🎊 Achievement unlocked: PR Merged. @davincios passed code review, survived CI, and shipped. Respect. 🤝

👋 Join us on Discord - OpenSRE : hang out, contribute, or hunt for features and issues. Everyone's welcome.

Greptile Summary

This PR upgrades the interactive shell with four new deterministic action kinds (shell execution, LLM provider switching, sample alert, synthetic RDS test), a SlashCommandCompleter with live bottom toolbar, persistent prompt history, a centralised theme module, and an LLM-driven JSON action-plan path in the CLI agent. The /model set command now applies provider/model changes live without requiring a restart.

• P1 – In execute_cli_actions, the return value of switch_llm_provider is not checked, so a failed provider switch (e.g. unrecognised provider) is unconditionally recorded as ok=True in session history.

Confidence Score: 4/5

Safe to merge after fixing the session-history ok flag for failed provider switches; all other findings are non-blocking P2 suggestions.

One P1 defect: failed LLM provider switches are silently recorded as successful in session history. The remaining findings are a duplicate regex (P2) and a missing error guard in history loading (P2). The feature is well-tested and the new deterministic pre-pass architecture is sound.

app/cli/interactive_shell/agent_actions.py (P1 return-value bug on llm_provider action), app/cli/interactive_shell/history.py (missing OSError guard)

Important Files Changed

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[User types in REPL] --> B{Starts with /}
    B -- Yes --> C[dispatch_slash]
    B -- No --> D[execute_cli_actions\ndeterministic pre-pass]
    D --> E{Any planned\nactions?}
    E -- No --> F[classify_input router]
    E -- Yes --> G[_print_planned_actions]
    G --> H{action.kind}
    H -- slash --> I[dispatch_slash]
    H -- llm_provider --> J[switch_llm_provider\nwrites .env + os.environ]
    H -- shell --> K[subprocess.run\nshell=True]
    H -- sample_alert --> L[run_sample_alert_for_session\nLangGraph pipeline]
    H -- synthetic_test --> M[subprocess: opensre tests synthetic]
    F --> N{InputKind}
    N -- cli_help --> O[answer_cli_help]
    N -- cli_agent --> P[answer_cli_agent\nLLM call]
    N -- new_alert / follow_up --> Q[LangGraph pipeline]
    P --> R{LLM returns\nJSON action plan?}
    R -- Yes --> S[_execute_action_plan\nswitch_llm_provider or slash]
    R -- No --> T[Render Markdown reply]

_{Reviews (1): Last reviewed commit: "feat(cli): improve interactive shell aut..." | Re-trigger Greptile}