Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/dork-labs/dorkosnpx agentmods add skills/dork-labs/dorkos/creating-pull-requestsWrote 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/skills/dork-labs/dorkos/creating-pull-requests)<a href="https://agentmods.dev/skills/dork-labs/dorkos/creating-pull-requests"><img src="https://agentmods.dev/badge/skills/dork-labs/dorkos/creating-pull-requests/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/skills/dork-labs/dorkos/creating-pull-requests"><img src="https://agentmods.dev/badge/skills/dork-labs/dorkos/creating-pull-requests.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 | $0.00109 | $0.12813 |
| Opus 5.5 | $0.00044 | $0.05125 |
| Sonnet 5.5 | $0.00022 | $0.02563 |
| Haiku 4.5 | $0.00011 | $0.01281 |
Grade C, and why
creating-pull-requests scanned grade C with 1 finding 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 24d 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.
Hidden instructionshighPrompt injection
Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.
<!-- The double quotes are load-bearing: verbatim what the post-commit hook writes. Prettier rewrites quotes inside an embedded fence, hence the ignore. --> How it starts
The opening of the file, as written. The whole thing — 875 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Creating Pull Requests
How DorkOS PRs are opened and how the automated Claude review behaves on them. This repo is routinely multi-agent, so the mechanics below keep PRs clean and the review loop cheap.
When to use
- You have finished a branch and are deciding when to open the PR.
- You are about to open a PR (from an agent or by hand).
- A PR already has review feedback and you want another pass once it is addressed.
- You want to dial a PR's review up, down, or off.
- A PR of yours is red, ejected from the merge queue, or not merging, and you need to know whether acting helps or only adds load.
The order: review the branch, then open the PR
The independent adversarial review runs against a pushed branch, before a PR exists. Opening the PR is the last step, not the first:
- Build in an isolated worktree and run the local gates.
- Push the branch. Open nothing.
- A reviewer fetches and checks out that branch. It does not need a PR.
- Findings, fixes, convergence.
- Squash to one clean commit carrying one changelog fragment.
- Then open the PR, already reviewed, so the repo's automated review spends its single pass on final content.
A pushed branch that will not become a PR is yours to delete. GitHub removes a
branch from origin only when its PR merges, so a review branch you abandon, or an
attempt that lost to a sibling, stays forever, and each one keeps a Neon preview
database alive. Delete it from origin (git push origin --delete <branch>) the
moment it is abandoned or superseded. A codex/archive/* twin goes the same way
once the PR it certifies has merged.
Every reason below is a cost measured on 2026-07-27/28, not a preference:
- Merge churn. A PR held open across review rounds watches
mainmove under it. One night of that took roughly a dozen branch updates (the pre-queue rule, retired 2026-07-28) and still left aDIRTYPR needing a semantic conflict resolved by hand. - The changelog gate. A fragment claims the commits that existed when it was
written. Review-fix commits arrive afterwards uncovered, and
fragment-presentfails. Three PRs failed this way in one night. One commit and one fragment removes the failure mode instead of working around it. - An open PR is an invitation to act. Another session armed auto-merge on a PR carrying two open blocking findings. Both were browser-only and untested, so every check was green with the defects live. A branch invites nothing.
- The automated review fires on open, so opening early spends it on a draft.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 24d ago Changed · +12 lines 62bb134ee795
- 26d ago Changed · +10 lines 6d104e696726
- 29d ago Changed · +35 lines 3511aafc540d
- 1mo ago Changed · +139 lines · +25 tokens per session 4a70a3a13a21
- 1mo ago Changed · +17 lines c4bedcb64ffb
- 1mo ago Changed · +5 lines 34aa4d05884f
- 1mo ago Changed · +17 lines 5945f805ceb3
- 1mo ago First seen · 640 lines · 84 tokens per session scan C ce3d4f212428
creating-pull-requests is a skill published in the GitHub repository dork-labs/dorkos (12 stars, last pushed today), licensed MIT. It adds 109 tokens to every session and 12,813 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it C with 1 finding (hidden instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
code-review-loop
Iterative review+fix loop for BASESHA..HEAD: generate findings, apply accepted fixes, run checks, commit, and re-review up to 3 iterations or until clean.
dedup
Dedupe-only pass for BASESHA..HEAD: remove duplicate code introduced by the diff or reuse existing shared utils; applies changes + commits.
plan-critique-loop
Critique and revise an existing plan doc up to 3 iterations, using accept/reject triage and stopping early when no important feedback remains. Use when refining a plan/.md before implementation.
submit-review
Reference for how a finished ticket reaches human review in the Gaffer factory. Submission is RUNNER-OWNED — the agent commits its work and evidences the ACs, then stops; the runner records the delivery and submits for review (and pushes / opens a PR only when the operator enabled that). Invoke when you think your…
cc-autopilot
Run cc-autopilot, a perpetual autonomous AUDIT/GENERATE→TRIAGE→BATCH→FIX→VERIFY→REVIEW→COMMIT loop that holds every change to a configured quality bar through a judge panel. Its fixers each own a file-disjoint slice of the tree so no two can ever touch the same file, and nothing unverified reaches git. Works on a web…
github
GitHub operations expert for PRs, issues, code review, Actions, and gh CLI.