orient cheaply. Report what you DID and the EVIDENCE — never a done/passed verdict.
  6. Update the deliverable (only if the goal type has one — see above). Fold this cycle's new findings into it, integrate rather than append, and keep it coherent.
  7. End the turn. The next cycle fires automatically after the idle interval.
Show full SKILL.md (408 more words)Show less

Finding schema

json
{
  "cycle": 7,
  "summary": "one line — what this cycle established",
  "key_insight": "the single most useful takeaway",
  "sources_checked": ["url or path", "..."],
  "sources_empty": ["searched but found nothing — useful to record"],
  "new_findings_count": 3,
  "evidence": "what you produced + the evidence behind it (the judge reads this)",
  "metric": {"name": "failing_tests", "value": 2}
}
  • new_findings_count is how the supervisor detects stagnation (several cycles of zero ⇒ it pauses you for direction). Be honest: a cycle that rediscovers known facts is 0.
  • evidence is what the judge subagent reads to assess marginal value and done-ness — make it substantive, not a teaser.
  • metric (optional) is for verifiable/measurable goals — the current value of the thing being driven (failing tests, lint warnings, the target metric). The supervisor trends it.
  • Never write a passed/done/verification field. Done-ness is decided off-worker.

Attendedness

brief.md says whether the loop is attended or unattended.

  • Attended: if the goal/scope is genuinely ambiguous in a way that would change your direction, you MAY write one {"question": ..., "why": ...} to questions.json and end the turn — the loop pauses and the user answers via a nudge (arriving as guidance.txt). Keep the bar high.
  • Unattended: never write questions.json. If a question arises, INVESTIGATE it yourself — research it in context, pick the best-reasoned answer, record the assumption in your finding, and proceed.

Self-retiring

You may stop the loop with loop_nudge_stop (with a brief reason) only in these cases — done-ness is otherwise the supervisor's call:

  • The STOP sentinel tripped or the status.json gate closed.
  • An unrecoverable infrastructure failure the host can't route around (disk full, network partition, auth provider down for multiple cycles) — log a one-line diagnosis to the finding, then loop_nudge_stop(reason="infra: <what>").

For monitor goals, never self-retire on "nothing new" — a quiet cycle is a valid no-op finding, not a reason to stop.

Staying silent

The chat panel is not the progress channel — the findings, FINDINGS.md, and the deliverable are. Stay silent in chat unless a hard blocker genuinely needs a user decision, or the STOP sentinel tripped / the status.json gate closed. If you hit a blocker, surface it once — don't re-post it every cycle.

Operating invariants

  • Never git push. Humans push.
  • Never run destructive operations (no rm -rf, no force-push, no dropping data).
  • Never read credential files as text (~/.aws/*, ~/.ssh/*, .env, ~/.netrc, cookie/auth files); never echo secrets into a finding. If an auth-path call raises, scrub the exception to its type name only.
  • One cycle = one atomic step. Compounding small steps is the whole design.
  • Keep going. Test failures, dead ends, and "I don't know how yet" are the job, not reasons to stop — find another angle and tick forward. The only stop conditions are the gate, the STOP sentinel, and unrecoverable infra failure. Everything else is a problem to solve, not a reason to halt.

© Sidiora-Labs, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in runtime/gideon/extensions/skills/bundled/loop-worker of Sidiora-Labs/centra-gideon-agent.

Open the folder on GitHubat commit 56a2de3

Compare with similar skills

Loop Worker next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Loop Worker compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Loop Worker this skillSidiora-Labs/centra-gideon-agent130—~2kAutomated safety check: WarnApache-2.0
Goalscodewhale-hq/Codewhale41k—~273Automated safety check: PassMIT
Worker Visualizernexu-io/open-design100k—~998Automated safety check: PassApache-2.0
Agent Worker Specialistruvnet/ruflo74k3 repos~1.4kAutomated safety check: PassMIT
Agent Goal Plannerruvnet/ruflo74k3 repos~842Automated safety check: PassMIT
Loop Workerruvnet/ruflo74k—~297Automated safety check: PassMIT

Similar skills

  • Goals

    codewhale-hq/Codewhale

    Set, review, and update the user's goals. An agent skill from codewhale-hq/Codewhale.

