| Age | Commit message (Collapse) | Author | Lines | |
|---|---|---|---|---|
| 2026-02-01 | execute_command: Making return loop friendly | Leonard Kugis | -4/+79 | |
| Default behaviour of aichat will terminate the request immediately, if a tool returns with exit code != 0. This is inappropriate if the model is capable to fix commands itself. Thus, this patch of execute_command will not forward the exit code, but print it to stderr and always return exit code 0. | ||||
| 2024-11-16 | refactor: improve bash code (#125) | sigoden | -7/+3 | |
| * refactor: extract guard_path to utils/guard_path.sh * add utils/guard_operation.sh | ||||
| 2024-11-14 | feat: add `LLM_OUTPUT=/dev/stdout` to all tools/agents bashscripts (#123) | sigoden | -0/+2 | |
| 2024-08-03 | chore: update description of some tools | sigoden | -1/+1 | |
| 2024-07-10 | feat: adjust the way of returning data to LLM (#69) | sigoden | -3/+10 | |
| 2024-07-03 | feat: rename `may_execute_*` to `execute_*` (#56) | sigoden | -0/+11 | |
 |
index : llm-functions-docker.git | |
| Wrapper scripts for docker to be used by LLM agents |
| aboutsummaryrefslogtreecommitdiffstats |