awwwards-doctor

Awwwards MCP Doctor

You are repairing the awwwards-mcp toolchain or working around it mid-task. Work in this order: diagnose → fix mechanically → fix parsers (if drift) → recover the user's task. Never delete user data; the only destructive step (cache DB moved aside) keeps the old file — in the OS temp dir for --fix, beside the cache after a full backup for the manual reset.

Profile note: this skill's tools are search_sites, get_site_details, capture_live_site, analyze_page_structure, capture_component, export_design_tokens and analyze_images. The examples below use those 1.x names (the default full profile). On AWWWARDS_MCP_PROFILE=v2, use search_references {request:{kind:"sites", ...}}, get_reference {request:{kind:"site_details", slug}}, capture_page {request:{kind:"screenshot", url}}, inspect_page {request:{kind:"structure", url}}, capture_page {request:{kind:"component", url}}, inspect_page {request:{kind:"tokens", url}} and audit_page {request:{kind:"images", url}} instead. Full mapping: v2-routing.md.

1. Run the doctor (awwwards-mcp source checkout)

Run these from the root of the awwwards-mcp source checkout (github.com/INSANE0777/Awwwards-mcp), the repository whose package.json is awwwards-mcp — not from an awwwards-skills checkout. The npm package does not ship these scripts; if you installed the server, use section 3 instead.

npm run doctor             # diagnose only
npm run doctor -- --fix    # diagnose AND apply available fixes
npm run doctor -- --json   # machine-readable report

The doctor checks five things and knows how to fix the mechanical ones:

Check	Failure meaning	What --fix does
network	awwwards.com blocked/challenged the request	nothing to fix locally — wait, retry later; never retry in a loop
parser-drift	awwwards.com markup changed under the parsers	captures the raw live page into docs/drift-<timestamp>/ (gitignored) as the fixture for the fix
playwright-chromium / ffmpeg-static	optional capture deps missing or broken	installs them (dev deps + npx playwright install chromium)
cache	SQLite DB corrupt/locked, or index older than 30 days	moves an unreadable DB aside (kept in temp) and rebuilds via npm run index
boot	dist/cli.js doesn't start	npm run build

Re-run the doctor after fixing; the verdict must be HEALTHY before you claim the MCP is repaired.

2. Fixing real parser drift (the only code-level failure)

If the doctor reports parser-drift, the anchor probes say which parser broke. Repair sequence — keep it mechanical, verify at each step:

1. Capture the evidence: the --fix snapshot in docs/drift-*/ IS the new truth. (The daily GitHub Action also opens/updates a parser-drift tracking issue with the drifted anchor names.)
2. Locate the new anchors: open the captured HTML and find the markup that replaced the old anchor (search the old anchor's purpose, e.g. the card JSON blob, the Elements section heading, the og:image meta).
3. Re-anchor the parsers in src/parsers.ts — change the minimum needed (one split/indexOf/regex literal), not the surrounding logic.
4. Update the probe list in scripts/parser-drift-probe.mjs (in the same awwwards-mcp checkout; PROBE_GROUPS) so the same failure is detectable next time.
5. Refresh the fixture: download the live page with the same UA the client uses and replace test/fixtures/listing.html (or the detail fixture). The offline tests run against fixtures only — a refreshed fixture proves the re-anchor against reality.
6. npm run typecheck && npm test — all tests must pass.
7. Commit as fix: re-anchor <parser> after awwwards.com markup change, reference the tracking issue.

Note: the annual-award tools (get_annual_*) come from the official JSON API (src/annual.ts), not the parsers, and the expanded award-cut listings (sites_of_the_month/year, nominees, winner_category_*) have their own trimmed fixtures under test/fixtures/awards/ — refresh those when their anchors drift, alongside the daily listing's.

3. End users (installed via npm, not a repo checkout)

The doctor scripts live in the repo, so for an npm-installed server:

1. Check the version notice: stderr at startup prints when a newer awwwards-mcp exists. Update with npm install -g awwwards-mcp@latest (or clear the npx cache) and restart the agent — a surprising share of "broken" reports are old parsers fixed in a newer release. AWWWARDS_AUTO_UPDATE=1 in the MCP server env opts into background self-update on startup.

2. Back up, then reset only the database. Stop the agent/MCP server first so the DB is not mid-write. Resolve the cache root with CACHE="${AWWWARDS_CACHE_DIR:-$HOME/.awwwards-mcp}" (the server reads the same variable). Copy it aside first, then move only the SQLite files; the next index/run rebuilds them:

   CACHE="${AWWWARDS_CACHE_DIR:-$HOME/.awwwards-mcp}"
   STAMP="$(date +%Y%m%d-%H%M%S)"
   cp -R "$CACHE" "$CACHE.backup-$STAMP"
   mkdir -p "$CACHE.moved-$STAMP"
   for f in "$CACHE"/cache.db*; do [ -e "$f" ] && mv "$f" "$CACHE.moved-$STAMP/"; done
   

   A blanket delete of the cache root would also destroy preferences.json (setup choices), skill-memory/findings.jsonl (learned findings), winner watches (rows inside cache.db), and captures in images/ and scene-captures/. The targeted reset loses only the search index and watch rows — the index rebuilds, watches must be re-added — while preferences, learned findings and captures are kept. Never rm -rf the cache root.

3. If it still fails, it's upstream (block or drift): open an issue at the awwwards-mcp repo attaching the raw HTML of the failing URL.

4. Recovering the in-flight task (the reason you noticed)

The user's build/research doesn't wait for the repair. Degrade gracefully:

• Stale cache is your friend: search_sites/get_site_details serve cached data when live fetches fail — say so in the reply ("7-day-old cached data").
• BlockedError means stop, not retry: the client never retries through blocks; tell the user and continue with cached/alternate data.
• Capture tools down (playwright missing): fall back to get_site_details screenshots (CDN-served, no browser needed) until npm run doctor -- --fix restores chromium.
• Token/image tools down: export_design_tokens reuses the capture_component pack (same playwright/chromium requirement), and analyze_images needs the same browser. Without playwright both return the raw browser/module error (e.g. Cannot find package 'playwright'), not the friendly install hint — capture_live_site / analyze_page_structure are the tools that return the hint. Cached get_site_details screenshots are the fallback. analyze_images never downloads assets — zero images means the page has no <img> elements (canvas/CSS-background content) or the scan found none, not a broken audit.
• One band, a preloader, or mode:"anchors": 1.14 detects inner scroll roots (scrollRoot) and waits out full-screen loaders; 1.15 returns mode:"anchors" for inner-scroll pages — six ordered viewport shots down the scroller as JSON paths (anchors[].scrollTop/file) with no inline image. That is the honest success payload, not a broken capture; use scrollCapture:"viewport" when you want one explicit viewport shot. A scrollRoot re-query error (0 or >1 matches) means the DOM changed between scan and shot — re-run or report the label and match count. If a page still returns a single band, its content may be in shadow DOM or canvas; inspect and report the limitation instead of treating it as a full band map.
• Parser drift mid-task: results may be partially empty (e.g. details without palette). Surface which part is missing; don't silently present degraded data as complete.
• After the MCP is healthy again, re-run the failed calls and reconcile with whatever the fallback produced.

Close the flywheel

Repairs are the highest-value lessons - they were expensive to learn. In a checkout of the awwwards-skills repository (the installed skill copy does not include this dev script), after a successful diagnosis + fix: node scripts/skill-memory.mjs record --skill awwwards-doctor --phase repair --symptom "..." --rule "..." [--evidence <report path>]. Distill folds recurring failure signatures into this installed copy so the next repair starts knowing them. When a rule proves out, promote --skill awwwards-doctor outputs it cleaned (no machine-local evidence paths) for the shipped-copy PR.

Anti-patterns

• Retrying blocked requests in a loop — politeness rules and block detection exist precisely so this doesn't hammer awwwards.com.
• Diffing raw HTML to detect drift — cosmetic page changes would false-alarm daily; only parser-anchor probes matter.
• "Fixing" parsers by loosening them to return empty results quietly — a parse that can't find its anchors must error loudly (that is the drift signal).
• Blanket-deleting the cache root to "reset" — it destroys preferences, learned findings, winner watches and captures. Back it up, then move only cache.db*.
• Treating the doctor's UNHEALTHY verdict as done because one check passed.

<!-- skill-memory:start --> <!-- skill-memory:end -->