Add cleared-compliance test employee + fix dead Groq model name - #18
Open
rahulkini31 wants to merge 2 commits into
Open
Add cleared-compliance test employee + fix dead Groq model name#18rahulkini31 wants to merge 2 commits into
rahulkini31 wants to merge 2 commits into
Conversation
Adds Arjun Reddy (Platform Engineering, L4) with overall_status: CLEARED so the agent can be exercised on the straight-through path (no ask_human pause), compared against EMP-2026-0847/0848 which are both PENDING. Verified with python agent.py --dry-run (JSON valid) and python agent.py --employee EMP-2026-0849 (agent went straight to provisioning + welcome email, as expected for a cleared employee).
call_llm() hardcoded 'llama-3.3-70b-versatile', which Groq has since retired from its catalogue — every fresh GROQ_API_KEY gets a 404 model_not_found on the very first LLM call, before a student sees any agent behavior at all. Swapped to openai/gpt-oss-120b (currently available on Groq's free tier) via a single GROQ_MODEL constant, and updated the --dry-run provider line to print it instead of the old hardcoded string. Verified: python agent.py --employee EMP-2026-0849 now completes the full ReAct loop end to end.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What I changed
EMP-2026-0849(Arjun Reddy, Platform Engineering, L4) todata/employees.json, and their compliance record — alreadyoverall_status: CLEARED— todata/compliance_state.json, so the agent can be exercised on the straight-through path (noask_humanpause), as distinct fromEMP-2026-0847/0848which are bothPENDING.agent.py'scall_llm()hardcoded the modelllama-3.3-70b-versatile, which Groq has since retired from its catalogue — every freshGROQ_API_KEYhit a 404model_not_foundon the very first LLM call, before a student saw any agent behavior at all. Swapped toopenai/gpt-oss-120b(currently available on Groq's free tier) behind a singleGROQ_MODELconstant, and updated the--dry-runprovider line to print it instead of the old hardcoded string.What I tested
python agent.py --dry-run— all file/tool checks pass, JSON parses cleanly for both edited files.python agent.py --employee EMP-2026-0849— with the model fix in place, the agent completes the full ReAct loop end to end: retrieves the profile, confirms complianceCLEARED, goes straight toprovision_access(noask_humanpause, correctly distinguishing this from the PENDING employees), and sends the welcome email toarjun.reddy@meridian.com— the real address from the profile, not the employee_id, per the system prompt's PII-in-logs-vs-email rule.EMP-2026-0847(PENDING) case too, to confirm the fix didn't change the ask_human path — still pauses for confirmation as before.Notes for the reviewer
The model-name fix (commit 2) goes slightly beyond what
docs/first_task.mdsuggests as a stretch change — it touchesagent.pyrather than a docs file. I included it because without it, the repo's core "run the real agent" instruction fails for literally any student picking up a fresh Groq key today, which seemed worth flagging and fixing rather than working around silently. Happy to split it into a separate PR if you'd rather review it independently from the first-task change.