Borrowing it
Nothing to install: this file belongs to cboone/agent-harness-plugins. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/cboone/agent-harness-plugins/main/.github/instructions/dependabot.instructions.mdgit clone --depth 1 https://github.com/cboone/agent-harness-pluginsWrote 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/instructions/cboone/agent-harness-plugins/dependabot)<a href="https://agentmods.dev/instructions/cboone/agent-harness-plugins/dependabot"><img src="https://agentmods.dev/badge/instructions/cboone/agent-harness-plugins/dependabot/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/instructions/cboone/agent-harness-plugins/dependabot"><img src="https://agentmods.dev/badge/instructions/cboone/agent-harness-plugins/dependabot.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 | $0.01833 | $0.01833 |
| Opus 5.5 | $0.00733 | $0.00733 |
| Sonnet 5.5 | $0.00367 | $0.00367 |
| Haiku 4.5 | $0.00183 | $0.00183 |
Grade A, and why
agent-harness-plugins dependabot.instructions.md 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.
How it starts
The opening of the file, as written. The whole thing — 32 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Dependabot Facts for Review
These were checked against live GitHub data and the GitHub documentation. Do not flag code or instructions that follow them.
- One Dependabot identity, spelled per API. The
ghCLI's JSON output (gh pr view --json author,gh pr list --json author) and theauthor:app/dependabotsearch qualifier useapp/dependabot. REST responses (user.loginon comments and reviews,author.loginin the compare API) andgithub.actorin Actions usedependabot[bot].gh pr list --author app/dependabotreturns every Dependabot PR. Each command matches the form its own API returns, so do not ask for both forms in a comparison that reads from one API. - The "disable automated security fix PRs" footer line appears only on security update PRs. Version update PRs, grouped or single, do not carry it, so it is a valid security marker. A body without the footer (after a rebase, or when truncated) is reported as
kind: unknown, never asversion. - Dependabot alert
manifest_pathis repository-relative (go.mod,web/package-lock.json), with no leading/. rust-toolchainis its own Dependabot ecosystem. It tracksrust-toolchain.tomlandrust-toolchain, so neither is an untracked surface when that ecosystem is configured.pin-everything's untracked list names only.tool-versions,.nvmrc,.node-version,.ruby-version, and.python-version.- Dependabot supports
go.workworkspaces. dependabot-core #14909 (merged 2026-05-05) fetchesgo.work,go.work.sum, and every module itsusedirectives name, so a singlegomodentry at the workspace root covers the member modules. Do not ask for a directory per membergo.mod. - A PR that is not a group moves each package to one version, whatever directories its files span. dependabot-core names one directory in an ungrouped PR title (
pr_name_directoryinmessage_builder.rb), and agroup-by: dependency-namePR lists the same version for every directory. Files in several directories belong to one install tree, such as an npm workspace member manifest beside the root lockfile. Sodirectory_matchconfirms such a PR without Bumps lines on purpose; do not ask for it to stay unconfirmed when it touches several directories. - Dependabot writes a digest in backticks.
