Key takeaways
- Coding agents like Claude Code and Codex now pick tools, SDKs and APIs for developers. Most AI visibility tools were built to track ChatGPT, Gemini and Perplexity, not these agents.
- In our research I found no source confirming that Profound tracks Claude Code or Codex. Verify that on Profound's own site or in a sales call before you commit.
- Of the tools researched, only Promptwatch is documented as tracking Claude Code and Codex on every plan. Gauge is a narrower vendor aimed at the same niche.
- Tracking mentions is not enough. Studies of coding agents show that being mentioned and being selected are very different outcomes.
- Pricing across the category varies widely and is reported inconsistently, so confirm every number on the vendor's pricing page.
Why developer-tool companies need a different kind of tracking
If you sell an API, a database, a hosting platform or an open-source package, a growing share of your "buyers" are not typing into a search box. They are asking Claude Code or Codex to scaffold a project, and the agent decides which payment provider, auth library or deployment target goes into the plan.
That changes what you need to measure. Classic AI visibility questions are "Was my brand mentioned in ChatGPT?" The coding-agent version is harder:
- Is my product in the agent's plan?
- Is it the default choice?
- Does the agent build the thing itself instead of buying from me?
- How does the agent describe me when it does pick me?
Those four questions come from Promptwatch's own write-up on coding-agent visibility, and they are a good checklist for evaluating any vendor, including ones that are not Promptwatch.
What the research says about how agents pick tools
Two independent studies are worth reading before you shop for software.
Amplifying ran 2,430 open-ended prompts through Claude Code across 20 categories and 3 Claude models, without naming any tool. In 12 of the 20 categories, the most common pick was a custom-built solution, not a third-party product. Where agents did pick a product, defaults were strong: GitHub Actions took 93.8% of CI/CD picks, Stripe 91.4% of payments and Vercel 100% of JavaScript deployment.
Armature published a study on September 3, 2026 that adds more detail. Claude Code built in-house in 19% of sessions, while Codex and Cursor did so in 10%. The three agents chose the same tool in only 42% of cases. Codex used web search in 94% of sessions, versus roughly 30% for Claude Code overall. And the mention-versus-selection gap is brutal: PayPal was mentioned 139 times and never selected, and LangChain was mentioned 194 times but selected 4 times.
The takeaway for buyers is simple. A tool that only counts brand mentions in chat answers will miss both build-versus-buy and mention-versus-selection. Ask any vendor how they would show you either.
Where Profound fits
Profound is an enterprise AI visibility platform. Its differentiators are Prompt Volumes, agents, crawler analytics, shopping intelligence and enterprise controls. It is a serious product, and for large brands tracking ChatGPT, Gemini and AI Overviews it is a reasonable shortlist item.
For developer-tool companies the open question is coverage. According to a competitor's blog (ZeroRank, October 2026, so treat it with caution), Profound's brand trial runs 50 prompts a day across ChatGPT, Gemini and Google AI Overviews, and Brand Enterprise is custom-quoted with up to 9 answer engines. Claude and Copilot coverage should be confirmed in a quote. The same source puts Agency Growth at $99 a month plus $399 per client workspace, covering ChatGPT, Perplexity and Google AI Overviews.
I could not find any source saying Profound tracks Claude Code or Codex. That does not prove it doesn't, but you should not assume it does. Ask directly, and ask to see a sample report that shows an agent's actual plan output.
The alternatives worth a look
Promptwatch
Promptwatch is the one platform in my research with documented coding-agent tracking. It tracks Claude Code and Codex on every plan, using the same visibility, mention, sentiment and competitor metrics it applies to ChatGPT, Gemini and Perplexity. It also covers OpenCode. Most coding-agent prompts map to the Transactional intent type, which makes sense: the agent is about to pick something and use it.
A few things matter specifically for developer-tool companies:
- It monitors the real user interfaces of these tools, not just API outputs, and answers can differ between the two.
