Skip to content

Conversation

@RajShah-1
Copy link
Contributor

@RajShah-1 RajShah-1 commented Dec 4, 2025

  • Introduced SLM as an agent over a set of given tools in an iterative loop.
  • Wired agent mode into config and CLI with reasoning and tool limits, plus a new model and tools for search, navigational reading, grep, and section summaries.
  • Many of the tools are stubs for now (section tools just gives first K characters of a section for example).

Note: Much of the code is AI generated. I am sharing this PR as a POC, and working on the details.

@RajShah-1 RajShah-1 requested a review from jarulraj December 4, 2025 04:35
* Introduced ContextRegistry, AgentOrchestrator, and AgentToolkit to run a budgeted investigation and synthesis loop over existing artifacts.
* Wired agent mode into config and CLI with reasoning and tool limits, plus a new model and tools for search, navigational reading, grep, and section summaries.
* Current agent loop is a first cut with simple prompts, no registry scoring or eviction policy, and some hardcoded paths and basic error handling.
@RajShah-1 RajShah-1 marked this pull request as draft January 23, 2026 15:40
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants