Re-collects, on the collector, the pages a Fetchproof brief rests on before a build starts or a release ships, using research.mjs --refresh-days, so a vendor that changed a limit, price or schema is caught in the corpus before it reaches code. Use before a release, when a brief is older than its refreshDays, when a builder reports behaviour that contradicts the brief, or when the user asks whether the research is still current. Collector only.
Installs into .claude/skills of the current project.
Are you the author of Freshness Recheck?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/stepenkoanatoli-freshness-recheck)
---
name: freshness-recheck
description: Re-collects, on the collector, the pages a Fetchproof brief rests on before a build starts or a release ships, using research.mjs --refresh-days, so a vendor that changed a limit, price or schema is caught in the corpus before it reaches code. Use before a release, when a brief is older than its refreshDays, when a builder reports behaviour that contradicts the brief, or when the user asks whether the research is still current. Collector only.
---
# Freshness recheck
A capture is true of the day it was fetched. Before the code that depends on it ships,
check the day has not moved on.
## Steps (collector only - a builder refuses to collect)
1. `node "$HOME/.agents/fetchproof/bin/research.mjs" --status` - the budget, and the
corpus's age.
2. `node "$HOME/.agents/fetchproof/bin/research.mjs" --refresh-days <n> --dry-run` -
which captures are older than `n` days and would be re-collected. Choose `n` from how
fast the fact goes stale (the map's D-6 row), not from habit. For the rows a
**volatile** requirement rests on (a price, rate limit, quota, API or schema version,
plan or tier boundary - auto-build marks them), `n` is the project's `refreshDays`
from `research/plan.json`, never longer.
`--refresh-days 0` does not re-collect a capture made today: it re-collects a capture
older than `n` days, and today's is 0 days old. For a same-day re-check, use `--force`
with `--dry-run` first. `--force` re-fetches every page the plan names and runs every
search again, those for closed unknowns included, keeping up to `perQuery` pages each
within the run's scrape budget; the dry run lists what it would spend. `--only <text>`
narrows the searches it runs again, not the plan's pages.
3. Run it without `--dry-run`. Every re-collection spends a credit; when the account runs
out the run stops and asks - never pass `--fallback` on your own initiative.
4. `node "$HOME/.agents/fetchproof/bin/preflight.mjs"` - a row superseded by a fresher
capture **fails the gate** where an unknown still cites it. Re-read the new capture,
move the citation, and rewrite the finding (`finding-rewriter`).
5. If a fact changed: record it as a contradiction in the brief (old capture vs new),
redraft the brief, and tell the builder which E-## rows moved - `cite-in-code`
comments name them.
## Rules
- `preflight`'s age limit is the machine config's `maxAgeDays` (180 by default) and holds
for every project on the machine. Lower it only on a machine that works mostly on
fast-moving vendors; for one project's fast facts, the volatile marking and
`refreshDays` are the tool.
- Never edit an old capture to match the new page. History stays; the new capture
supersedes it.
