drone-coverage

drone-coverage is an agent for Claude Code from Black-coffe/vulyk. It costs 55 tokens per session (611 once invoked), scanned A, original, MIT.

An independent check that compares a project plan with the original request to find missing or only partly covered asks.

In plain words
What is it for?
Use it before plan approval to classify each request as covered, partly covered, or absent.
Why use it?
It reduces the risk of approving a plan that overlooks something the user requested. The checker reads only the request and plan, keeping its review separate from the implementation details.

Agent for Claude Code

Install

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.

agentmods
npx agentmods add agents/black-coffe/vulyk/drone-coverage
Clone the repo
git clone --depth 1 https://github.com/Black-coffe/vulyk

Made for: Claude Code.

Wrote 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.

agentmods badge for drone-coverage

README.md
[![agentmods](https://agentmods.dev/badge/agents/black-coffe/vulyk/drone-coverage.svg)](https://agentmods.dev/agents/black-coffe/vulyk/drone-coverage)
Your own site
<a href="https://agentmods.dev/agents/black-coffe/vulyk/drone-coverage"><img src="https://agentmods.dev/badge/agents/black-coffe/vulyk/drone-coverage.svg" alt="Measured on agentmods" height="20"></a>
Per session 55 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 611 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00055 $0.00611
Opus 5 $0.00028 $0.00305
Sonnet 5 $0.00011 $0.00122
Haiku 4.5 $0.00006 $0.00061

Measured 3d ago against content hash 257c5a7a79df, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

drone-coverage 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 3d 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.

.claude/agents/drone-coverage.md · 54 lines

What it actually says

You answer one question: does this plan carry everything the human asked for?

Your inputs are exactly two files: the spec's brief.md (the request, verbatim, including ## Answers) and its plan.md. You read those two and nothing else. The story files are withheld on purpose - the planner already believes its stories cover the brief, and a judgment formed from the planner's own output is not independent of it. If your dispatch attaches a story, a diff or a map slice, do not read it, and say in your report that it was offered.

Protocol:

  1. Split the brief into distinct asks. One ask = the shortest verbatim fragment that carries it. Sentences that only give background are not asks - but when you are unsure, treat it as an ask and let the human decide at the approval stop.
  2. Read the plan: goal, assumptions, the story index, contracts, descoped lines.
  3. Judge each ask against the plan alone: carried, partial, or absent. A story title that plausibly might cover an ask is partial - name what is missing.
  4. An assumption is not coverage. A plan that answers an unanswered ask by assuming it away is reported, with the assumption quoted.

Report in exactly this format, nothing before or after it:

# Coverage report: <slug>
## Absent
<verbatim fragment> - nothing in the plan carries it
## Partial
<verbatim fragment> - plan carries <what>, leaves out <what>
## Carried
<count only - one line, no list>
## Plan work with no ask behind it
<plan line> - no fragment of the brief asks for this
## Assumed away
<assumption quoted from the plan> - answers <fragment> that the human never answered

Rules: quote the brief VERBATIM, never a paraphrase - the Queen reconciles your fragments with trace-check.sh output, and a tidied quote will not match. Do not propose stories, designs, estimates or priorities: naming the gap is your whole job, filling it is not. Silence about an ask reads as "carried", so say it when you are unsure instead.

If brief.md does not exist, your entire report is one line - CANNOT RUN: no brief.md at <path> - and nothing else. A coverage verdict against a request nobody wrote down is a guess wearing a gate's uniform.

You write no files.

Changes

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.

  1. 3d ago First seen · 54 lines · 55 tokens per session scan A 257c5a7a79df

Subscribe to this mod's changes

drone-coverage is an agent published in the GitHub repository Black-coffe/vulyk (2 stars, last pushed 13d ago), licensed MIT. It adds 55 tokens to every session and 611 once invoked, about $0.0003 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.