readiness-audit 2.4.0 → 2.5.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -1,5 +1,19 @@
1
1
  # Changelog
2
2
 
3
+ ## 2.5.0 (2026-11-03)
4
+
5
+ From a project's agent that was asked why llms-full.txt should hold a site's whole text. D4 said "the whole corpus" without saying what that meant, and the scorecard only checked that the file existed. One new warning, no new failures: a site whose llms-full.txt leaves out pages llms.txt links now warns in D4, which lowers its required score until fixed but doesn't change the exit code.
6
+
7
+ - **D4 warns about pages llms-full.txt doesn't list.** Of the llms.txt links it samples, each one that answers HTML (besides the homepage) must appear as a link in llms-full.txt. Files like `/openapi.json` are surfaces, not pages, and aren't counted. A passing D4 says how many sampled pages llms-full.txt lists.
8
+ - **STANDARD.md:** D4 describes three depths. llms.txt is the map, llms-full.txt is the catalog an agent works from in one fetch (every page, with what it is and the facts an agent needs to choose or use it), and each page's markdown copy has the rest. For an API or docs site, the catalog holds the reference itself; for things people read, watch or listen to, each item's summary, facts and link. The harm on each side is named, with the documented read limits of three agent fetch tools. D4 says llms-full.txt isn't the control for what AI may use (that's D15's `Content-Signal`, or auth), and that llms-full.txt is a convention the llms.txt proposal doesn't define. §16 cites the three tools' limits.
9
+
10
+ From an outside review of 2.4.0:
11
+
12
+ - **The §14 stub links a release tag, not `main`.** Agents read the stub every session and act on it, and a link to `main` gives them whatever was pushed last, before any release, tag or test. The stub says `blob/v[version]/STANDARD.md`, and §14 and docs/adopting.md say to fill in the latest tag and change it after reading the CHANGELOG. The status page that `--matrix` writes links the standard at the tag of the scorecard version that wrote it.
13
+ - **"Last verified" is 2026-10-25,** the newest date a source in §16 was read. A test fails when it's earlier.
14
+ - **The README's sample output is fresh** (animalhouse.ai, 2026-11-04), and a test fails when its version isn't the package's.
15
+ - **"Score a site's readiness",** not "any site's", in the README and package.json, and SECURITY.md says why to audit only sites you own or may test: one run sends dozens to a few hundred requests, several at a time, and POSTs to any MCP endpoint it finds.
16
+
3
17
  ## 2.4.0 (2026-11-02)
4
18
 
5
19
  From a review by the agents that run the scorecard daily for two projects. No scoring changes, and no new warnings or failures.
package/README.md CHANGED
@@ -4,7 +4,7 @@
4
4
  [![npm](https://img.shields.io/npm/v/readiness-audit)](https://www.npmjs.com/package/readiness-audit)
5
5
  [![License: MIT](https://img.shields.io/badge/license-MIT-blue)](LICENSE)
6
6
 
7
- **Score any site's readiness for AI agents and search:** llms.txt, robots.txt rules for AI crawlers, the sitemap, structured data, link previews, OpenAPI, MCP servers, agent skills and the A2A path. A standard built from production request logs, and a command-line audit that checks a live site against it. Read-only, no account, no API key.
7
+ **Score a site's readiness for AI agents and search:** llms.txt, robots.txt rules for AI crawlers, the sitemap, structured data, link previews, OpenAPI, MCP servers, agent skills and the A2A path. A standard built from production request logs, and a command-line audit that checks a live site against it. Read-only, no account, no API key.
8
8
 
9
9
  ## Quick start
10
10
 
@@ -15,21 +15,20 @@ npx readiness-audit example.com
15
15
  It needs Node 22.23.4, 24.22.1 or 26.12.1, or later on the same line (each supported line's current security release), and has no dependencies.
16
16
 
