Back to skills
cli
by reporails
CopilotโNot yet rated
Your rating
Brief
AI Instructions Diagnostics | Supported agents: Claude | Codex | Copilot | Cursor | Gemini
Description
# Reporails CLI (v0.6.8)
> **AI Instruction Diagnostics for coding agents. Validates the entire agentic instruction system against 120+ rules across six rule packs (core + per-agent). Supports Antigravity, Claude, Codex, Copilot, and Cursor.**
>
> *Beta phase - moving fast, feedback welcome.*
## X-ray
```bash
npx @reporails/cli xray
# or
uvx --from reporails-cli ails xray
```
`xray` opens a local page on `127.0.0.1` that shows what filled a Claude Code session's context window, over time. It is built from the transcripts Claude Code keeps under `~/.claude/projects/`, so it covers Claude Code sessions. Pass `--session <id>` to pick a session; the newest is the default.
[](https://www.youtube.com/watch?v=2PmPI4rZCgM)
The full 28-second demo is [on YouTube](https://www.youtube.com/watch?v=2PmPI4rZCgM).
## Quick Start
```bash
npx @reporails/cli check
# or
uvx --from reporails-cli ails check
```
No install, no account. The headline is a single **Quality** score (the analysis service's verdict on how well-formed your instructions are); fix the findings that move it, run again, watch it climb:
```
Reporails โ Diagnostics
โโ Main (1) 6 directive / 1 constraint ยท 30% prose
โ CLAUDE.md 6 dir / 1 con ยท 30% prose
โ โ Unresolved imports: docs/setup.md CORE:S:0024
โ โ Missing directory layout โ show the project structure with a treโฆ CORE:C:0035
โ ... and 19 more
โ 6 vague ยท 7 brief ยท 1 weak
โ โ 2 Pro diagnostics โ topic overlap, unbalanced topics
โ
โโ 35 findings
โโ Rules (1) 4 directive ยท 20% prose
โ testing 4 dir ยท 20% prose
โ โ No frontmatter block found CLAUDE:S:0012
โ โ Missing frontmatter block โ start with --- delimited YAML frontmโฆ CORE:S:0006
โ ... and 1 more
โ 4 vague ยท 4 brief
โ โ 1 Pro diagnostics โ topic overlap
โ
โโ 11 findings
โโ Cross-file โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ .claude/rules/testing.md โ CLAUDE.md โ 3 overlaps
โ .claude/rules/testing.md โ CLAUDE.md โ 1 repetition
Line-level detail โ sign in with ails login, then upgrade to Pro
โโ Summary โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Quality 1.0 / 10 โโโโโโโโโโโโโโโโโโโโ (1.0s)
Fix now 1 error. Start with CORE:S:0024.
Findings 46 total ยท -v to list every one
Agent: Claude
Level: L3 Scoped
Scope:
instructions: 10 directive / 4 prose (27%)
1 constraint
Main (1): โโโโโโโโโโโโโโโ 1.0 35 findings ยท 1 error
Rules (1): โโโโโโโโโโโโโโโ 1.0 11 findings
Top rules (by finding count):
CORE:E:0004 x11 warn Instruction Elaboration
CORE:C:0042 x10 warn Specificity Gap
CLAUDE:S:0012 x2 warn Path Scope Declared
CORE:C:0005 x1 warn Testing Framework Documented
+ 3 Pro diagnostics (3 warnings)
1 cross-file repetition
1 element pair overlaps in topic โ keep each topic in one file
Pro adds the remedies and the order to apply them.
โ sign in with ails login, then upgrade to Pro
```
## Install permanently
```bash
npx @reporails/cli install
# or
uvx --from reporails-cli ails install
```
Puts `ails` on your PATH, installs the reporails plugin into Claude Code and Codex when they are on your machine, and prints the install steps for the other supported agents (also listed in [Agent Support](https://github.com/reporails/cli/blob/main/docs/agent-support.md#plugin-support)). `ails install --project` installs the plugin for the current repository only, shared with collaborators through its Claude Code settings; Codex installs for your user. `ails update` upgrades `ails` and refreshes the plugin in each agent that has it. With Pro, the step after install is `/reporails:ails heal` in Claude Code, which rewrites your instruction files.
