fix(docs): precision improvements + feat(agent): SWE-bench SOTA optimizations#166
Merged
Merged
Conversation
…ent): add SWE-bench prompt profile, budget injection, deep_analysis tool - README: add version mapping table, clarify 3-tier vs 4-tier permissions, restructure benchmark table with deployable/research-ceiling tiers, attribute LiteLLM properly, document tool count precisely, label stock_price as infrastructure utility - Findings doc: remove leaked local file path - Architecture doc: improve GEPA attribution with naming convention note - Agent: inject budget prompt after N writes/verify-failures to prevent agent-in-loop over-editing regression (30.4% vs 34.8% single-shot) - System prompt: add SWEBENCH_TASK_PROMPT with minimal-edit rule, verify-feedback handling, and budget awareness - Prompt profiles: add swebench profile for targeted task structure - Tools: integrate DeepAnalysisTool (3-step reasoning) as core tool using existing LLM client, no hardcoded credentials
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Changes
Docs (7 files, 330+ / 27-)
README.mdexperiments/swebench_lite/findings_2026_04_21.md~/.claude/plans/pathGODSPEED_ARCHITECTURE.mdAgent (3 files)
src/godspeed/agent/system_prompt.pySWEBENCH_TASK_PROMPTwith problem-solving protocol, minimal-edit rule, verify-feedback handling, budget awarenesssrc/godspeed/agent/prompt_profiles.pyswebenchprofile with targeted preamble and plan-stylesrc/godspeed/agent/loop.pyTools (1 new file)
src/godspeed/tools/deep_analysis.pyVerification
pytest tests/test_agent_loop.py tests/test_system_prompt.py tests/test_prompt_profiles.py: 135 passed, 1 skippedruff check+ruff format --check: 0 issuesgit diff --staged: no secrets, no API keys