17
17
  ```
18
- Agent and search readiness: animalhouse.ai (readiness-audit v2.3.0)
18
+ Agent and search readiness: animalhouse.ai (readiness-audit v2.5.0, 2026-11-04T02:41Z)
19
19
 
20
20
  Required
21
21
  PASS D1 robots.txt 1 group with consistent rules, Sitemap listed
22
- PASS D4 llms.txt and llms-full.txt llms.txt 16.2 KB, llms-full.txt 65 KB; 10 sampled links resolve
23
- PASS D12 No false declarations no OAuth, A2A, plugin or server-card claims without the service
24
- PASS W1 Titles and descriptions 27 sampled pages (18 static, 3 templates), all with their own
25
- WARN W6 Page structure heading levels skipped on /agent/[name] (h1 to h3)
26
- PASS M4 Hosted MCP edge both eras, JSON for every Accept, browsers redirected
22
+ PASS D4 llms.txt and llms-full.txt llms.txt 16.2 KB, llms-full.txt 65 KB; 10 sampled links resolve without a redirect; llms-full.txt lists the 3 sampled pages
23
+ PASS D12 No false declarations no OAuth, A2A, plugin or server-card claims without the service behind them
24
+ PASS W1 Titles and descriptions 27 sampled pages (18 static, 3 templates), all with their own title and description
25
+ PASS M4 Hosted MCP edge both eras, JSON for every Accept, browsers redirected, aliases redirect
27
26
  ...
28
27
  Recommended
29
28
  PASS D7 JSON 404s and a JSON 405 on POST / JSON 404 for unknown well-known paths, JSON 405 for POST / (Allow: GET, HEAD, OPTIONS)
30
29
  WARN D8 Markdown for agents Vary: Accept missing on the HTML response
31
30
  ...
32
- Score on 2026-10-24: required 17 of 19, recommended 10 of 11, next level 1 of 2
31
+ Score on 2026-11-04: required 19 of 19, recommended 10 of 11, next level 1 of 2
33
32
  ```
34
33
 
35
34
  To switch a project over, give its coding agent one line:
package/STANDARD.md CHANGED
@@ -2,7 +2,7 @@
2
2
 
3
3
  **What it is:** a common standard for making a site findable and usable by AI agents and search engines, built from six projects' own playbooks (animalhouse.ai, inbed.ai, magnifica.family, drifts.bot, achurch.ai, botsmatter.live), checked against the specs ([§16](#16-specs-and-sources)), scored against each live site, and revised after each project's agent reviewed it.
4
4
  **Scorecard:** `npx readiness-audit <domain>` (see the [README](README.md))
5
- **Last verified:** 2026-10-05
5
+ **Last verified:** 2026-10-25
6
6
 
7
7
  > **Start here**
8
8
  > 1. Score your site: `npx readiness-audit <domain>`. It needs Node 22.23.4, 24.22.1 or 26.12.1, or later on the same line, and no dependencies.
@@ -134,7 +134,10 @@ Each item says what it is, why it's at its level, and how you know it's done. Se
134
134
 
135
135
  **D3 Server-rendered pages.** Every page's content is in the HTML the server sends, not loaded later by script, with a real logo through the framework's icon convention and a favicon set (missing icons show up as steady 404s). *Why: crawlers, link-preview bots and agents that fetch pages read the HTML, and many don't run JavaScript.* inbed's chat pages once shipped none of their 48 messages in the HTML. The page-level details (titles, canonicals, structured data, previews, headings, links) are W1 to W7 in [§5](#5-search-people-and-answer-engines).
136
136
 