- IP-verified AI crawler logs show whether GPTBot, ClaudeBot, PerplexityBot and others can actually read your docs, and which pages they fetch. If your docs return errors to bots, no amount of monitoring fixes that.
- The Content Agent and Action Board turn gaps into work: new docs pages, comparison pages and integration guides, published to Webflow, Framer or WordPress.
- Visitor analytics attribute actual traffic and conversions from AI platforms, so you can connect this to signups rather than mentions.
- A hosted MCP server and REST API let you pull data into Claude Code, Cursor or your own dashboards, which suits engineering-led teams.
One limit to know: on Brands plans each project actively tracks 4 models at a time, picked from the full list. If you want Claude Code, Codex, ChatGPT and Perplexity running together, that uses your four. Pricing starts at $95 a month (Essential) and $245 (Professional), with Business at $579 a month, which adds the automated Content Agent. Confirm current numbers at promptwatch.com/pricing.
Promptwatch's own data is relevant here too. In its software-prompt report (Sep 28 to Oct 4, 2026), GitHub was the most cited software domain in ChatGPT at 1.2% of software citations, followed by learn.microsoft.com at 0.98% and G2 at 0.73%. Documentation and help centers earned the largest share by source type at 2.64%, and vendor sites 2.01%. Vendor sites and docs together earn about three times what review sites and analysts do. See Promptwatch Data on software citations in ChatGPT. The practical advice: publish crawlable docs, keep a public GitHub presence, and keep G2 and Capterra profiles current.
Gauge
Gauge is positioned squarely at this niche. Its site says Gauge Agents helps developer tools get selected and implemented by coding agents like Claude Code and Codex, and it talks about "agent-led growth" with mention rate, rankings and citation rates. I found no pricing and did not verify how deep the coverage is, so treat it as a lead to investigate rather than a recommendation.
Scrunch
Scrunch covers how AI agents read your site, which is relevant to docs-heavy companies. Core is reported at $250 a month for 125 prompts across 4 engines. I could not confirm Claude Code or Codex tracking, so ask.
Peec AI and Otterly.AI
Both are straightforward prompt trackers with lower price points. Peec AI is reported at around $80 to $95 for 50 prompts at the entry level, and sources disagree, so check peec.ai/pricing. Otterly.AI starts at $29 a month for 15 prompts, rising to $189 and $489. They are fine for brand mention tracking on mainstream assistants. They do not, in anything I found, cover coding agents, crawler logs or content workflows.
AthenaHQ, Goodie AI and Writesonic
AthenaHQ lists a free tier with $25 in credits, then Starter at $295 a month. Goodie AI is reported at $295 and $495 a month billed annually. Writesonic starts at $79 a month annually. All three are GEO-style platforms aimed at marketing teams. None of my sources confirmed coding-agent coverage.
Semrush and Ahrefs Brand Radar
If you already live in an SEO suite, the add-ons are tempting. Semrush's AI visibility plan is reported at $165.17 a month annually with 50 daily prompts, and Ahrefs Brand Radar starts at $50 a month for custom prompts, with the full-model tier at $699. They are a good fit for tracking brand presence in mainstream AI search next to your keyword data. They are a weak fit for the question "what does Claude Code put in its plan?"
Comparison table
Coding-agent coverage is based on what I could verify. "Unverified" means no source I found confirmed it, not that the tool lacks it.
