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.
npx skills add chan4lk/specclaw --skill loopgit clone --depth 1 https://github.com/chan4lk/specclawWrote 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.
[](https://agentmods.dev/skills/chan4lk/specclaw/loop)<a href="https://agentmods.dev/skills/chan4lk/specclaw/loop"><img src="https://agentmods.dev/badge/skills/chan4lk/specclaw/loop/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/chan4lk/specclaw/loop"><img src="https://agentmods.dev/badge/skills/chan4lk/specclaw/loop.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 3 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Tool Misuse · line 94 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
- high Tool Misuse · line 94 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
- high Tool Misuse · line 94 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
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 5.1 | $0.00015 | $0.02555 |
| Opus 5.5 | $0.00006 | $0.01022 |
| Sonnet 5.5 | $0.00003 | $0.00511 |
| Haiku 4.5 | $0.00002 | $0.00255 |
Grade A, and why
loop 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 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.
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.
How it starts
The opening of the file, as written. The whole thing — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.
specclaw loop
First, run specclaw-ensure-init .specclaw — idempotently creates .specclaw/ if it doesn't exist (silent if already initialized; auto-inits using the current directory's basename as the project name).
Autonomously drive the change to all-green. The controller (specclaw-loop) owns every mechanical decision — gate evaluation, signatures, caps, no-progress / regression / oscillation detection, the reward-hack guard, and state/log persistence. This skill only orchestrates the LLM steps: reflect, spawn a fix agent, commit. Follow the controller's JSON verbatim — do not second-guess a halt.
Step 0 — Validate prerequisites
The loop presumes /specclaw:plan and at least one /specclaw:build pass have already run — it remediates an existing implementation, it does not create one. Confirm the change is built:
specclaw-validate-change .specclaw <change> verify
If it fails (tasks not all complete / no build), tell the user to run /specclaw:build first and stop.
Read loop.enabled from .specclaw/config.yaml:
grep -A1 '^loop:' .specclaw/config.yaml
If loop.enabled is false, tell the user the autonomous loop is disabled and to run /specclaw:build and /specclaw:verify normally, then stop. Otherwise continue.
Step 1 — Init
specclaw-loop init .specclaw <change>
Seeds loop-state.json and loop-log.md (idempotent — never clobbers an existing state or log). Send a loop started notification:
🦞 **Loop Started**
**Change:** <change>
**Mode:** autonomous build→verify→review
**Caps:** max <loop.max_iterations> local iterations
Step 2 — Local loop
Repeat up to loop.max_iterations (default 5). The controller enforces the cap in Step 2c — do not track it yourself.
a. Evaluate gates:
specclaw-loop gates .specclaw <change>
Parse the JSON: {all_green, passing_count, gates:[{name,green,errors,files}]}. The four gates are tasks-complete, tests, verify, review.
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 Changed · -75 tokens per session 1b589a2ba7a3
- 29d ago Changed · +13 lines 02151029a1b0
- 1mo ago First seen · 168 lines · 90 tokens per session scan A a34fdf7fcdfd
loop is a skill published in the GitHub repository chan4lk/specclaw (13 stars, last pushed yesterday), licensed MIT. It adds 15 tokens to every session and 2,555 once invoked, about $0.0001 per session on Opus 5.5. 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.
Other skills, from other repositories
upload-assets
An asset-upload workflow that sends files to a console’s content-delivery network and prepares a review-request payload for a person to submit.
from-intake
A workflow for reviewing material brought from another workspace and carrying relevant brand evidence into the project's assets. It checks inventories, brand guides, logos, and image sources before anything is used.
review-spec
A workflow for comparing an existing feature specification with the project's review checklist and implementation. It looks for missing requirements, differences between documented screens and actual screens, and lessons from rejected submissions.
wire-sdk
An SDK integration guide for 앱인토스, a platform for building apps inside Toss. It covers permissions, persistent storage, sharing links, screen controls, safe-area spacing, and secure WebSocket connections.
optimize-seo-geo
A workflow for filling only the SEO and GEO sections of a post's front matter. SEO helps a page appear in search results, while GEO prepares information for answer systems that may quote or cite it.
loki-mode
An autonomous software factory that knows what it is supposed to deliver, and proves it did. Use when the user says Loki Mode or asks to build, fix or verify software autonomously.