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 sauravvarma/garage --skill spec-researchgit clone --depth 1 https://github.com/sauravvarma/garageWrote 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/sauravvarma/garage/spec-research)<a href="https://agentmods.dev/skills/sauravvarma/garage/spec-research"><img src="https://agentmods.dev/badge/skills/sauravvarma/garage/spec-research/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/sauravvarma/garage/spec-research"><img src="https://agentmods.dev/badge/skills/sauravvarma/garage/spec-research.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 | $0.00175 | $0.03916 |
| Opus 5 | $0.00088 | $0.01958 |
| Sonnet 5 | $0.00035 | $0.00783 |
| Haiku 4.5 | $0.00017 | $0.00392 |
Grade A, and why
spec-research 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 9d 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 — 205 lines — stays where its author put it; the contents beside it link to each section on GitHub.
spec-research — turn a thin spec into a reviewable draft
This skill exists because most "the AI shipped the wrong thing" bugs aren't coding bugs — they're scoping bugs. The spec described the happy path; nobody enumerated the empty / error / role-gated / mid-fetch branches; the code shipped what was described. /spec-research is the spec-completion phase between scaffolding (/spec-first-project-setup) and implementation (/code-agent).
The output is always a proposal, never a unilateral edit. The user accepts/edits/rejects rows. The skill does the research so the user can review, not the other way around.
When to invoke
Invoke directly whenever an IDEAS doc is missing detail you'd need to implement or QA against. Also invoked indirectly: /code-agent and /visual-qa route here when their preflights detect a thin spec.
Typical triggers:
- "The IDEAS doc has no Page states table — derive one."
- "We're porting Sleep Insights from mweb — read the source and reconcile against the IDEAS doc."
- "This feature has only 2 locked decisions; what else needs to be locked before I can implement?"
- "This route's IDEAS doc was written 3 commits ago — has it drifted from the code?"
If the IDEAS doc doesn't exist at all, route the user to /spec-first-project-setup to scaffold it first; this skill fills, it doesn't scaffold.
Inputs
- Required: path to a
[FEATURE]-IDEAS.mdin the project (or the project root + a feature name to fuzzy-match). - Required: project root with
CLAUDE.mdanddocs/REPO-CONVENTIONS.md— to know the data flow, parser rules, token system. - Optional but high-value:
- Source file or directory to port from (mweb component, legacy app, etc.)
- API contract documentation or example responses
- Sibling IDEAS docs in the same route family (shared decisions cross-reference)
- Design comp (Pencil, Figma) — read structure, not just visuals
If the IDEAS doc references mweb / a legacy source and that source isn't readable, ask the user for a pointer before proceeding. Don't guess what the source renders.
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.
- 9d ago First seen · 205 lines · 175 tokens per session scan A 4eee01bc5fe8
spec-research is a skill published in the GitHub repository sauravvarma/garage (3 stars, last pushed yesterday), licensed MIT. It adds 175 tokens to every session and 3,916 once invoked, about $0.0009 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.
Other skills, from other repositories
designlang-tokens
Use when styling UI for cal.com — references the extracted design system tokens instead of inventing colors, spacing, or typography.
stakeholder-brief
Write a one-page stakeholder brief translating design system health or status into business language. Trigger when someone says: stakeholder update, exec brief, leadership summary, status report for leadership, system status for non-designers, write a brief for the business, or anything about communicating design…
token-audit
Audit a design system's token definitions for naming violations, missing semantic tiers, and structural debt. This audits how tokens are defined and organised, NOT how they are consumed in code. Trigger when someone says: audit my tokens, token naming review, are my tokens consistent, token health check, review my…
figma-variable-audit
Audit Figma variable collections against token architecture best practices. Trigger when someone says: audit my Figma variables, check my Figma tokens, are my variables structured correctly, Figma variable health, review my variable collections, variable naming check, or anything about auditing the quality or…
governance-encoder
Convert governance policies into machine-executable JSON constraint files that AI agents and CI pipelines validate against automatically. This produces rule engine files in .ai/governance/, NOT narrative decision records or documentation. Trigger when someone says: encode governance rules, governance as code, automate…
system-benchmark
Benchmark a design system against industry standards and comparable public systems, producing a qualitative comparison across dimensions with specific, named reference points. Goes beyond an internal health assessment to answer 'how does our system compare to what good looks like out there?' Trigger when someone says…