Acquires missing code repositories for the selected idea (Phase A) and conducts comprehensive code survey mapping academic concepts to implementations (Phase B).
Acquires missing code repositories for the selected idea (Phase A) and conducts comprehensive code survey mapping academic concepts to implementations (Phase B).
Acquires missing code repositories for the selected idea (Phase A) and conducts comprehensive code survey mapping academic concepts to implementations (Phase B). Outputs acquired_code_repos, updated_prepare_res, and model_survey for downst...
Acquires missing code repositories for the selected idea (Phase A) and conducts comprehensive code survey mapping academic concepts to implementations (Phase B). Outputs acquired_code_repos, updated_prepare_res, and model_survey for downst...
Trigger Rules
Use this skill when the user request matches its research workflow scope. Prefer the bundled resources instead of recreating templates or reference material. Keep outputs traceable to project files, citations, scripts, or upstream evidence.
Resource Use Rules
Read from references/ only when the current task needs the extra detail.
Treat scripts/ as optional helpers. Run them only when their dependencies are available, keep outputs in the project workspace, and explain a manual fallback if execution is blocked.
Execution Contract
Resolve every relative path from this skill directory first.
Prefer inspection before mutation when invoking bundled scripts.
If a required runtime, CLI, credential, or API is unavailable, explain the blocker and continue with the best manual fallback instead of silently skipping the step.
Do not write generated artifacts back into the skill directory; save them inside the active project workspace.
Upstream Instructions
Inno Code Survey (Repo Acquisition + Code Survey)
Merges _acquire_missing_repos, _update_prepare_res_with_new_repos, and _conduct_code_survey from run_infer_idea_ours.py (lines 639–828, 1038–1052) into a single two-phase skill.
Directory structure
skills/inno-code-survey/
├── SKILL.md ← this file
├── prompts/
│ ├── build_repo_acquisition_query.md ← Phase A query template
│ └── build_code_survey_query.md ← Phase B query template
├── references/
│ ├── repo_acquisition_agent.md ← Phase A agent system prompt & tools
│ └── code_survey_agent.md ← Phase B agent system prompt & tools
└── scripts/
└── github_search_clone.py ← GitHub search + clone helper
Path conventions
All file paths use semantic directory names under the project root:
Path
Contents
Ideation/references/papers/
Downloaded arXiv LaTeX sources (, , )
.tex
.txt
.md
Experiment/code_references/<repo_name>/
Cloned GitHub repositories
Experiment/code_references/model_survey.md
Code survey implementation report
Experiment/code_references/logs/
Phase A & B agent cache files
Inputs
These are aligned with outputs from inno-idea-generation and inno-prepare-resources:
Input
Source
Description
selected_idea
Ideation/ideas/selected_idea.txt or final_selected_idea_data
The finalized selected idea (full markdown)
download_res
inno-prepare-resources output
Result log from downloading arXiv paper sources
prepare_res
inno-prepare-resources output (JSON)
Contains reference_codebases and reference_paths
context_variables
Shared context dict
Accumulated pipeline context
instance.json
<project_path>/instance.json
Paths are absolute when created by Dr. Claw (Experiment.code_references, Ideation.references); use as-is or resolve with path.join(project_path, value) if relative. Also date_limit from context.
Outputs
Output
Description
Consumer
acquired_code_repos
Dict of {name: path} for newly cloned repos
Phase B, cache
updated_prepare_res
prepare_res JSON with new repos merged into reference_codebases / reference_paths
Maps to _acquire_missing_repos (lines 745–792) + _update_prepare_res_with_new_repos (lines 639–686).
Step A1: Analyze the selected idea and identify gaps
Read selected_idea and identify 2–3 missing technical components — novel or specialized parts that are likely NOT in the standard repos already present in Experiment/code_references/.
Step A2: Search GitHub using the "Cascade" strategy
For each missing component, perform 6 distinct queries using progressive decomposition:
Level 1 (Specific): Search for the exact mechanism name
Construct the query using selected_idea, download_res, and extra_repo_info (from Phase A):
I have an innovative idea related to machine learning:
{selected_idea}
I have carefully gone through these papers' github repositories and found download
some of them in my local machine, in the directory `Experiment/code_references/`, use `ls`, `tree`,
and `find` to navigate the directory.
And I have also downloaded the corresponding paper (LaTeX sources, markdown, txt),
with the following information:
{download_res}
{extra_repo_info_block}
Your task is to carefully understand the innovative idea, and thoroughly review
codebases and generate a comprehensive implementation report for the innovative
idea. You can NOT stop to review the codebases until you have get all academic
concepts in the innovative idea.
Note that the code implementation should be as complete as possible.
Step B2: Survey all repos in Experiment/code_references/
Use Linux commands to navigate and read code:
Action
Command
List repos
ls Experiment/code_references/ or tree Experiment/code_references/ -L 1
View repo structure
tree Experiment/code_references/<repo>/ -L 3
Find Python files
find Experiment/code_references/<repo>/ -name "*.py" -type f
