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.
git clone --depth 1 https://github.com/emaraschio/cursor-commandsWrote 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/commands/emaraschio/cursor-commands/blind-spot-pass)<a href="https://agentmods.dev/commands/emaraschio/cursor-commands/blind-spot-pass"><img src="https://agentmods.dev/badge/commands/emaraschio/cursor-commands/blind-spot-pass/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/commands/emaraschio/cursor-commands/blind-spot-pass"><img src="https://agentmods.dev/badge/commands/emaraschio/cursor-commands/blind-spot-pass.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00044 | $0.01321 |
| Opus 5 | $0.00022 | $0.00660 |
| Sonnet 5 | $0.00009 | $0.00264 |
| Haiku 4.5 | $0.00004 | $0.00132 |
Grade A, and why
blind-spot-pass 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 9d 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 — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Overview
Before building, run a blind spot pass on the user's rough plan. Treat the prompt as the map and the real project as the territory: classify gaps across four knowledge quadrants (with Dependencies, Edge cases, and Blast radius under unknown unknowns), ask 5 to 10 high-leverage interview questions, and halt with a two-step handshake before building. Default intensity is standard; red-team only when the user asks to challenge or break the plan. Plan-only on delivery and on pass approval. Full workflow: .cursor/skill-contracts/blind-spot-pass/SKILL.md (user install: ~/.cursor/skill-contracts/blind-spot-pass/SKILL.md).
Defaults
| Setting | Default |
|---|---|
| Deliverable | Five-section pass in chat (four quadrants + interview questions; §4 has required sub-bullets) |
| Intensity | standard always; red-team only if user asks to challenge / break / red-team |
| Handshake | Approve or edit the pass, then later execute now; pass approval is not build |
| Execution | None in same turn unless explicit skip-the-pass-and-build |
| Question count | 5 to 10, ranked by leverage on structure, architecture, audience, scope, workflow, quality |
| Persistence | Offer save under docs/blind-spot-passes/; write only after user confirms |
| Examples | Generic names only (product-a, service-a, repo1) |
Steps
- Read
.cursor/skill-contracts/blind-spot-pass/SKILL.mdfor the full agent contract; if that file is missing, read~/.cursor/skill-contracts/blind-spot-pass/SKILL.md. - Execute phases in order (Intake → Discovery → Draft the pass → Deliver); do not skip discovery or start building.
- Report the blind spot pass (sections 1 to 5) with §4 sub-bullets and the two-step approval handshake.
- Offer optional save under
docs/blind-spot-passes/when the host hasdocs/; write only after the user confirms.
Anti-patterns
- Classify all four quadrants. Trigger: drafting the pass from a rough plan. Wrong: skipping unknown knowns or unknown unknowns, or collapsing quadrants into a single list. Correct: populate every quadrant explicitly, marking thin quadrants when the context is sparse. Reason: the pass exists to surface what the prompt omitted, and skipped quadrants hide unstated assumptions.
- Require §4 Dependencies, Edge cases, and Blast radius. Trigger: writing unknown unknowns. Wrong: a mushy risk paragraph with no sub-bullets. Correct: include all three sub-bullets, or mark N/A when genuinely empty; do not invent facts. Reason: vague §4 lets the model skip failure modes the pass is meant to surface.
- Ask high-leverage questions, not trivia. Trigger: producing the interview questions. Wrong: listing shallow yes/no questions that would not change the output. Correct: ask 5 to 10 questions ranked by leverage on structure, architecture, audience, scope, workflow, or quality. Reason: low-leverage questions waste the interview gate before building.
- Stay a blind spot pass. Trigger: user did not ask to challenge or break the plan, or pastes a step-by-step implementation plan for technical audit. Wrong: defaulting to a hostile free-form risk report, or replacing the four quadrants with a principal-engineer plan audit. Correct: use standard unless they ask for red-team / challenge / break; stay map/territory and point them at a dedicated plan-preflight skill, not this one. Reason: this skill audits the ask, not a finished build plan; unsolicited red-team burns tokens.
- Two-step handshake: pass approval is not build. Trigger: user says "approved" or "LGTM" on the pass. Wrong: scaffolding or coding in that turn. Correct: acknowledge and wait for a later execute now (unless they explicitly skip the pass and build now). Reason: conflating review with build defeats the gate.
- Implementation notes after execute now. Trigger: follow-up build after execute now. Wrong: silent assumptions during the follow-up build. Correct: maintain an Implementation notes section logging each material assumption (what, why, and what would change if wrong). Reason: unlogged assumptions recreate the blind spots the pass was meant to surface.
- Docs save only after confirm. Trigger: host has a
docs/directory and the pass is ready. Wrong: writingdocs/blind-spot-passes/...unprompted. Correct: offer the path and write only after the user confirms. Reason: unsolicited docs clutter the host workspace. - Do not claim a native product feature. Trigger: describing what this deliverable is. Wrong: implying a built-in Cursor product feature. Correct: present this as a portable planning document for Cursor agents. Reason: a native-product claim misleads the user about what shipped.
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.
- 9d ago First seen · 56 lines · 44 tokens per session scan A 6ae411e1cc53
blind-spot-pass is a command published in the GitHub repository emaraschio/cursor-commands (9 stars, last pushed 1mo ago), licensed MIT. It adds 44 tokens to every session and 1,321 once invoked, about $0.0002 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.
Other commands, from other repositories
grok_execute
Explicitly turn the task-local Grok executor mode on or off.
grok_init
Discover, verify, and persist the local HTTP proxy used by Grok Build Supervisor.
handle-pr-comments
Fetch all PR comments, classify by severity, and present for user approval before acting.
self-review
Review only changed files in this branch Your primary goal is to provide valuable, trustworthy feedback while avoiding false positives and low-impact commentary.
create-pr
Commit changes and create a PR with proper formatting.
back-to-master-delete-branch
Cleanup current branch and switch to default branch with latest changes.