    41k GitHub stars~273 tokensUpdated today
    Auto-check passed

  • Worker Visualizer

    nexu-io/open-design

    A real-time data/particle/simulation visualizer whose heavy compute runs in a Web Worker (off the main thread), optionally sharing memory with the UI via SharedArrayBuffer, and renders to a canvas…

    100k GitHub stars~998 tokensUpdated yesterday
    Auto-check passed

  • Agent Worker Specialist

    ruvnet/ruflo

    Agent skill for worker-specialist - invoke with $agent-worker-specialist

    74k GitHub starsUsed in 3 repos~1.4k tokens
    Auto-check passed

  • Agent Goal Planner

    ruvnet/ruflo

    Agent skill for goal-planner - invoke with $agent-goal-planner

    74k GitHub starsUsed in 3 repos~842 tokens
    Auto-check passed

  • Loop Worker

    ruvnet/ruflo

    Run Ruflo background workers using Claude Code native /loop scheduling

    74k GitHub stars~297 tokensUpdated today
    Productivity & AutomationAuto-check passed

  • Goal Plan

    ruvnet/ruflo

    Create and execute Goal-Oriented Action Plans (GOAP) with precondition analysis, cost optimization, and adaptive replanning

    74k GitHub stars~807 tokensUpdated today
    Auto-check: notes

More from Sidiora-Labs/centra-gideon-agent

All 17 skills in this repo

  • Artifacts

    Sidiora-Labs/centra-gideon-agent

    Persist, version, and iterate on LLM-generated UI (widgets, HTML, markdown).

    130 GitHub stars~1.2k tokensUpdated today
    Auto-check passed

  • Best Of N

    Sidiora-Labs/centra-gideon-agent

    Sample several candidate answers in parallel, judge them against stated criteria, and present the winner with the runners-up kept one click away.

    130 GitHub stars~1.2k tokensUpdated today
    Auto-check passed

  • Document Authoring

    Sidiora-Labs/centra-gideon-agent

    Author a real document — a report, proposal, slide deck, spec, or paper — by settling audience, claim, and structure before drafting prose, then writing to that structure.

    130 GitHub stars~970 tokensUpdated today
    Auto-check passed

  • Editorial Document

    Sidiora-Labs/centra-gideon-agent

    Write long-form editorial documents as clean, semantic HTML saved as an artifact (kind=document) — reports, briefs, explainers, proposals, write-ups.

    130 GitHub stars~1.3k tokensUpdated today
    Auto-check passed

  • Gideon API

    Sidiora-Labs/centra-gideon-agent

    The operator's manual for DRIVING Gideon's API and tools — how to find an exact tool/route signature, the mandatory verify-after-mutate loop, and what NOT to hand-roll.

    130 GitHub stars~1.4k tokensUpdated today
    Auto-check passed

  • Infographic Syntax

    Sidiora-Labs/centra-gideon-agent

    Author infographics with the AntV declarative DSL — pick a template, fill a small indented data tree, and save as an artifact (kind=infographic) that renders to crisp SVG and streams as you write it.

    130 GitHub stars~1.1k tokensUpdated today
    Auto-check passed

Questions about Loop Worker

What does Loop Worker do?

Per-cycle protocol for an autonomous goal loop worker — the unified autonomous goal engine that runs one self-directed cycle per turn until its goal is met, then self-retires. Loop Worker is an agent skill from Sidiora-Labs/centra-gideon-agent. Per-cycle protocol for an autonomous goal loop worker — the unified autonomous goal engine that runs one self-directed cycle per turn until its goal is met, then self-retires.

How do I install Loop Worker in Claude Code?

Run `npx skills add Sidiora-Labs/centra-gideon-agent --skill loop-worker -a claude-code`. Or copy the skill folder (runtime/gideon/extensions/skills/bundled/loop-worker in Sidiora-Labs/centra-gideon-agent) into .claude/skills/loop-worker in your project. Claude Code loads it when a task matches its description.

How do I install Loop Worker in Codex?

Run `npx skills add Sidiora-Labs/centra-gideon-agent --skill loop-worker -a codex`. Or copy the skill folder (runtime/gideon/extensions/skills/bundled/loop-worker in Sidiora-Labs/centra-gideon-agent) into .agents/skills/loop-worker in your project. Codex loads it when a task matches its description.

Can I use Loop Worker in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add Sidiora-Labs/centra-gideon-agent --skill loop-worker -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/loop-worker, .gemini/skills/loop-worker, .github/skills/loop-worker and .opencode/skills/loop-worker in your project.

What does Loop Worker need to run?

Going by SKILL.md and its folder, Loop Worker needs the command-line tools its instructions call (git).

Does Loop Worker access the network?

SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Loop Worker safe to install?

Our automated static check of SKILL.md flagged 2 warning(s): mentions a credentials file (ssh keys, cloud or package-manager tokens). Read the flagged lines before installing; the check is not a guarantee either way.

What licence does Loop Worker use?

Loop Worker is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Loop Worker use?

About 2k tokens (SKILL.md is roughly 8.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Loop Worker?

Skills that share tags, products or a category with Loop Worker: Goals (codewhale-hq/Codewhale, 41k stars), Worker Visualizer (nexu-io/open-design, 100k stars), Agent Worker Specialist (ruvnet/ruflo, 74k stars) and Agent Goal Planner (ruvnet/ruflo, 74k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Loop Worker?

Sidiora-Labs (a GitHub organization) maintains it in Sidiora-Labs/centra-gideon-agent, which has 130 GitHub stars. The repository holds 17 skills in this directory. The repository was last updated on October 7, 2026.

Source: Sidiora-Labs/centra-gideon-agent on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.