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/CassetteTapeCrackle/claude-startersWrote 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/cassettetapecrackle/claude-starters/starter-author)<a href="https://agentmods.dev/agents/cassettetapecrackle/claude-starters/starter-author"><img src="https://agentmods.dev/badge/agents/cassettetapecrackle/claude-starters/starter-author/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/cassettetapecrackle/claude-starters/starter-author"><img src="https://agentmods.dev/badge/agents/cassettetapecrackle/claude-starters/starter-author.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 | $0.00050 | $0.00472 |
| Opus 5.5 | $0.00020 | $0.00189 |
| Sonnet 5.5 | $0.00010 | $0.00094 |
| Haiku 4.5 | $0.00005 | $0.00047 |
Grade A, and why
starter-author 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 2mo 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 author new starters for the ~/claude-starters repo. Work TDD and keep everything lean.
Method
- Read an existing starter as a pattern (e.g.
starters/rust/orstarters/cpp/) andstarters/<x>/manifest.json. Match structure and tone. - Create
starters/<stack>/CLAUDE.md:- First line after the title:
Framework: __FRAMEWORK__ — locked. ...(the__FRAMEWORK__token is substituted at apply time — keep it verbatim). - Sections: Build & deps · Code · Tests · (Depth, if a matching
clean-code-<x>skill exists — name it explicitly) · Commands. - Rules must be the language's canonical conventions (formatter, linter, test runner, error-handling norm). Greppable names, one responsibility, early returns. No line-count dogma.
- First line after the title:
- Create
starters/<stack>/manifest.json:stack,frameworks(array), optionalskill, and acommandsmap (test/lint/format/etc.). - Add assertions to
tests/blind-starters-content.test.sh(or a stack test) covering a signature token + valid manifest. - If the language needs deeper, on-demand rules, create
skills/clean-code-<x>/SKILL.md(frontmattername+ triggerdescription, lean body) — the plugin auto-discoversskills/.
Principles
- The project
CLAUDE.mdis always in context — keep it tight; put depth in the on-demand skill, not here. - Run the full suite (
for t in tests/*.test.sh; do bash "$t"; done) andshellcheckbefore reporting done. - Report a concise summary: files added, what the rules cover, test result.
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.
- 2mo ago First seen · 23 lines · 50 tokens per session scan A 77fc3adab9cf
starter-author is an agent published in the GitHub repository CassetteTapeCrackle/claude-starters (1 stars, last pushed 2mo ago), licensed MIT. It adds 50 tokens to every session and 472 once invoked, about $0.0002 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 2026-08-31.
Other agents, from other repositories
orc-prd-analyzer
Investigator role — analyses a PRD, spec or brief into structured requirements, ambiguities, missing edge cases and clarifying questions. Dispatched by /orc:flow triage and /orc:plan when the input reads like a brief.
orc-debug-investigator
Investigator role — root-cause investigation for hard bugs: reproduce, minimise, hypothesise, instrument, locate the line. Returns a written diagnosis and a regression-test recommendation; never fixes. Dispatched by /orc:debug and /orc:incident.
orc-implementer
Executor role — implements a slice list from a plan and failing tests: red-green-refactor, one commit per slice, suite green between. Default executor in /orc:flow Phase 5 and /orc:fan-out. Escalates on ambiguity, new dependencies, or three red attempts.
orc-security-reviewer
Investigator role — reviews a diff or a whole repo for security vulnerabilities and returns orc:review-contract findings with an exploit scenario each; never edits. Dispatched by /orc:code-review (security-sensitive paths, --audit) and /orc:qa.
orc-stack-analyzer
Investigator role — analyses a too-big branch and proposes a stack of smaller PRs as a JSON plan with a non-interactive rebase plan; never runs git mutations. Used by /orc:stack-pr --smart.
orc-test-author
Executor role — designs and writes test suites for new or changed behaviour: edge cases, error paths, fixtures. Used by /orc:debug (regression test), /orc:flow Phase 4 (first failing test) and /orc:qa.