HN
Today

Launch HN: Kampala (YC W26) – Reverse-Engineer Apps into APIs

Kampala, a new Launch HN, offers a man-in-the-middle proxy that turns complex application workflows into reliable APIs, sidestepping the flakiness of traditional browser automation. Born from the founders' struggles with integrating legacy dental systems, it preserves HTTP/TLS fingerprints to appear legitimate, offering a robust solution for agentic reverse engineering. This technical utility resonates deeply with developers on HN, who frequently encounter the pain points of interacting with API-less systems.

22
Score
16
Comments
#4
Highest Rank
5h
on Front Page
First Seen
Apr 16, 3:00 PM
Last Seen
Apr 16, 7:00 PM
Rank Over Time
65457

The Lowdown

Kampala, a new offering from Zatanna, introduces a novel approach to automating interactions with websites, mobile, and desktop applications by reverse-engineering them into stable APIs. Co-founder Alex Blackwell explains that existing automation methods, like browser automation or computer use agents, are often brittle, slow, and non-deterministic, especially when dealing with legacy dashboards or systems lacking public APIs. Kampala was developed out of necessity while the founders were building integrations for dental tech, facing numerous insurance payer and practice management systems.

Key aspects of Kampala include:

  • MITM Proxy: Operates as a man-in-the-middle proxy to intercept and analyze network traffic.
  • Robustness: Designed to overcome limitations of other MITM tools by not manipulating TLS/HTTP2 fingerprints, thus evading anti-bot detection. It also handles session tokens and anti-bot cookies effectively.
  • Workflow Creation: Allows users to create scripts/APIs either through an agent harness that prompts for actions or by manually performing a workflow once and having an AI agent replicate it.
  • Export and Hosting: Generated APIs/scripts can be exported, run locally, or hosted by Zatanna.
  • Underlying Principle: Focuses on automating at the "requests layer" – the language computers truly understand – rather than relying on visual representations like screenshots for LLMs.

Kampala aims to provide a reliable, deterministic solution for automating tasks that traditionally involve navigating complex, API-deficient interfaces, offering a significant improvement over current brittle automation practices.

The Gossip

Code-Crafted Connectivity

Many commenters shared their own DIY solutions for similar problems, often leveraging tools like Claude or Playwright to analyze HAR files or intercept traffic and generate API documentation or scripts. This highlighted a community desire for such tools and demonstrated varying approaches, from fully AI-driven analysis to custom Python scripts for authentication handling, with the author engaging by suggesting how Kampala could enhance these methods.

Proxying Ponderings & Technical Nuances

Users questioned Kampala's differentiation from established tools like Charles Proxy or Proxyman. A co-founder of Caido highlighted the significant challenges of nailing authentication and maintaining perfect HTTP/TLS fingerprinting, especially against sophisticated anti-bot measures. The author, Alex Blackwell, elaborated on Kampala's strategies, including using specific TLS libraries and acknowledging current limitations with HTTP/3 and TCP fingerprinting.

Naming Nomenclature Narratives

Several commenters expressed curiosity and amusement regarding the unique names "Kampala" and "Zatanna." The author clarified that "Zatanna" came from a DC comic character, while "Kampala" was a randomly generated workspace name that stuck, adding a lighthearted personal touch to the brand story.

Practical Queries & Use Cases

Users posed practical questions regarding specific challenges, such as handling session re-authentication mid-script, and noted potential uses like dataset creation. The author also engaged with comments discussing the potential for importing HAR files into Kampala and its capabilities for A/B testing payloads, suggesting a broader utility for integration and data collection.