137
- **D4 `/llms.txt` and `/llms-full.txt`**: a short map (what this is, how to start, links to every other surface) and the whole corpus in one fetch. Generate both and test them against the sitemap: agents follow the links, so every one must open. The scorecard samples them and warns about missing pages (an endpoint that answers 401 for want of a key is fine; templates like `{id}` are skipped). The file is markdown; serve it as `text/markdown` or `text/plain`, and use the same type wherever you link it.
137
+ **D4 `/llms.txt` and `/llms-full.txt`**: three depths. llms.txt is the map: what this is, how to start, links to every other surface. llms-full.txt is the catalog an agent works from in one fetch: every page llms.txt links, with its URL, what it is, and the facts an agent needs to choose it or use it. Each page's markdown copy (D8) has the rest. Generate both files and test them against the sitemap: agents follow the links, so every one must open. The scorecard samples llms.txt's links, warns about missing pages (an endpoint that answers 401 for want of a key is fine; templates like `{id}` are skipped), and warns about sampled pages llms-full.txt doesn't list. Both files are markdown; serve them as `text/markdown` or `text/plain`, and use the same type wherever you link them.
138
+ - **Put in what agents act on.** For an API or docs site, that's the reference itself: an agent writes a call from the parameters, not from a summary of them. For things people read, watch or listen to (articles, episodes, recordings), it's each item's summary, its facts (length, date, credits) and its link, not the body. A test: from this file alone, can an agent answer your site's common questions and pick the right link? *Harm prevented:* a file as thin as llms.txt sends the agent to open every page, and a file of every page's full text gets read in part. Agent fetch tools return the first part of a long file: 5,000 characters by default in the MCP reference fetch server, 20,000 in OpenClaw's `web_fetch`, 100,000 in Claude Code's WebFetch. Some tell the agent where to resume, but each further part is another fetch.
139
+ - **It isn't the control for what AI may use.** Everything in llms-full.txt is already on your pages. To say what AI may do with your content, use D15's `Content-Signal` (`ai-train=no` keeps search and AI answers while declining training), or keep the content behind auth. Leaving a page's body out of llms-full.txt doesn't keep it from crawlers.
140
+ - llms-full.txt is a convention: the llms.txt proposal defines only llms.txt and per-page markdown.
138
141
  - *Why required: agents doing tasks use it, though the AI crawlers mostly don't.* Across 56 domains in a week of October 2026, bots in Cloudflare's legacy AI categories made 49 of 1,406 llms.txt requests (GPTBot 5, ClaudeBot none in the sample) and requested robots.txt and the sitemap instead. On the 16 sites with an API, 13% of the addresses that requested llms.txt also got a successful API answer that week, 9% with a user agent that wasn't a browser's ([research](docs/research/who-reads-discovery-files.md)). In a controlled benchmark across 20 documentation sites, markdown with a linked llms.txt cut agents' wasted fetches from 2.23 per task with HTML to 0.11 ([Mintlify](https://www.mintlify.com/blog/llms-txt-agent-benchmark), 2026-07-17). Google Search ignores llms.txt (it said so in June 2026). At animalhouse, 79 requests in four and a half days came from:
139
142
  - agent directories (AgenstryBot, BrickBlueBot, agentprobe, MCPHarbor, AgentTrustBot);
140
143
  - llms.txt directories and validators;
@@ -799,7 +802,7 @@ Step-by-step instructions for a project's coding agent: [docs/adopting.md](docs/
799
802
  # Agent and Search Readiness
800
803
 
801
804
  This project follows the Agent and Search Readiness Standard:
802
- https://github.com/geeks-accelerator/agent-and-search-readiness/blob/main/STANDARD.md
805
+ https://github.com/geeks-accelerator/agent-and-search-readiness/blob/v[version]/STANDARD.md
803
806
 
804
807
  Score this site: `npx readiness-audit@2 <domain>`. Current status: docs/readiness-status.md
805
808
 
