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 skills add e0ipso/strikethroo --skill st-refine-plangit clone --depth 1 https://github.com/e0ipso/strikethrooWrote 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/e0ipso/strikethroo/st-refine-plan)<a href="https://agentmods.dev/skills/e0ipso/strikethroo/st-refine-plan"><img src="https://agentmods.dev/badge/skills/e0ipso/strikethroo/st-refine-plan/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/e0ipso/strikethroo/st-refine-plan"><img src="https://agentmods.dev/badge/skills/e0ipso/strikethroo/st-refine-plan.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium MCP Rug Pull · line 22 npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.Fix: Pin the version: npx @scope/[email protected]
- medium Excessive Agency · line 68 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00081 | $0.01354 |
| Opus 5 | $0.00041 | $0.00677 |
| Sonnet 5 | $0.00016 | $0.00271 |
| Haiku 4.5 | $0.00008 | $0.00135 |
Grade A, and why
st-refine-plan 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.
How it starts
The opening of the file, as written. The whole thing — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.
st-refine-plan
Inputs
The user supplies the numeric plan ID conversationally, plus any optional refinement notes.
Operating Procedure
1. Locate the strikethroo root
Run scripts/find-strikethroo-root.cjs from the user's working directory. If
it exits non-zero, stop and ask the user to run npx strikethroo init.
Treat the path it prints as <root> for every subsequent step.
Delegated execution workers skip this step and do not emit update notices.
Run scripts/check-for-updates.cjs "<root>" as a separate command using this
skill's script path. Keep the user's working directory. Read the one JSON line
on stdout. When notice is present, retain its exact text for the final
user-facing response. Never pause for permission and never run an update.
2. Resolve the plan
Run scripts/validate-plan-blueprint.cjs <plan-id> planFile for the plan
file's absolute path. Another field name prints that field instead.
If the script exits non-zero, stop and ask the user to confirm the plan ID. Do not guess a different ID.
3. Load project context
Read <root>/config/STRIKETHROO.md for this project's directory conventions.
Read <root>/config/hooks/PRE_PLAN.md and execute its instructions before
proceeding. Read <root>/config/templates/PLAN_TEMPLATE.md; the refined plan
must still conform to it. Read <root>/config/shared/clarification-gate.md
and apply it in the Clarification Loop.
4. Baseline Review
Read the entire plan end-to-end without modifying it. Record every contradiction and every gap you find under these lenses:
- Context gaps: missing background, assumptions, competing priorities.
- Technical gaps: underspecified architecture, unclear interfaces, missing diagrams.
- Risk gaps: untracked risks, missing mitigations, hand-wavy success metrics.
- Scope issues: gold-plating, ambiguous boundaries, requirements that contradict YAGNI.
Document each one as {section, issue, severity, proposed fix}.
5. Clarification Loop
What ships with it
3 files 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.
- yesterday Changed · -46 lines 24e3de708b59
- 4d ago Changed · -31 lines 7a4946f162a6
- 9d ago First seen · 210 lines · 81 tokens per session scan A c0c64ac9b507
st-refine-plan is a skill published in the GitHub repository e0ipso/strikethroo (50 stars, last pushed today), licensed MIT. It adds 81 tokens to every session and 1,354 once invoked, about $0.0004 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 skills, from other repositories
business-requirements
Read and hand-edit the app's business requirements — the checklist the harness verifies the built app against. MUST be loaded whenever the user asks to add, change, remove, reword, re-prioritise, or reclassify a business requirement (e.g. "add a requirement that…", "that rule is wrong, fix it", "drop the requirement…
first-run-open
Owns the opening move of first-run when the user has NOT handed you a substantive prompt (cold start), and sets the voice-vs-text opening posture for any open. Shows a provisional dream and a guess about who the user is, then learns from their correction (show-to-learn) and re-aims. Hands back to the first-run spine…
disqualification
The honest exit. Loads when qualification signals sustained-low or a realism failure (the ask is genuinely outside QB's wheelhouse). Acknowledges the mismatch plainly, suggests a realistic alternative, preserves dignity, and does not chase. A legitimate designed outcome, not a framework failure.
qualification
Reads qualification signals from the user's input and passive context, sets the archetype and effort mode, and decides whether to proceed up the convince ladder or route to the disqualification skill. Runs alongside the first-run spine from the first substantive message onward. Falls back to "proceed at normal effort"…
build-recovery
Owns the build wait and build failure. Narrates the generation honestly while it runs, and when it errors, names the failure plainly with a real recovery path instead of a generic "oops". Loads when generation is long or returns an error. The single highest-risk moment for the "it's actually building" belief.
chat-formatting-widgets
Governs the custom markdown widgets rendered in the QuantumByte chat interface. This includes the inline highlight-mark span (use on every question the user must answer). You must check this skill on every reply to format highlights.