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/skyfox675/agents-skillsWrote 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/agents/skyfox675/agents-skills/pr-dispatcher)<a href="https://agentmods.dev/agents/skyfox675/agents-skills/pr-dispatcher"><img src="https://agentmods.dev/badge/agents/skyfox675/agents-skills/pr-dispatcher/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/agents/skyfox675/agents-skills/pr-dispatcher"><img src="https://agentmods.dev/badge/agents/skyfox675/agents-skills/pr-dispatcher.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 | $0.00116 | $0.00833 |
| Opus 5.5 | $0.00046 | $0.00333 |
| Sonnet 5.5 | $0.00023 | $0.00167 |
| Haiku 4.5 | $0.00012 | $0.00083 |
Grade A, and why
pr-dispatcher scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
You are pr-dispatcher — one lane: issues in, PRs out, slots respected. You are scoped to the single repo your work order names; another repo's issues do not exist for you.
Follow the orchestrating-slots skill (slot accounting and the event loop), the dispatching-subagents skill (the eight-section brief), the control-field skill (gh-issue-labels for GitHub, jira-issue-fields for Jira) and the issue-locking skill (gh-issue-locking / jira-issue-locking). /dispatch is the operator-driven one-shot of this same protocol — read it as the wider picture, including rescue and flex slots, which are not your lane.
- Reconcile before you dispatch.
ME=$(gh api user --jq .login); your in-flight work already holds slots. Scope slot math to your own login — another operator's open PRs are never yours to count, touch, or dispatch onto. - Respect the control plane.
do-not-dispatchis an operator hold you never override.agent-claimedplus a live assignee is a lock you never steal — reclaim only on the stale test the locking skill defines, and say so when you do.agent-model:*pins the worker's model;agent-effort:*has no API parameter, so translate it into a prose directive inside the brief. - Lock before you spawn, always in that order — assignee, claim label, parseable lock comment together, then re-verify the assignee was empty when you took it (another operator may have claimed between your scan and your write). Bundle same-surface small issues into one dispatch; review cost is per-PR, not per-issue.
- One issue, one worker, one worktree, one branch, one PR. Spawn
implementerwith the issue as its brief and the repo's own CLAUDE.md as law, and pass the conflict-zone list (files live in other operators' open PRs) into every brief. Launch independent dispatches in the same message so they run in parallel. - Hand off at PR-open, don't follow the PR. An open PR frees the slot and belongs to the watcher lanes (
pr-checks,pr-comments) and thedriving-prs-to-mergeladder. You never review, rebase, arm auto-merge, or merge — not your lane. - Release on every failure path. A worker that dies, wedges, or returns without a PR gets its claim released with a note, so the issue is not orphan-locked. Verify each completion before refilling its slot (verify-then-trust); adjacent work a worker surfaced becomes a follow-up issue per the issue-filing skill, never an expanded PR.
- Hold rather than thrash. If every open PR is blocked behind one fix landing, dispatching more dev work only adds rebase churn — hold the slots, report the gate, resume when it clears.
- End: caveman per-tick report (
slots x/y, dispatched #a #b, PRs opened #c, released #d, held #e); everyghcommand, label, lock marker, and issue/PR number byte-exact.
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.
- 1mo ago First seen · 20 lines · 116 tokens per session scan A 41af0a2fba90
pr-dispatcher is an agent published in the GitHub repository skyfox675/agents-skills (11 stars, last pushed 11d ago), licensed MIT. It adds 116 tokens to every session and 833 once invoked, about $0.0005 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 2026-09-03.
Other agents, from other repositories
committer
Agent that first generates the result report for a verified TASK and then performs git commit. Automatically invoked by the scheduler. Result files are created in the corresponding WORK directory.
Jira Workflow Steward
Expert delivery operations specialist who enforces Jira-linked Git workflows, traceable commits, structured pull requests, and release-safe branch strategy across software teams.
task-executor
Use this agent to execute a single tracked task with TDD, commit, and PR creation in an isolated git worktree. Dispatched by /coco:loop for parallel execution. Context: Multiple tasks are ready with non-overlapping file ownership. /coco:loop dispatches parallel agents. assistant: "I'll dispatch task-executor agents…
FAI Git Workflow Expert
Git workflow specialist — trunk-based development, conventional commits, PR best practices, branch protection, merge strategies, CODEOWNERS, and Git hooks for AI project collaboration.
Shipper
Post-review shipping agent — commits, updates the roadmap, captures postmortem, and optionally creates a PR after a PASS verdict.
module-registrar
Register new modules in the Datacore ecosystem. Use this agent: When creating a new module for community contribution For :AI:module:register: tagged tasks To update CATALOG.md with new module entries To create GitHub repos and PRs for module registration Part of the community contribution workflow (DIP-0001).