Conversation
…e files Agent-Logs-Url: https://github.com/githubnext/autoloop/sessions/149e05de-84fa-4e41-9a14-cf8948cdb980 Co-authored-by: mrjf <180956+mrjf@users.noreply.github.com>
|
@copilot resolve the merge conflicts in this pull request, and fix failing tests, if any |
…Python scheduler Agent-Logs-Url: https://github.com/githubnext/autoloop/sessions/705c2b0a-fd32-4450-bddb-a715daf0dce1 Co-authored-by: mrjf <180956+mrjf@users.noreply.github.com>
Resolved the conflict in Warning Firewall rules blocked me from connecting to one or more addresses (expand for details)I tried to connect to the following addresses, but was blocked by firewall rules:
If you need me to access, download, or install something from one of these locations, you can either:
|
|
@copilot resolve conflicts |
…ry-max-file-size # Conflicts: # .github/workflows/autoloop.lock.yml # .github/workflows/scripts/autoloop_scheduler.py # workflows/scripts/autoloop_scheduler.py Co-authored-by: mrjf <180956+mrjf@users.noreply.github.com>
Merged the latest Warning Firewall rules blocked me from connecting to one or more addresses (expand for details)I tried to connect to the following addresses, but was blocked by firewall rules:
If you need me to access, download, or install something from one of these locations, you can either:
|
Long-running programs grow their per-program state file unboundedly, eventually exceeding repo-memory's default size limit (breaking the scheduler) and flooding the agent's context window with stale iteration history.
Changes
repo-memory.max-file-size: 30720inworkflows/autoloop.mdfrontmatter — 30 KB budget, sized for ~10 detailed iteration entries plus ~5 compressed-range summaries. Comment notes it's tunable per program cadence.### Iters 50–100 — ✅ (metrics 20→55): brief summary of what worked across this rangestate_file_size_bytes/state_file_max_bytesadded to/tmp/gh-aw/autoloop.jsonvia a newget_state_file_size()helper andSTATE_FILE_MAX_BYTESconstant inworkflows/scripts/autoloop_scheduler.py(mirrored to.github/workflows/scripts/), so the agent can act on the 80% threshold rather than only reacting to overflow. Documented in the "Reading Programs" section..github/workflows/autoloop.lock.yml.origin/maininto the branch and re-applied the scheduler-side changes against the new Python scheduler (the upstream had refactored the inline JS heredoc intoworkflows/scripts/autoloop_scheduler.py).