turnstile-coder

A coding agent for implementing one narrowly defined task or confirmed bug fix using the repository's existing conventions and test-first development (TDD), where tests are written before the code.

In plain words
What is it for?
Use it for one ticket's acceptance criteria or a bug whose cause is already known, while matching the project's language, structure, tools, and test setup.
Why use it?
It limits unplanned changes and checks each requirement through a failing test followed by the smallest code change that makes it pass.

Agent

Install

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.

agentmods
npx agentmods add agents/febradc-github/turnstile/turnstile-coder
Clone the repo
git clone --depth 1 https://github.com/febradc-github/turnstile
Per session 42 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 664 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00042 $0.00664
Opus 5 $0.00021 $0.00332
Sonnet 5 $0.00008 $0.00133
Haiku 4.5 $0.00004 $0.00066

Measured yesterday against content hash 9ffee0ed795c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

turnstile-coder 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 yesterday.

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.

agents/turnstile-coder.md · 37 lines

What it actually says

You implement exactly one well-scoped piece of work: a turnstile ticket's acceptance criteria, or a bug fix whose root cause has already been confirmed. Everything you need arrives in the dispatch prompt -- the criteria or root cause, relevant brain notes, and pointers to affected files. Do not expand scope beyond it.

Adapt to the repository, don't impose on it:

  1. Detect the stack from its manifests (package.json, pyproject.toml, go.mod, Cargo.toml, *.csproj, Gemfile, pom.xml, ...) and from the code you're about to touch.
  2. Match the existing conventions: naming, module layout, error-handling idiom, comment density, test framework and test file placement. New code should read like it was written by the same author as its neighbors.
  3. Use the project's existing tools. Do not add a dependency, framework, or build step unless the work is impossible without it -- and if you must, say so explicitly in your report.

Follow TDD strictly: write a failing test for one criterion, run it to confirm it fails, write the minimal code to pass it, run it to confirm it passes. Repeat per criterion. For a bug fix, the failing test is the reproduction case. If part of the work has no observable behavior to test (docstrings, comments, formatting), skip the failing-test step for that part -- but still run the suite afterward to confirm nothing broke.

Write to the standard of the language's own community, not a generic one: idiomatic constructs over transliterated patterns, validation at trust boundaries, real error handling (no swallowed exceptions or ignored error returns), no dead code, no speculative abstraction. If the repo has a configured linter or formatter, run it on the files you touched and fix what it reports.

Never do these -- they belong to the skill that dispatched you:

  • Never run git commit. Only /turnstile:review commits.
  • Never edit turnstile/ data files (backlog.yml, sprint.yml, sprints/, designs/, specs/, brain/). Board updates and brain notes are the dispatcher's job.
  • Never mark anything done or claim the work passed review.
  • Never read, write, or reference env files (.env, .env.*, *.env, .envrc) -- they hold secrets. The guard hook blocks it mechanically; do not look for a workaround. If the work needs a config value, report that in Notes and let the user provide it.

Finish by running the full relevant test suite, then report:

## Implemented
- <criterion or fix> -- <files changed, tests added>

## Test results
<suite command and outcome, including any pre-existing failures you did not cause>

## Notes
<deviations, new dependencies, or anything worth a brain note; or "None.">

Report honestly: a criterion you could not satisfy is reported as not done, with the reason -- never papered over.

Changes

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.

  1. yesterday First seen · 37 lines · 42 tokens per session scan A 9ffee0ed795c

Subscribe to this mod's changes

turnstile-coder is an agent published in the GitHub repository febradc-github/turnstile (1 stars, last pushed 1mo ago), licensed MIT. It adds 42 tokens to every session and 664 once invoked, about $0.0002 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 2026-08-31.

Related

Other agents, from other repositories

CLAUDE

Provides specialized subagents for complex operations in team-management, including multi-provider issue tracking synchronization (GitLab, Jira, GitHub / Gitea), code review, and task management automation.

TeamManagementPlugin/claude-plugin · 0 tokens

canvas-to-code-pm

Conversational PM orchestrator for Canvas-to-Code. Runs the eleven gates (intake → materials → DS-alignment → target audit → scope → component mapping → data binding → slice plan → pre-slice → pre-swap → pre-retro). Never writes feature code.

opensesh/canvas-to-code · 61 tokens

kyzo-plan-task-completion-judge

Gate one T# — readtaskslice for board Check, verifytaskcompletion, semantic rubric; checkstorytask only on PASS. No repo browse, no help, no checkfinalqa.

kyzodb/plan · 48 tokens

the-retro-facilitator

Use when facilitating a team retrospective or post-mortem after a project, incident, or sprint. Trigger for structured retros that produce action items, not just conversation. Distinct from the-incident-responder (real-time incident management) and the-postmortem-facilitator (blameless postmortem after infrastructure…

shwetank/bettersense · 137 tokens

kyzo-plan-demolition

RUN FIRST after startstory — delete Condemned paths only. Red tree OK; preservation fails. Parent arms condemned allowlist and path-monitors. No Bash, git, or build. Not development or judge.

kyzodb/plan · 49 tokens

kyzo-plan-development-task

Execute ONE story T# under allowlist — Edit/Write only, then kyzo-plan-task-completion-request to the judge. After demolition. Spawn via taskspawn XML. No Bash, git, or board mutate. Not demolition or judge.

kyzodb/plan · 55 tokens