## Free vs Pro
Anonymous mode needs no account, and signing in is free. Signing in raises the hourly limit from 5 to 100 checks; anonymous and signed-in free accounts share the same payload cap and the same full diagnosis โ every finding, the score, and the fix text of the local checks in `-f json` and MCP โ and `ails check --heal` writes fixes on a Pro account only (an anonymous or free run gets the diagnosis and no changes). `ails check` also shows messages about your account (a failed payment, Pro ending) when there are any. Pro is the paid subscription: it raises the rate and payload caps and unlocks the remedies for the findings worth fixing, the exact line of each cross-file repetition and topic overlap, and the ordered remediation workflow your coding agent runs end to end.
```bash
# Sign in through your browser; one sign-in per machine
ails login
```
Full breakdown: [Tiers and Limits](https://github.com/reporails/cli/blob/main/docs/tiers.md).
## In CI
Run on every PR so instruction-quality regressions (vague or buried instructions, oversized files, weak reinforcement, instructions repeated across files) get caught the same way test or lint regressions do โ before merge, not after a teammate's agent has been silently misbehaving for a week.
```yaml
- uses: reporails/cli/action@0.6.8
with:
api-key: ${{ secrets.REPORAILS_API_KEY }} # optional - a Pro key unlocks the full diagnostic detail
strict: "true" # exit 1 if any rule fires
min-score: "7.0" # exit 1 if Quality < 7.0
```
Create an API key on [reporails.com/account](https://reporails.com/account) and store it as `REPORAILS_API_KEY` in your CI secret store. See [Configuration โ Authentication](https://github.com/reporails/cli/blob/main/docs/configuration.md#authentication).
The action keeps the analysis model (~275 MB) in the repository's Actions cache: the first run downloads it, and later runs restore it instead of downloading it again, even when the check fails. Running `ails` directly in a workflow? See [Configuration โ Caching the model in CI](https://github.com/reporails/cli/blob/main/docs/configuration.md#caching-the-model-in-ci).
## Documentation
- [Getting Started](https://github.com/reporails/cli/blob/main/docs/getting-started.md) - install, first run, what the output means
- [Agent Support](https://github.com/reporails/cli/blob/main/docs/agent-support.md) - which agents are recognized and what's covered
- [Tiers and Limits](https://github.com/reporails/cli/blob/main/docs/tiers.md) - Free vs Pro, what each tier includes
- [Configuration](https://github.com/reporails/cli/blob/main/docs/configuration.md) - disabling rules, project / global config, exclude paths
- [Score Guide](https://github.com/reporails/cli/blob/main/docs/score-guide.md) - how the score is built and what it tells you
- [Capability Levels](https://github.com/reporails/cli/blob/main/docs/capability-levels.md) - the L0-L7 ladder and what each level requires
- [Rules CLI](https://github.com/reporails/cli/blob/main/docs/rules-cli.md) - `ails rules list --capability=skill` and friends โ preflight rules before authoring
- [FAQ](https://github.com/reporails/cli/blob/main/docs/faq.md) - common questions
## Built and validated for
- **Antigravity** โ [Google](https://antigravity.google)
- **Claude** โ [Anthropic](https://github.com/anthropics)
- **Codex** โ [OpenAI](https://github.com/openai)
- **Copilot** โ [GitHub](https://github.com/github)
- **Cursor** โ [Anysphere](https://github.com/cursor)
## License
[BUSL 1.1](https://github.com/reporails/cli/blob/main/LICENSE) - converts to Apache 2.0 three years after each release.
The analysis model files are licensed separately under the [Reporails Model Licence](https://github.com/reporails/cli/blob/main/LICENSE-weights).
Installation
Install the skill from https://github.com/reporails/cliThis skill is not published on the ClawHub registry. Paste the prompt into your coding agent โ Claude Code, Cursor, Codex โ and it will install the skill straight from the source repository.
Open the repository