mirror of
https://github.com/agentscope-ai/ReMe.git
synced 2026-10-13 04:01:35 +00:00
Review feedback on #563: the two skill docs I added told agents not to `read` a target marked `(unindexed)`. That is wrong. `indexed` in `TraverseGraph` only records whether the path has a node in the file graph, while `read` resolves against the filesystem (`ReadStep` -> `read_file_lines_safe`), so a workspace file that has not been indexed yet is flagged by `traverse` and still reads fine. Reproduced locally: `a.md` links `target.md`, the file exists on disk, `traverse` prints `(unindexed)` and `read` returns its body. Correct the wording in both skill copies and in `_render_answer`'s docstring: `(unindexed)` means traverse has no frontmatter or hop metadata for the path, not that the path is missing. Add `test_traverse_unindexed_node_is_still_readable`, which seeds a real file on disk that is absent from the graph and asserts both the marker and a successful `read`, so the claim cannot drift again. |
||
|---|---|---|
| .. | ||
| claude_code | ||
| hermes_agent | ||
| README.md | ||
Agent Integrations
This directory contains host-specific adapters that connect external agents to ReMe. An integration may use the host's
plugin API, hooks, MCP configuration, or client interface, but it does not extend ReMe's runtime through the
reme.plugins entry-point group.
The shared TypeScript client and the DeepSeek Harness and OpenClaw adapters live in
../typescript.
Installable extensions of ReMe itself include Auto Fin and Daily Paper.