@@ -809,6 +812,7 @@ Step-by-step instructions for a project's coding agent: [docs/adopting.md](docs/
809
812
  Project-specific notes:
810
813
  - (anything the standard doesn't cover)
811
814
  ```
815
+ - **Pin the standard to a release tag.** Put the latest tag from the [releases page](https://github.com/geeks-accelerator/agent-and-search-readiness/releases) in place of `v[version]`. Your agents read the stub every session and act on what it links. A link to `main` gives them whatever was pushed last, before any release, tag or test; a tag gives them the text you reviewed. To take a newer standard, read the CHANGELOG entries since your tag, then change the tag.
812
816
  - **Pin the scorecard to a major version** (`npx readiness-audit@2 <domain>`), so a new release can't fail your build without warning.
813
817
  - **Run it in CI** after each deploy or on a schedule, and fail on a required regression. `examples/github-action.yml` is a starting point.
814
818
  - **Record declined items** (principle 9) in your CLAUDE.md or AGENTS.md, with the reason.
@@ -825,7 +829,7 @@ Level: **R** required where it applies, **Rec** recommended, **N** next level. W
825
829
  | D1 | robots.txt: the same rules for every crawler, Sitemap, side effects disallowed | R | agents, honesty | score |
826
830
  | D2 | sitemap.xml with real lastmod dates | R | traffic | score |
827
831
  | D3 | Server-rendered pages, real logo and favicons | R | agents, discovery | review |
828
- | D4 | llms.txt and llms-full.txt, generated, linked from skills and README | R | traffic | score, test |
832
+ | D4 | llms.txt and llms-full.txt, generated, linked from skills and README; llms-full.txt lists every page llms.txt links | R | traffic | score, test |
829
833
  | D5 | `Link` headers: service-desc, service-doc, describedby, api-catalog | Rec | discovery | score |
830
834
  | D6 | security.txt with Expires once | R | hygiene | score |
831
835
  | D7 | JSON 404s for unknown well-known paths (`agent.json` included), JSON 405 with `Allow` on `POST /` | Rec | agents | score |
@@ -954,6 +958,9 @@ Status as of 2026-10-05. Several are drafts, so re-check before relying on a det
954
958
  | Discovery-file requests ("Who requests agent discovery files?") | this repo's research, 56 domains on Cloudflare, 2026-10-17 to 10-24, corrected 2026-10-24 | docs/research/who-reads-discovery-files.md |
955
959
  | Negotiated markdown requests | GoodBarber's own logs, 2026-09-04 to 09-18, read 2026-10-24 | https://www.goodbarber.com/blog/do-ai-agents-read-markdown-what-two-weeks-of-our-own-server-logs-say-a1622/ |
956
960
  | Markdown and llms.txt in agent tasks | Mintlify benchmark, 2026-07-17, read 2026-10-24 | https://www.mintlify.com/blog/llms-txt-agent-benchmark |
961
+ | MCP reference fetch server's `max_length` (default 5,000 characters) | source, read 2026-10-25 | https://github.com/modelcontextprotocol/servers/blob/main/src/fetch/src/mcp_server_fetch/server.py |
962
+ | OpenClaw `web_fetch` limits (`maxChars` default 20,000) | docs and source, read 2026-10-25 | https://docs.openclaw.ai/tools/web-fetch |
963
+ | Claude Code WebFetch (100,000 characters per call) | Claude Code docs, read 2026-10-25 | https://code.claude.com/docs/en/tools-reference |
957
964
  | Crawler impersonation ("Bots in AI clothing") | agent-security-pranks research, 56 domains on Cloudflare, 2026-10-16 to 10-23, read 2026-10-24 | https://github.com/geeks-accelerator/agent-security-pranks/blob/main/docs/research/bots-in-ai-clothing.md |
958
965
  | 405 and `Allow` | RFC 9110 §15.5.6 | https://www.rfc-editor.org/rfc/rfc9110.html#section-15.5.6 |
959
966
  | Headings | W3C WAI tutorial, read 2026-10-21 | https://www.w3.org/WAI/tutorials/page-structure/headings/ |
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "readiness-audit",
3
- "version": "2.4.0",
4
- "description": "Score any site's readiness for AI agents and search (GEO, AEO, AI SEO): llms.txt, robots.txt AI crawler rules, structured data, link previews, OpenAPI, MCP, agent skills and A2A. Read-only CLI, no account.",
3
+ "version": "2.5.0",
4
+ "description": "Score a site's readiness for AI agents and search (GEO, AEO, AI SEO): llms.txt, robots.txt AI crawler rules, structured data, link previews, OpenAPI, MCP, agent skills and A2A. Read-only CLI, no account.",
5
5
  "type": "module",
6
6
  "bin": {
7
7
  "readiness-audit": "bin/readiness-audit.mjs"
package/src/audit.mjs CHANGED
