- Home
- Skills
- Workflow Automation
- Spec Lint: check a task before another agent runs it
Works with the AI Agents you already use
Spec Lint: check a task before another agent runs it
Preflight check for agent task specs: finds the missing guardrails a worker agent needs and gives you paste-ready lines.
$5
Spec Lint: check a task before another agent runs it
Example session with this skill installed
lint this spec before I dispatch it: tasks/fix-dates.md
- Read your context and instructions
- Compiled the spec lint
Spec Lint: fix-dates.md → Codex in ~/work/billing
Task class: implement. Rules checked: 3 from CLAUDE.md.
BLOCKER Secrets (G1) Missing. Your rule (CLAUDE.md): never create, read, or delete env files.
BLOCKER Commit (G2, conflict) Spec says "commit when done"; your rule: never commit without approval.
WARN Bounded loops (G14) Weak. "Keep iterating until it passes" has no attempt limit.
OK: destructive git, file scope, stop and ask.
Paste-ready fix
Must NOT do
- Do not push, merge, tag, open a PR, or publish anything. Leave changes uncommitted for review. (replaces: "commit when done")
- Make at most 2 attempts. If it still fails, stop and report what you tried. (replaces: "keep iterating until it passes")
Add these lines to the spec?
Connects securely to your tools. The creator never sees your data.
About this skill
Before you hand a task to Codex, a Claude subagent, or any worker agent, Spec Lint checks the spec for missing guardrails (secrets, commits, destructive git, test runs, live systems, cost caps, file scope, stop conditions) and gives you the exact lines to paste in.
A worker agent starts cold. It doesn't see your CLAUDE.md or the corrections you made last week. Whatever the spec leaves out, it decides alone: it opens your env file to debug, commits to your branch, restarts a prod pod, or runs a paid eval with no limit.
What you get
- A short report: BLOCKER / WARN / NOTE, one plain-language line per gap.
- Conflicts with your own rules called out, with your rule winning every time.
- Paste-ready guardrail lines, marked where they replace a spec line.
- Your rules override its defaults: set your test, commit and branch policy once in CLAUDE.md or AGENTS.md.
- Never edits the spec without your yes, never dispatches, never opens secret files, and never suggests letting a worker write to a live system.
Optional: Brain (free, separate: pip install agent-memory-brain). If you use it, Spec Lint also checks the spec against the rules you've approved there. Not required.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.
It checks the spec text; it can't make a worker obey it. It reads action rules (never / always / ask first) from your instruction files, not every preference. It doesn't rewrite your task's goal, except that writes to live systems become proposals you confirm.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 5 days ago
- Software
Git
Permissions
File Scopes
Never opens env, key or credential files. Optionally calls Brain's rules_list if the Brain MCP is installed.