Repository navigation
feat(slack): let bare readers inspect the bot’s recent run - #141
Conversation
Shashankss1205
left a comment
There was a problem hiding this comment.
Evaluation of 4073e86: bare readers disclose the inferred trace/run and prefer per-channel completed history, with an explicit workdir fallback. Typed reader arguments keep their existing behavior. Inferred arguments re-enter the admission gate; candidate paths are confined before reads, corrupt newest traces do not silently select an older answer, and multiple runs require an explicit choice. Channel history is process-local as documented.
Validation: 17 new checks pass, including demo stage0 followed by metrics through the real CLI, missing history, channel isolation, malformed traces and escaping symlinks. The focused Slack/documentation run passes 151 tests. Full integrated local suite: 2,388 passed, 13 skipped, 13 paid tests deselected. Ruff/diff checks and all 133 installed-wheel module imports pass. GitHub CI is pending; merge after #140 and successful checks.
No blocking code findings. diff correctly retains an explicit comparison baseline.
|
Final merge check for 4073e86: all GitHub CI test jobs (Python 3.12, 3.13 and 3.14), lint, distribution validation and the live-test guard passed. #140 is merged; the remaining diff contains only the five Slack-related files. The reviewed head is unchanged, with no unresolved review threads or requested changes. Merging with ancestry preserved for #143. |
After a bot run, bare
metrics,viz,replayandtracenow select a recent trace and state exactly which path and run id were selected. Remember completed traces per channel in a thread-safe bot-local history; fall back to the newest trace underslack-runs/in a new channel or after restart, and disclose that fallback.Explicit arguments keep their current behavior. Resolved arguments pass the existing confinement/allowlist gate; empty history, corrupt traces and multi-run ambiguity produce actionable refusals.
diffkeeps its explicit comparison baseline. Channel history is process-local, as documented.Add 17 regression checks, including the real
demo stage0 → metricssubprocess flow, channel selection and symlink confinement; update the Slack cookbook and selected-test figure. The pre-integration behavior fails 13 new checks. Focused Slack/documentation checks: 151 passed. The final integrated tree through #143 passes 2,395 tests on both Python 3.12 and 3.13 locally; 13 skipped and 13 paid tests excluded. Ruff/diff checks and all 133 installed-wheel module imports pass. GitHub CI results are recorded in the review before merge.No Slack credentials, network connection or paid model calls are used by the tests.
Fixes #49.