Add essay: Every tool ships an agent now#15
Conversation
Compares five vendor approaches to AI agents: Sentry's Junior (open-source runtime), PostHog (narrow in-app assistant), CodeRabbit (expanding from review gate), Notion (workspace as coordination hub), and phase-specific agents (PDERO lifecycle framework). Includes four custom SVG figures, card illustration, comparison table, embedded tweet, and SEO/GEO optimizations (138-char meta description, Article + BreadcrumbList schema, llms.txt entry, statistics for factual density). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (6)
📝 WalkthroughWalkthroughThis pull request adds a new blog post exploring agent architecture paradigms—vendor-scoped versus open/composable runtimes—supported by four custom SVG figure components that illustrate key architectural concepts, a home page card visualization, and content indexing updates. ChangesAgent Architecture Blog Post and Visualizations
🎯 2 (Simple) | ⏱️ ~12 minutes
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
|
|
||
| Cramer's tweet frames this as a binary: vendor chatbots are broken, generalized agents win. The implementations tell a more nuanced story. PostHog found that narrow scope produces better answers. Notion found that being the coordination layer matters more than being the agent itself. CodeRabbit found that one strong integration point can expand into many. Junior found that the runtime should be open and composable. And phase-specific agents suggest the organizing principle might not be tools at all, but the lifecycle stages that every engineering task passes through. | ||
|
|
||
| What every approach found is that credential isolation, plugin systems, MCP interop, and execution sandboxes consume more engineering effort than prompting. The hard work isn't making the LLM smarter. It's giving the right agent the right access, to the right data, with the right permissions, at the right moment in the lifecycle. |
There was a problem hiding this comment.
🔴 Essay closing line uses banned "not X. It's Y" inversion and grand-closer pattern
The essay's final sentence (content/posts/every-tool-ships-an-agent.mdx:140) is: "The hard work isn't making the LLM smarter. It's giving the right agent the right access, to the right data, with the right permissions, at the right moment in the lifecycle."
This violates two explicit CLAUDE.md writing rules simultaneously:
- "This is not X. It is Y." inversions — listed as the first LLM tell to avoid in the Writing Voice section.
- "Do not write grand closing lines that could be taglines" — the rule says to "End pieces with a specific observation, a pointer to what comes next, or nothing at all."
The sentence also fits the "dramatic reveal closers like 'the bottleneck was never X'" pattern that CLAUDE.md explicitly bans.
Prompt for agents
The last line of the essay at content/posts/every-tool-ships-an-agent.mdx:140 uses a 'not X, it's Y' inversion pattern and functions as a grand tagline closer, both explicitly banned by CLAUDE.md's Writing Voice rules. The sentence is: 'The hard work isn't making the LLM smarter. It's giving the right agent the right access, to the right data, with the right permissions, at the right moment in the lifecycle.' Replace this with a specific observation or a pointer to what comes next. CLAUDE.md says: 'End pieces with a specific observation, a pointer to what comes next, or nothing at all.' One approach: state a concrete next step or unanswered question, e.g. referencing how MCP convergence might reshape these architectures, or pointing the reader toward a specific resource. Another approach: just end the essay on the preceding sentence ('...consume more engineering effort than prompting.') and cut the grand closer entirely.
Was this helpful? React with 👍 or 👎 to provide feedback.
|
|
||
| "Vendor-specific chatbots are broken by design," he wrote. The Sentry agent, the Linear agent, any agent scoped to a single product. Fine for point queries. Nice to get started with. But agents with generalized access outperform them "in every single" use case that crosses tool boundaries. | ||
|
|
||
| Then he open-sourced [Junior](https://github.com/getsentry/junior), an agent runtime Sentry's team built internally. Seven plugin packages, a sandboxed execution environment, and an egress proxy for credential isolation. It's not a Sentry chatbot. It's a framework for building a single agent that talks to all your tools at once. |
There was a problem hiding this comment.
🔴 "It's not X. It's Y" inversion pattern in early paragraph
Line 16 contains: "It's not a Sentry chatbot. It's a framework for building a single agent that talks to all your tools at once." This is a textbook instance of the "This is not X. It is Y." inversion pattern that CLAUDE.md's Writing Voice section lists as the first example of an LLM-sounding sentence structure to avoid. The rule says: "If a sentence sounds like it could appear in any AI-generated essay on any topic, rewrite it."
Prompt for agents
Line 16 of content/posts/every-tool-ships-an-agent.mdx has: 'It's not a Sentry chatbot. It's a framework for building a single agent that talks to all your tools at once.' This matches the 'This is not X. It is Y.' inversion banned by CLAUDE.md. Rewrite to convey the same information without the dramatic negation-then-assertion structure. For example, you could restructure as a single sentence that describes what Junior actually is, or integrate the contrast into the preceding sentence about what was open-sourced. The key is to avoid the two-sentence 'not X / is Y' pivot.
Was this helpful? React with 👍 or 👎 to provide feedback.
|
|
||
| ## Where the approaches converge | ||
|
|
||
| Five patterns, one shared realization: the model matters less than the plumbing. |
There was a problem hiding this comment.
🔴 Section opener uses enumeration-then-verdict / rehearsed rhetorical pivot
Line 116 reads: "Five patterns, one shared realization: the model matters less than the plumbing." This is a neat rhetorical pivot that feels rehearsed — one of the LLM-sounding structures CLAUDE.md's Writing Voice section explicitly flags. The "[count], one [dramatic noun]: [pithy conclusion]" formula could appear in any AI-generated essay. CLAUDE.md says: "Write like a person drafting a blog post, not a model producing content."
Prompt for agents
Line 116 of content/posts/every-tool-ships-an-agent.mdx reads: 'Five patterns, one shared realization: the model matters less than the plumbing.' This is a neat rhetorical pivot flagged by CLAUDE.md as an LLM tell. Rewrite the section opener to be less formulaic. For example, you could state the convergence observation directly without the dramatic count-then-reveal structure, or simply let the comparison table below speak for itself with a more conversational lead-in.
Was this helpful? React with 👍 or 👎 to provide feedback.
1 participant
Summary
SEO/GEO audit
Notion
Test plan
🤖 Generated with Claude Code