| Tool | Claude Code / Codex tracking | Entry price (reported) | Crawler logs | Content workflows | Best for |
|---|---|---|---|---|---|
| Promptwatch | Yes, documented on every plan | $95/mo Essential | Yes, IP-verified | Yes, Content Agent and Action Board | Dev-tool companies that want tracking plus fixes |
| Profound | Unverified | Custom (Enterprise) | Yes | Agents | Large brands on mainstream AI engines |
| Gauge | Positioned for it, depth unverified | Not found | Not verified | Not verified | Teams exploring agent-led growth |
| Scrunch | Unverified | $250/mo Core | Agent-readability focus | Limited | Docs-heavy sites |
| Peec AI | Unverified | Roughly $80-95/mo | No | No | Simple prompt tracking |
| Otterly.AI | Unverified | $29/mo Lite | No | No | Small budgets |
| AthenaHQ | Unverified | $295/mo Starter | Not verified | Some | Marketing-led GEO |
| Semrush AI Visibility | Unverified | $165.17/mo annual | No | SEO suite | Existing Semrush users |
| Ahrefs Brand Radar | Unverified | $50/mo custom prompts | No | No | Existing Ahrefs users |
Prices come from vendor-run comparison posts (Frase, ZeroRank, Tim Soulo's list) and differ between sources. Prompt and engine counts vary so much that price per prompt can mislead, and some tools charge extra per model. Always confirm on the vendor's pricing page.
How to evaluate any vendor for coding-agent visibility
Run this checklist in the demo. It takes twenty minutes and filters out most of the field.
- Ask which coding agents are tracked, and whether they are tracked through the real interface or an API model that approximates it.
- Ask to see a prompt like "set up payments for a Next.js app" with the full agent output, including whether your product appears in the plan, not just in prose.
- Ask how they separate mention from selection. PayPal-style results (139 mentions, zero selections) are the failure mode you are trying to catch.
- Ask whether they can show build-versus-buy. If the agent writes its own solution, you want to know that, because it means your competition is a code snippet.
- Ask about crawler logs. Check that bots can reach your docs. Promptwatch's crawler traffic data (June 8 to September 6, 2026) shows OpenAI at 79.8% of IP-verified AI crawler requests in the week of Aug 31 to Sep 6, down from 94.8% in early June. If a provider is large in that chart but near zero in your logs, you are probably blocked by robots.txt, a CDN or a WAF rule. See Promptwatch Data on AI crawler traffic.
- Ask how many models run at once on your plan, since some tools cap this.
What to actually do with the data
Tracking is only useful if something changes afterward. A few moves that follow from the research:
Fix crawlability before anything else
Agents that use web search, and Codex did in 94% of sessions in Armature's study, can only recommend what they can read. Make docs crawlable, avoid gating quickstarts, and check your CDN rules.
Write for the queries agents run
Promptwatch's query fanout data shows ChatGPT averaged 2.15 searches per response in early December and about 1.84 by early March, with query length falling from about 117 characters to the high 80s. The report's advice is to write headings that read like short search queries and cover sub-questions like "X vs Y", pricing, alternatives and how-tos. One caveat: that page is dated March 1 but includes April data, so check the date range before citing it. More in Promptwatch's ChatGPT query fanout report.
Win the default slot
The Amplifying numbers show defaults are close to winner-take-all in some categories. If your category has a default, your job is to become the alternative the agent reaches for when the default does not fit, with clear integration pages and comparison content.
Keep third-party profiles current
GitHub, G2 and Capterra show up in software citations. A stale G2 profile is a cheap thing to fix.
A note on agencies
If you would rather hand this to specialists, an SEO and GEO agency can run the monitoring and the content work together. 1001 SEO Media is the agency behind this site and works with Promptwatch for AI visibility, though it is worth interviewing several firms and asking about their experience with developer audiences.
Where to find more tools
This category changes monthly. For a wider list of options, the GEO software directory at bestgeosoftware.com is a useful place to browse, and Promptwatch Data publishes ongoing research on how AI systems cite sources.
Bottom line
If your buyers are developers, the agent is now part of the funnel. Profound is a capable enterprise platform, but I could not verify that it tracks Claude Code or Codex, so make that the first question you ask. Of the options researched, Promptwatch is the one with documented coverage, crawler logs and content tooling in one place, and Gauge is worth a call if you want a niche specialist. Start with a free baseline: run twenty realistic agent prompts for your category, note whether you are in the plan, the default, or replaced by custom code, then pick the tool that can show you the same thing at scale.