Skip to content

Reveng 0.2.0

Latest
Latest

Choose a tag to compare

View all tags
@marc0der marc0der released this 22 May 11:38
· 2 commits to main since this release
v0.2.0
This tag was signed with the committer’s verified signature.
marc0der Marco Vermeulen
GPG key ID: 47F7ABD6F9FBD428
Verified
Learn about vigilant mode.
7e47d7a
This commit was signed with the committer’s verified signature.
marc0der Marco Vermeulen
GPG key ID: 47F7ABD6F9FBD428
Verified
Learn about vigilant mode.

Reveng 0.2.0 — workspace-distribution release.

Workspace-distribution model

Reveng's agents and skills are now copied into the workspace's .claude/ directory by reveng init, replacing the previous --plugin-dir model. The runner discovers them via Claude Code's standard project-level discovery, and exits with a friendly error if the workspace hasn't been initialised. See specs/workspace-distribution.md for the full design.

Runner UX

  • Streaming progress to stderr: tool calls (e.g. ▸ Read: src/main.py), assistant text, and a ✓ Done line with elapsed time. -v still dumps raw stream-json.
  • Interactive bailout prompt when running outside reveng sandbox on a TTY; non-interactive runs proceed with a warning.

Structural changes

  • Agents migrated to agents/<name>/AGENT.md directory pattern, preparing for future template extraction.
  • Each agent's skills: frontmatter now declares which skills it can invoke (subagents otherwise can't see them).
  • Workspace CLAUDE.md template shipped via reveng init includes a hard constraint against using auto-memory in the workspace (clean-slate runs).
  • .claude-plugin/ removed; reveng is now plainly a harness around Claude Code, not a marketplace plugin.

Project framing

Reveng is the successor to DEFRA/claude-legacy-reveng-plugin, which this repository supersedes.

Full changelog: v0.1.0...v0.2.0

Assets 2
Loading