feat: add Terminal-Bench weekly regression workflow [LET-7791]#1232
Open
feat: add Terminal-Bench weekly regression workflow [LET-7791]#1232
Conversation
Add scheduled GitHub Actions workflow (Mon + Thu) to run a curated subset of Terminal-Bench 2.0 tasks against Letta Code using Harbor. Tests two models (sonnet-4.6-low, gpt-5-minimal) at concurrency 4, reports results via GitHub Issue, and fails on regressions. - Harbor BaseInstalledAgent adapter (from recovery-bench) - 12 curated tasks (7 easy, 5 medium) - report.py: parses Harbor job results, updates GH Issue tracker - baseline.json: manually populated after initial runs 👾 Generated with [Letta Code](https://letta.com) Co-Authored-By: Letta <noreply@letta.com>
Build from main branch instead of npm so regression tests run against the latest code, not the last published release. 👾 Generated with [Letta Code](https://letta.com) Co-Authored-By: Letta <noreply@letta.com>
devanshrj
force-pushed
the
devansh/terminal-bench-regression
branch
from
68ecfdf to
36c8200
Compare
devanshrj
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
sonnet-4.6-lowandgpt-5-minimal) at concurrency 4, reports results via GitHub Issue, fails on regressionsWhat's included
benchmarks/terminal_bench/letta_code_agent.py) —BaseInstalledAgentimplementation copied from recovery-benchinstall-letta-code.sh.j2) — clones repo, builds with bun, installs globally inside TB containersregression-tasks.txt) — 7 easy + 5 medium, covering file ops, coding, debugging, system adminreport.py) — parses Harbor job results, creates/updates a pinned "Terminal-Bench Regression Tracker" GitHub Issue, exits non-zero on regressionsbaseline.json) — empty, to be manually populated after initial runsterminal-bench-regression.yml) — cron + manual dispatch, matrix over models, artifact uploadTest plan
ANTHROPIC_API_KEY,OPENAI_API_KEY,LETTA_API_KEYworkflow_dispatchrun and verify Harbor executes tasksbaseline.jsonfrom first run results👾 Generated with Letta Code