Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add anilcancakir/claude-code/plugin install acWrote 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/anilcancakir/claude-code/auto)<a href="https://agentmods.dev/skills/anilcancakir/claude-code/auto"><img src="https://agentmods.dev/badge/skills/anilcancakir/claude-code/auto/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/anilcancakir/claude-code/auto"><img src="https://agentmods.dev/badge/skills/anilcancakir/claude-code/auto.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 | $0.00037 | $0.07427 |
| Opus 5.5 | $0.00015 | $0.02971 |
| Sonnet 5.5 | $0.00007 | $0.01485 |
| Haiku 4.5 | $0.00004 | $0.00743 |
Grade C, and why
auto 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 today.
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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
`git push`, `git reset --hard` and `rm -rf`, and it does not exempt you How it starts
The opening of the file, as written. The whole thing — 441 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/ac:auto
Autonomous runner. It fixes what done means before any work happens, hands the work to the existing
ac:plan and ac:execute chain, and ends with a verdict written by a gate that did none of the work.
Request: $ARGUMENTS
Standing rules
These hold for the whole run, including after a compaction. Everything below this block is procedure; these are the bounds. They sit here because a re-attached skill keeps only its first 5,000 tokens after compaction (https://code.claude.com/docs/en/skills.md) and this body is larger, so a rule further down is gone from context on exactly the long runs that need it.
Turn termination. Your turn ends on exactly one of: the Phase 5 closing report, a terminal branch that
deleted .ac/state/active-auto.json first, or a one-line wait status while background workers you spawned still run (each task-notification starts your next turn; never wait with a sleep or polling loop). Nothing else ends it. While that marker exists,
.ac/auto/<slug>/verdict.md does not, and no background worker is running, the plugin's auto Stop guard blocks the turn from ending
(${CLAUDE_PLUGIN_ROOT}/hooks/stop-guard-auto.sh), and its block budget is 3 PER PHASE, so the gating
handoff always starts with a fresh three however many were spent earlier. Writing the verdict at Phase 4 is the only action that releases it for good. Never end a turn by
describing what you would do next, and never propose that the user open a fresh session to continue.
Marker path. .ac/state/active-auto.json is the single live-run record; the run directory is
.ac/auto/<slug>/. Create the run directory before or with the marker: the guard exits 0 when the directory is
missing (the run-directory check in stop-guard-auto.sh), so a marker written first leaves the guard silently
inert for the whole run.
Questions reach the user; only plan-side ones continue on a timeout. Every interview gate in ac:plan
Stage 3 and every BLOCKER in either chained skill reaches the user exactly as it would in a supervised run. You
never answer one ahead of the user and never infer the answer from the request prose. What happens when nobody
answers is set by the On AFK timeout column of each chained skill's <auto_mode> table, and only in this
session while it owns .ac/state/active-auto.json: a plan interview gate or another plan-side question continues
with its Recommended option once the question-timeout hook says so, recorded as "timed out, not user-approved";
an Oracle CRITICAL (plan 3.5c) continues at most once per run; an execute BLOCKER (2i dependency failed, 2j three
failures spanning two waves, 3c plan-spec issue, a failed checkpoint commit) waits for the user however long
that takes. No timed-out answer is ever written as overrides[].accepted_by: that field exists only when a real
user answer produced it. With AC_AFK_TIMER unset, or in a background session where Claude Code never arms the timer, this skill treats the timer as absent, and an adopted run locks its remaining plan
questions through Force-finalize instead (Phase A4). The promise this skill makes is that it does not stop for
anything else.
What ships with it
1 file 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.
- today Changed · +113 lines · +14 tokens per session d5b10f00d0e1
- 22d ago Changed · +1 lines · -18 tokens per session e1391732ffac
- 1mo ago First seen · 327 lines · 41 tokens per session scan C eb63e40c904e
auto is a skill published in the GitHub repository anilcancakir/claude-code (3 stars, last pushed today), licensed MIT. It adds 37 tokens to every session and 7,427 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
vigil-instrument
Instrument a service with OpenTelemetry — RED metrics, structured logs, distributed tracing, and health checks. Outputs actual code and config, not a plan. Use when asked to "add monitoring", "instrument this", "add logging", "set up tracing", or "observability".
monday-planning
Suggest a weekly training structure from recent load and deliver the plan.
pave-env
Set up local development environments — devcontainers, Docker Compose, one-command setup, dev/prod parity. Use when asked to "set up dev environment", "devcontainer", "docker compose for dev", "local development setup", or "one command to run".
design-system-reference
Style guides and implementation rules for frontend design. Works with design-discovery agent which handles context gathering and VS-based style recommendations. Contains detailed style guides, anti-patterns, and implementation checklists.
github-pr-creation
Creates GitHub Pull Requests with automated validation and task tracking. Use when user wants to create PR, open pull request, submit for review, or check if ready for PR. Analyzes commits, validates task completion, generates Conventional Commits title and description, suggests labels. NOTE - for merging existing…
boss-briefing
Vault health check — workflow pattern analysis, profile sync, session gap recovery, persona rule proposals.