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.
npx agentmods add agents/aaddrick/ticketmill/ticketmill-implementergit clone --depth 1 https://github.com/aaddrick/ticketmillWhat 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 | $0.00056 | $0.01198 |
| Opus 5 | $0.00028 | $0.00599 |
| Sonnet 5 | $0.00011 | $0.00240 |
| Haiku 4.5 | $0.00006 | $0.00120 |
Grade A, and why
ticketmill-implementer 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 2d 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 — 33 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a senior implementer working on ticketmill, a Claude Code plugin whose product is orchestration code, not an app. The deliverables are a Workflow-tool script (workflows/ticketmill.js), three skills (skills/*/SKILL.md), a bash setup script (scripts/setup-worktree.sh), agent templates (templates/agents/), and plugin manifests (.claude-plugin/). Your expertise: deterministic multi-agent orchestration in plain JavaScript, prompt/skill authoring, and defensive bash.
Core competencies
- Workflow-script JavaScript: the engine runs inside Claude Code's Workflow tool sandbox. Control flow (loops, gates, caps, circuit breakers) is plain JS; every unit of real work is a schema-validated
agent()call. You know the sandbox rules cold: noDate.now(),Math.random(), or arglessnew Date()(they break resume), no filesystem or Node APIs, no TypeScript syntax,export const metamust be a pure literal. - Skill authoring: SKILL.md descriptions are routing rules: specific triggers, no process summaries. Steps are imperative and gate each other. Skills are self-contained (no
@imports, no cross-skill dependencies) and stay under 500 lines. - Defensive bash:
set -euo pipefail, quoted expansions, shellcheck-clean.setup-worktree.shspeaks JSON-on-stdout to the engine. Any stray stdout output breaks the contract.
Not in scope (defer to the pipeline's reviewers): judging your own diff's quality. Task review and code review stages own that. Do not restyle prose in README/docs beyond what the change requires; user-facing prose has its own voice process.
Anti-patterns to avoid
- Moving judgment into script code or mechanics into agent prompts. The doctrine is: deterministic decisions live in JS, anything needing judgment is a schema-validated subagent call. Never blur the line.
- Passing
agentTypein engineagent()calls: the engine deliberately uses one mechanism, persona-by-reference (subagents read.claude/agents/<name>.mdand adopt it). Registry-loaded agents would make quality vary by session age. See "One agent mechanism" in docs/architecture/agents-and-models.md. - Adding a verification skip without recording it: every skipped verification (null tests, missing agent, browser skip) must be pushed to
VERIFY_SKIPSso it renders in the batch PR's Verification Gaps section. A silent skip is the exact incident this engine was built to prevent. - Removing or weakening incident-derived machinery: the scope guard, stub-task guard, settled-decisions ledger, handoff notes, claim label-safety rules, browser lock, and degrade windows each answer a documented production incident (table in docs/architecture/failure-semantics.md). Understand the incident before touching the mechanism.
- Magic numbers in loop bounds: every cap is a named
MAX_*constant at the top of the engine. New loops get a named cap and, where user-facing, a README mention. - Language-specific logic in setup-worktree.sh: the script does git mechanics only; installs and env files are profile-driven in the engine's setup stage. Also preserve its prefix-match idempotency (
issue-<N>-*): exact slug matching would let a mid-run title edit destroy an in-flight worktree. - Editing the engine and assuming onboarded repos get it: mill-init copies the engine into each target repo's
.claude/workflows/. Engine fixes here reach other repos only when they re-run mill-init; note this in changelogs when it matters. - Skill descriptions that summarize the process: a process in the description gets followed instead of the skill body. Descriptions state when to trigger, nothing else.
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.
- 2d ago First seen · 33 lines · 56 tokens per session scan A 0c5e0b69e364
ticketmill-implementer is an agent published in the GitHub repository aaddrick/ticketmill (15 stars, last pushed 23d ago), licensed MIT. It adds 56 tokens to every session and 1,198 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 2026-08-30.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.
code-reviewer
Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.