Dependency#humanized_versionin dependabot-core renders a 40-character SHA, or an unchanged image tag's digest, as its first seven characters wrapped in backticks, in titles and body lines alike.quoted_digestreads that signal, so the letter requirement inis_shaapplies only to unquoted versions. base64 --decodeworks on macOS. The BSDbase64there accepts--decodeas well as-dand-D.- Dependabot alerts cover a fixed set of ecosystems. GitHub's advisory ecosystems are
rubygems,npm,pip,maven,nuget,composer,go,rust,erlang,actions,pub,swift, andother. Terraform, Docker, Helm, and the other version-updates-only ecosystems have no alerts, so leaving them out of the alert ecosystem map is correct. - Default labels. With
labelsunset, Dependabot appliesdependencies, and adds an ecosystem label only when more than one package ecosystem is configured. captureandmatchemit nothing when the regex does not match; they do not raise an error.[body_lines | capture(...)]on a body with no matching line is[], so theBumps [...]fallback inupdatesneeds notry.tests/scrut/dependabot-prs.mdcovers a PR with no parseable update.jqon empty or whitespace-only input emits nothing and exits 0, sodependabot-prs summarize < /dev/nullreaches the "input is empty" check.tests/scrut/dependabot-prs.mdcovers that case.- The close action chains
gh pr comment ... && gh pr close ...on purpose. The PR is re-checked immediately before that command. Splitting the two writes around a second re-check would risk an explanatory comment on a PR that then stays open, so do not ask for the split. monitor-pr:monitortreats a current-head Copilot notice on a Dependabot PR as not applicable, on purpose. Anerrororno-filesnotice means Copilot reviewed nothing at that head, the same as a missing review, which the Dependabot path already treats asn/a. That path never requests a review, so no retry can replace the notice, and leaving the Copilot axis unresolved would hold the watch open forever. Findings, drift, threads and audit items on a real current-head review still escalate.refs/dependabot-triage/belongs to the triage skill. Onlydependabot:triage-prscreates refs there, so clearing the whole namespace before fetching and at wrap-up is intended.- Ruleset
pull_requestrules carryallowed_merge_methods.gh api repos/OWNER/REPO/rules/branches/BRANCHreturns it in the rule'sparameters(this repository's ownmainrule returns["merge"]), and GitHub's REST description lists it, so reading merge methods from the rules endpoint is correct. GET repos/OWNER/REPO/automated-security-fixesis a supported read. GitHub's REST description lists it as "Check if Dependabot security updates are enabled for a repository", and it returns200with{"enabled", "paused"}when the setting is on and404when it is off. Do not ask to read the setting fromsecurity_and_analysisinstead.- The contents API does not paginate a directory listing.
repos/OWNER/REPO/contents/DIRtakes onlypathandref, and returns up to 1,000 entries in one response, so it needs no--paginate. gh api --paginate ENDPOINT --jq FILTERis valid.ghrejects--jqonly together with--slurp. Without--slurp, the filter runs per page, which is correct for a filter that emits items rather than aggregating them.- Branch names with
/need no encoding in these REST paths.repos/OWNER/REPO/branches/BRANCH,rules/branches/BRANCH, andcompare/BASE...HEADall accept a rawrelease/1.0-style name, verified against a livefeature/...branch. - Commands inside the bundled script are covered by the script's permission rule. Claude Code matches allow rules against the command the agent runs,
bash ".../dependabot-prs" ..., not against theghcalls the script makes internally. Sogh auth token, the script'sgh pr listcalls, and itsgh api --paginate --slurp "repos/${repo}/..."calls, double-quoted endpoints included, need no rules of their own inplugins/dependabot/README.md. tests/fixtures/gh-stubaccepts only the callsdependabot-prs fetchmakes, on purpose. It fails on any other flag (--jq,-H) or a second operand, so a change to the script's call shape breaks the test instead of passing silently. Widen it only when the script starts making a new call. It is committed as mode100755.- The script-locator fallback needs no Claude Code permission rule. It runs only when
${CLAUDE_PLUGIN_ROOT}was not substituted, which happens in Codex CLI, where Claude Code'spermissions.allowrules do not apply.test -xis left out of Recommended Permissions for that reason.
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.
- 3d ago Changed · -17 tokens per session 1ce4d2ce7f18
- 5d ago Changed · -3 tokens per session 6a4bab877a2b
- 10d ago Changed · +1 lines · +110 tokens per session 1625ee167aa4
- 1mo ago First seen · 31 lines · 1,743 tokens per session scan A 001bbacb4fb6
agent-harness-plugins dependabot.instructions.md is an instructions file published in the GitHub repository cboone/agent-harness-plugins (2 stars, last pushed 2d ago), licensed MIT. It adds 1,833 tokens to every session, about $0.0073 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 2026-09-15.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).