agentleFS
Sign inStart free
jjmrochajjmrocha/coding-skills/research/SKILL.md

research

Skill1 starsChanged 30 days agoMITView on GitHub

Use when you have a question about this codebase or what we've documented, and need a sourced, cited answer read from the actual source — the KB and the code — not answered from memory.

189 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use daily-arxiv, exp-eval and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文，过滤相关性，auto-ingest 高优先级论文，检测 SOTA 更新

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门：Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • truongtv22Skill

    Problem-Solving Techniques

    Apply systematic problem-solving techniques for complexity spirals (simplification cascades), innovation blocks (collision-zone thinking), recurring patterns (meta-pattern recognition), assumption constraints (inversion exercise), scale uncertainty (scale game), and dispatch when stuck. Techniques derived from Microsoft Amplifier project patterns adapted for immediate application.

    Also in 5 projects that use this

  • liufeng1976Skill

    setup-ts-deep-modules

    Wire dependency-cruiser into a TypeScript repo so each package is a deep module — implementation hidden in subfolders, reachable only through its entry-point files. User-invoked.

    Also in 43 projects that use this

  • skyllwtSkill

    exp-pilot-eval

    Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-run

    Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.

    Also in 4 projects that use this

  • +5 more from Lambenthan
  • +1 more from skyllwt

The file

Create a free account to mount
Files
SKILL.md

Research

Core principle: Bring a sourced answer to the user. Every claim traces to something you read this session — never memory, never assumption. Find the truth in the most authoritative source and cite it (file:line, doc, KB entry).

Where the answer lives

The answer is always in the code — read it (preferably with serena) to get and confirm it. If kb_path is configured, load knowledge-base first to locate the right surfaces and files faster, then read the code to validate.

Code is ground truth. When the KB and code disagree, the code wins — report the drift, and fix the KB.

External / wider-world questions — best practice, library choice, "is X true?" — aren't this skill's job. Use deep-research-agent specialist from using-software-specialists skill; don't answer them from memory.

Read with serena, not from memory

For codebase questions, load the actual symbol — don't recall it: get_symbols_overview (map a file) → find_symbol (read one symbol's body, not the whole file) → find_referencing_symbols (trace callers, "what happens when X?"). Grep + Read only when serena can't resolve it. Reads fewer tokens and cites exactly.

Match effort to the question

  • Direct lookup ("do we have X?", "payload of event X?") → find it, cite file:line, done.
  • Tracing ("what happens when X?") → follow it with find_referencing_symbols; give confidence per finding, report conflicts, set the stopping criterion upfront.

Asking the user questions

Clarify an ambiguous request before investigating. If what the user wants answered isn't explicit, ask first — researching the wrong question wastes time and tokens. Keep it tight: ask only what you can't determine yourself and what changes where you look. Once the question is clear, investigate — don't ask the user to recite what you can read, and never ask what the research is supposed to tell you.

Output

Answer + cited source. For non-trivial findings, add confidence and named gaps. The deliverable is the answer — stop there. If the user then wants to act on it, that's a separate step (e.g. using-software-specialists to plan and implement).

Red Flags

ThoughtReality
"I'm pretty sure we have an endpoint for that"Don't answer code from memory. Read the symbol with serena, cite file:line.
"Let me just Read the whole file"Use serena's symbol tools — load only what answers the question.
"The KB says so — good enough"Docs drift. Validate against the code; it's ground truth.
"Let me go research the best-practice approach"That's external tech research — hand off to using-software-specialists.
"Let me ask what they're trying to build"They came with a question, not a goal. Answer it first.
"Let me research more to be thorough"Know your stopping criterion. Good-enough beats never-finished.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free