@@ -11,7 +11,7 @@ import { isIP } from 'node:net';
11
11
  import { request } from './http.mjs';
12
12
  import { SKILL_NAME, sampleSkillEntries, skillNameProblem, cdnCacheNote, repeatsSummary, hasVariantDescription, spread, samplePages, sharedCardText, API_LINK, MCP_LINK, linkRels, apiPresence, DISCOVERY_FILES, AI_BOTS, missingSections, richResultPages, list, norm, pathDepth, decodeText, headings, headingSkips, sitemapEntries, llmsLinks, baseMapper, formatRecorded } from './helpers.mjs';
13
13
 
14
- export const VERSION = '2.4.0';
14
+ export const VERSION = '2.5.0';
15
15
  export const REPO = 'https://github.com/geeks-accelerator/agent-and-search-readiness';
16
16
  const UA = `readiness-audit/${VERSION} (+${REPO})`;
17
17
  const AID_PROTOCOLS = ['mcp', 'a2a', 'openapi', 'grpc', 'graphql', 'websocket', 'local', 'zeroconf', 'ucp'];
@@ -272,12 +272,22 @@ export async function audit(domain, { mcpPath = '/mcp', base = null, concurrency
272
272
  const linkProblems = [];
273
273
  if (broken.length) linkProblems.push(`broken links in llms.txt: ${list(broken)} (${broken.length} of ${sample.length} sampled)`);
274
274
  if (redirected.length) linkProblems.push(`llms.txt links that redirect (link the destination): ${list(redirected)}`);
275
+ // llms-full.txt is the catalog: every page llms.txt links appears in it, so an agent can choose from one fetch.
276
+ // Pages are the sampled links that answer HTML, besides the homepage; files like /openapi.json are surfaces, not pages.
277
+ let pagesListed = 0;
278
+ if (ok(llmsFull) && !llmsFull.tooBig) {
279
+ const covered = new Set(llmsLinks(llmsFull.text, `${BASE}/llms-full.txt`, siteHosts).map(norm));
280
+ const pagesLinked = answers.filter(({ u, r }) => ok(r) && /text\/html/.test(r.type) && norm(r.url) === norm(u) && new URL(u).pathname !== '/');
281
+ const uncovered = pagesLinked.filter(({ u }) => !covered.has(norm(u))).map(({ u }) => u.replace(BASE, '') || '/');
282
+ if (uncovered.length) linkProblems.push(`pages llms.txt links that llms-full.txt doesn't list: ${list(uncovered)} (${uncovered.length} of ${pagesLinked.length} sampled pages)`);
283
+ else pagesListed = pagesLinked.length;
284
+ }
275
285
  const kb = (r, digits) => (r.tooBig ? r.error.replace(/^.*?over /, 'over ') : `${(r.text.length / 1024).toFixed(digits)} KB`);
276
286
  const sizes = `llms.txt ${kb(llms, 1)}, llms-full.txt ${kb(llmsFull, 0)}`;
277
287
  if (!ok(llms) || !/^\s*# \S/.test(llms.text)) miss(id, name, level, (ok(llms) ? 'llms.txt has no H1 title on its first line' : `/llms.txt answered ${llms.status}`) + cached(llms));
278
288
  else if (!ok(llmsFull)) miss(id, name, level, `/llms-full.txt missing${cached(llmsFull)}`);
279
289
  else if (linkProblems.length) record(id, name, level, 'warn', `${linkProblems.join('; ')}${cached(llms)}`);
280
- else pass(id, name, level, `${sizes}${sample.length ? `; ${sample.length} sampled link${sample.length === 1 ? '' : 's'} resolve without a redirect` : ''}`);
290
+ else pass(id, name, level, `${sizes}${sample.length ? `; ${sample.length} sampled link${sample.length === 1 ? '' : 's'} resolve without a redirect` : ''}${pagesListed ? `; llms-full.txt lists the ${pagesListed} sampled page${pagesListed === 1 ? '' : 's'}` : ''}`);
281
291
  }
282
292
 
283
293
  // ---- D5 Link headers (recommended; a wrong service-desc is a false declaration) ----
@@ -888,7 +898,7 @@ export function matrix(reports, command, recorded = {}) {
888
898
  const ids = reports[0].results.map((r) => [r.id, r.name, r.level]);
889
899
  const out = [
890
900
  '# Agent and Search Readiness Status', '',
891
- `Generated ${new Date().toISOString().slice(0, 10)} by \`${command}\` (readiness-audit v${VERSION}). Don't edit it by hand: regenerate it. What each ID means: [the standard](${REPO}/blob/main/STANDARD.md).`, '',
901
+ `Generated ${new Date().toISOString().slice(0, 10)} by \`${command}\` (readiness-audit v${VERSION}). Don't edit it by hand: regenerate it. What each ID means: [the standard](${REPO}/blob/v${VERSION}/STANDARD.md).`, '',
892
902
  'The scorecard sees only what is visible from outside a site. Items checked in review and tests (listed at the end) aren\'t in these numbers, so a high score can hide real gaps.', '',
893
903
  '✓ pass, ✗ fail, ! warning, – not applicable.', '',
894
904
  `| ID | Check | Level | ${domains.join(' | ')} |`, `|---|---|---|${domains.map(() => '---').join('|')}|`,