Skip to content
View hoklims's full-sized avatar
14 followers · 9 following

Achievements

Achievement: Pull Sharkx3Achievement: Pair Extraordinairex3Achievement: YOLOAchievement: Quickdraw

Achievements

Achievement: Pull Sharkx3Achievement: Pair Extraordinairex3Achievement: YOLOAchievement: Quickdraw

Block or report hoklims

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
hoklims/README.md

Guillaume Fossier

Software engineer · Reliable AI-assisted development

AI coding agents can write the code, the tests and the explanation. If all three share the same wrong assumption, passing tests can create false confidence.

I build tools to record decisions before outcomes are known, trace what code changes put at risk, and check that regression tests distinguish the intended bug from harmless changes.

Selected work

  • AssertLedger — Check whether a regression test actually catches its target bug, with replayable evidence.
  • semctx — Trace a Git diff to affected symbols, contracts, invariants and tests.
  • StacksFinder MCP — Bring deterministic tech-stack recommendations to AI assistants through MCP.
  • Material Workbench — Generate reproducible materials and terrain tiles from recipes, through a CLI, desktop app or MCP.

Research & methods

  • Latent Compass — Record coding-agent decisions and evaluate policies offline, with explicit limits on what the evidence establishes.
  • Top-down Diagnosis — A bilingual guide and reusable templates for tracing software problems from intent to implementation.

I care about reproducible results, inspectable evidence and clearly stated uncertainty.

Get in touch

Pinned Loading

  1. assertledger assertledger Public

    Verify that regression tests catch the bug they claim to prevent. Deterministic, replayable evidence for agent-assisted development.

    TypeScript 2

  2. semctx semctx Public

    Local-first change-impact analysis: map Git diffs to symbols, contracts, invariants and tests. Deterministic PASS/WARN/BLOCK verdicts; not a code-search tool.

    TypeScript 3 2

  3. stacksfinder-mcp stacksfinder-mcp Public

    MCP server for StacksFinder - deterministic tech stack recommendations for LLM clients

    TypeScript 9 3

  4. material-workbench material-workbench Public

    Recipe-driven, deterministic procedural materials and isometric terrain tiles for game engines. Python engine + Tauri desktop app + MCP server.

    Python 1

  5. latent-compass latent-compass Public

    Local decision recorder and offline policy evaluation for coding agents. Advisory tooling with explicit evidence and authority boundaries.

    Python 2 1

  6. top-down-diagnosis top-down-diagnosis Public

    Portable top-down diagnosis for coding agents, with a bilingual guide and reusable context templates.

    HTML