H
harrjdk/truth-seeker
MCP server for LLM-driven analysis of compiled Java apps (legacy Spring / Jakarta EE / J2EE): truth tables,…
updated 2026-11-17
View on GitHub ↗Link checked 2026-11-17 · HTTP 200 · opens the vendor's own site
- Star count
- 0 ★ · harrjdk/truth-seeker · read from the GitHub search API · (checked 2026-11-19 20:00:48)
- Source
- read directly from its GitHub page · type: open-source project
- Harvest history
- first seen 2026-11-16 · last seen 2026-11-20 · seen 5× · listed by 2 sources
- Score
- score 203 · 2 sources · repo
FAQ (answers come from our records)
Who is it for?
We filed it under these tags: Open Source. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository
harrjdk/truth-seeker;first catalogued 2026-11-16;last seen 2026-11-20。There is no official certification or quality rating here — we record sources and dates only.README excerpt (English, untranslated) · Read the full README on GitHub →
# truth-seeker
Excerpted from the repository README (abridged)
An MCP server that lets an LLM client (Claude Code, Claude Desktop, Codex, or a local model with MCP support)
analyze compiled Java applications — especially legacy Spring, Jakarta EE / Java EE and J2EE apps — and produce
truth tables, state machine diagrams, call graphs and unit tests straight from bytecode.
Works on .jar, .war, .ear, Spring Boot fat jars and target/classes directories, including nested jars.
Never loads or runs the target application. Classes are parsed with ASM; static initializers don't execute.
No source code needed. Debug info (-g) improves names but isn't required.
The server runs on Java 17+. It reads bytecode from Java 1.1 through 25, including javac and Eclipse (ECJ) output.
## Tools
| Tool | What it gives the model |
| scan_archive | Inventory: bytecode levels, bundled libraries, XML descriptors, stereotypes (Spring, EJB 2.x/3.x, Struts 1/2, servlets, JPA, MDBs, Quartz, Spring Batch…) and complexity hot spots |
| list_endpoints | HTTP routes (Spring MVC, JAX-RS, @WebServlet, web.xml, struts-config.xml), JMS/Kafka/Rabbit listeners, MDBs, @Scheduled, EJB timers, SOAP |
| describe_class | Kind, annotations, injected collaborators (field / constructor / setter (XML-wired) injection), fields, methods with cyclomatic complexity |
| disassemble | Raw bytecode for a class or one method, as ground truth |
| call_graph | Callers/callees as a Mermaid flowchart. Interface calls are expanded to the implementations in the archive |
| truth_table | A method's decisions and the simplified boolean logic per outcome. The summary (default) gives one row per outcome, with path counts and side effects listed once. detail: "full" gives one row per path, and expand adds the 2ⁿ table |
| state_machine | Mermaid / PlantUML state diagrams inferred from enum state. Also reports derived state (a status computed from other facts, e.g. from the last event). Flags unguarded transitions, never-entered or never-produced states, and state-guard rejections |
| generate_unit_tests | JUnit 5 + Mockito test classes (one class or classNames): one test per path, with solved inputs, stubs, assertions and verify() calls. Tests whose preconditions can't be arranged are @Disabled with exact TODOs. With outputDir, it writes the files and returns only a compact summary |
| verify_tests | Compiles test sources against the archive and runs them on JUnit 5 in a separate JVM. Reports passed / failed / disabled, with failures grouped by cause and disabled tests grouped by unresolved precondition. With mutate: true it also runs mutation testing (below). JUnit and Mockito are bundled, so it works offline. Runs the application's code |
All tools take path. The archive index is cached per path, so follow-up calls are fast.
Outputs are sized for an LLM's context: Markdown summaries by default, with full detail on request.
format: "json" (truth_table, state_machine), generate_unit_tests and verify_tests also return MCP structuredContent.
All tools except generate_unit_tests (writes files) and verify_tests (runs code) are annotated read-only, so clients can auto-approve them.
### The generate → verify loop
text
generate_unit_tests path=app.war classNames=[...] outputDir=build/gen-tests
analyze compiled Java applications — especially legacy Spring, Jakarta EE / Java EE and J2EE apps — and produce
truth tables, state machine diagrams, call graphs and unit tests straight from bytecode.
Works on .jar, .war, .ear, Spring Boot fat jars and target/classes directories, including nested jars.
Never loads or runs the target application. Classes are parsed with ASM; static initializers don't execute.
No source code needed. Debug info (-g) improves names but isn't required.
The server runs on Java 17+. It reads bytecode from Java 1.1 through 25, including javac and Eclipse (ECJ) output.
## Tools
| Tool | What it gives the model |
| scan_archive | Inventory: bytecode levels, bundled libraries, XML descriptors, stereotypes (Spring, EJB 2.x/3.x, Struts 1/2, servlets, JPA, MDBs, Quartz, Spring Batch…) and complexity hot spots |
| list_endpoints | HTTP routes (Spring MVC, JAX-RS, @WebServlet, web.xml, struts-config.xml), JMS/Kafka/Rabbit listeners, MDBs, @Scheduled, EJB timers, SOAP |
| describe_class | Kind, annotations, injected collaborators (field / constructor / setter (XML-wired) injection), fields, methods with cyclomatic complexity |
| disassemble | Raw bytecode for a class or one method, as ground truth |
| call_graph | Callers/callees as a Mermaid flowchart. Interface calls are expanded to the implementations in the archive |
| truth_table | A method's decisions and the simplified boolean logic per outcome. The summary (default) gives one row per outcome, with path counts and side effects listed once. detail: "full" gives one row per path, and expand adds the 2ⁿ table |
| state_machine | Mermaid / PlantUML state diagrams inferred from enum state. Also reports derived state (a status computed from other facts, e.g. from the last event). Flags unguarded transitions, never-entered or never-produced states, and state-guard rejections |
| generate_unit_tests | JUnit 5 + Mockito test classes (one class or classNames): one test per path, with solved inputs, stubs, assertions and verify() calls. Tests whose preconditions can't be arranged are @Disabled with exact TODOs. With outputDir, it writes the files and returns only a compact summary |
| verify_tests | Compiles test sources against the archive and runs them on JUnit 5 in a separate JVM. Reports passed / failed / disabled, with failures grouped by cause and disabled tests grouped by unresolved precondition. With mutate: true it also runs mutation testing (below). JUnit and Mockito are bundled, so it works offline. Runs the application's code |
All tools take path. The archive index is cached per path, so follow-up calls are fast.
Outputs are sized for an LLM's context: Markdown summaries by default, with full detail on request.
format: "json" (truth_table, state_machine), generate_unit_tests and verify_tests also return MCP structuredContent.
All tools except generate_unit_tests (writes files) and verify_tests (runs code) are annotated read-only, so clients can auto-approve them.
### The generate → verify loop
text
generate_unit_tests path=app.war classNames=[...] outputDir=build/gen-tests
Listed badge (optional)
Copy this line into the README if you want the listing mark. Optional, and it does not affect being listed.
[](https://anclum.com/en/s/harrjdk-truth-seeker/)
- Source
- github · https://github.com/harrjdk/truth-seeker
- Browse by field
- Open source
- Listed by
- 2 independent sources
- added
- today
More entries
Filed under
Every tag opens the full list for that filter. Counts are the live catalogue figures.