code-intelligence
Use for code ownership, structure, symbols, types, references, call relationships, structural patterns, and diagnostics.
Skill by som1shi
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/som1shi-code-intelligence
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/code-intelligence
What's inside
- Skill name
code-intelligence- Description
- Use for code ownership, structure, symbols, types, references, call relationships, structural patterns, and diagnostics. Select the smallest relevant combination of symbol/module, LSP, AST, and diagnostic tools.
- License
- MIT
- Instructions
- 37 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Code Intelligence Use semantic code tools when code structure, behavior, types, relationships, or diagnostics matter. Plain file and text tools remain correct for filenames, comments, logs, configuration text, and exact strings. ## Select evidence by question When a situational Pi Lens tool is unavailable, call `pi_lens_activate_tools` with the needed tool names. Use the activated tools on the next model turn. - **Repository-wide orientation:** use `project_report`, when available, for questions about architecture, subsystems, hubs, or dependency cycles. - **Ownership and shape:** `symbol_search`, then `module_report`; read exact bodies with `read_symbol` or `read_enclosing`. - **Types and relationships:** `lsp_navigation` for definitions, references, implementations, hover, symbols, rename previews, and call hierarchy. - **Structural patterns:** `ast_grep_search`; use `ast_grep_replace` for structural rewrites and dry-run before applying. Use `ast_grep_outline` for syntax-only structure and `ast_grep_dump` when the AST shape is unclear. - **Diagnostics:** `lsp_diagnostics` for focused language-server checks and `lens_diagnostics` for aggregate edited-file or project findings. Cached diagnostics do not prove unobserved files are clean; select an active check when the claim needs fresh coverage. Use every evidence group that answers a material question, but do not call groups mechanically. Gather the minimum evidence that can settle ownership, implementation, or correctness. ## Read before editing Before changing an identifiable function, class, method, callback, or symbol, read its actual body with `read_symbol` or `read_enclosing`. For multi-file changes, inspect relevant structure first. Use plain `read` when semantic tools are unavailable or the target is not source code. ## Structural search Use specific valid code patterns and scope them to relevant paths. If a search returns no match, simplify it once. Use `ast_grep_dump` when node kinds or nesting are unclear, then fall back to text search only when structure is not the right query. Dry-run structural replacements before applying them. ## LSP use Use the tool's registered `path` and `paths` fields. Start from a real file in the project when workspace symbol or project resolution needs context. Query references from the definition when usage-site results can be partial. Run targeted LSP diagnostics after each coherent code edit group and again after the final relevant edit. State the concrete reason when diagnostics are unavailable or do not apply. ## Trust the result boundary Code-intelligence findings guide investigation and review. They do not authorize edits outside the approved behavior. Validate diagnostics and reviewer findings against the real producer, reachable state, scope, and contract before changing code.
From som1shi/pi-config, MIT.
More from som1shi/pi-config
- 0.524score
commit
som1shi
Git commit and branch naming conventions only — ss/ branch prefix, terse imperative commit messages.
SkillBFreshness: BClaude Code +9 - 0.521score
brainstorming
som1shi
Use for unclear intent or a material goal, behavior, placement, or design choice unresolved after inspection.
SkillBFreshness: BClaude Code +9 - 0.521score
delegation
som1shi
Use subagents for nontrivial work, review support, research, reconnaissance, validation, and exceptional isolated implementation.
SkillBFreshness: BClaude Code +9 - 0.521score
context-mode
som1shi
Use Context Mode for large command, test, log, API, document, data, browser, or MCP output that would otherwise flood the model context.
SkillBFreshness: BClaude Code +9 - 0.521score
code-quality-review
som1shi
Deep code-quality and simplification review for explicit review requests or concrete useful background quality checks.
SkillBFreshness: BClaude Code +9 - 0.471score
behavioral-proof
som1shi
Selects proportionate evidence for material behavior changes and bug fixes.
SkillCFreshness: CClaude Code +9 - 0.455score
diagnose
som1shi
Disciplined diagnosis loop for hard bugs and performance regressions.
SkillCFreshness: CClaude Code +9 - 0.450score
improve-codebase-architecture
som1shi
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/.
SkillCFreshness: CClaude Code +9 - 0.450score
tdd
som1shi
Test-driven development with red-green-refactor loop.
SkillCFreshness: CClaude Code +9 - 0.450score
grill-with-docs
som1shi
Grilling session that challenges your plan against the existing domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as…
SkillCFreshness: CClaude Code +9 - 0.445score
grill-me
som1shi
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree.
SkillCFreshness: CClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 18, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/som1shi/pi-config
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from som1shi
- 0.524score
commit
som1shi
Git commit and branch naming conventions only — ss/ branch prefix, terse imperative commit messages.
SkillBFreshness: BClaude Code +9 - 0.521score
brainstorming
som1shi
Use for unclear intent or a material goal, behavior, placement, or design choice unresolved after inspection.
SkillBFreshness: BClaude Code +9 - 0.521score
delegation
som1shi
Use subagents for nontrivial work, review support, research, reconnaissance, validation, and exceptional isolated implementation.
SkillBFreshness: BClaude Code +9 - 0.521score
context-mode
som1shi
Use Context Mode for large command, test, log, API, document, data, browser, or MCP output that would otherwise flood the model context.
SkillBFreshness: BClaude Code +9 - 0.521score
code-quality-review
som1shi
Deep code-quality and simplification review for explicit review requests or concrete useful background quality checks.
SkillBFreshness: BClaude Code +9 - 0.471score
behavioral-proof
som1shi
Selects proportionate evidence for material behavior changes and bug fixes.
SkillCFreshness: CClaude Code +9
Cite this page
code-intelligence skill. Agent Mart. https://agentmart.dev/skills/som1shi-code-intelligence (last updated 2026-10-24)