From devflow
Use when you need to enrich any task with project context — codebase structure, semantic analysis, patterns, and conventions
npx claudepluginhub nexuz-sys/devflow --plugin devflowThis skill uses the workspace's default tool permissions.
Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Processes code review feedback technically: verify suggestions against codebase, clarify unclear items, push back if questionable, implement after evaluation—not blind agreement.
Enriches tasks with deep project context by leveraging dotcontext's semantic analysis (Full mode), .context/ files (Lite mode), or direct exploration (Minimal mode).
Announce at start: "I'm using the devflow:context-awareness skill to gather project context."
Always run these three to get comprehensive context:
context({ action: "buildSemantic" }) # AST analysis, symbols, deps, cross-file refs
context({ action: "getMap" }) # Codebase map (file tree + descriptions + stats)
context({ action: "detectPatterns" }) # Architecture patterns and conventions
Returns:
Task-specific queries:
context({ action: "searchQA" }) # Search project Q&A knowledge base
explore({ action: "analyze", path: "<file>" }) # Deep analysis of specific file
explore({ action: "search", query: "<term>" }) # Semantic code search
explore({ action: "getStructure" }) # Full directory structure
Tip: buildSemantic + getMap + detectPatterns should be run together — they complement each other. buildSemantic gives symbols, getMap gives structure, detectPatterns gives architectural context.
Read these files in order (stop when you have enough context):
Project overview: .context/docs/project-overview.md
Codebase map: .context/docs/codebase-map.json
Development workflow: .context/docs/development-workflow.md
Testing strategy: .context/docs/testing-strategy.md
Tooling: .context/docs/tooling.md
ADR Guardrails: .context/docs/adrs/README.md
Agent playbooks (for domain-specific context):
.context/agents/<relevant-agent>.mdSkill files (for task-specific guidance):
.context/skills/<relevant-skill>/SKILL.mdStandard code exploration:
README.md, CONTRIBUTING.md, package.json/Cargo.toml/go.modTip: If Minimal mode, suggest running /devflow init to scaffold .context/ and unlock Lite mode.
.context/ CompatibilityDevFlow reads .context/ files generated by either system:
npm i -g @dotcontext/cli && dotcontext mcp:install claude --local, then use MCP tools) — produces the richest content with semantic analysis/devflow init) — produces good content via codebase explorationBoth use the same dotcontext v2 frontmatter format (scaffoldVersion: "2.0.0"). Files from either source are interchangeable.
Priority rule: If a file has status: filled, trust it regardless of origin. If status: unfilled, it's a scaffold waiting to be enriched — DevFlow can fill it during context gathering.
After gathering context, summarize for the consuming skill:
## Project Context
**Type:** [CLI / Web / Library / API / Monorepo]
**Stack:** [key technologies]
**Patterns:** [key patterns found — e.g., "service layer + repository pattern"]
### Relevant Files
- `path/to/file.ts` — [what it does, why it's relevant]
### Conventions
- [coding conventions discovered]
### Dependencies
- [key dependencies relevant to the task]
### Active ADR Guardrails
- [list of guardrails from relevant ADRs]
### Recent Activity
- [recent changes in the area being modified]
dotcontext caches semantic context automatically. Re-building is only needed when:
Context files are static — read once per session unless the task modifies them.
No caching — explore as needed.
| Thought | Reality |
|---|---|
| "I know this codebase" | Context changes. Gather fresh context, especially after other agents worked. |
| "Context gathering is slow" | In Full mode it's cached. In Lite mode it's a few file reads. Worth it. |
| "I'll just read the relevant file" | Isolation without context leads to inconsistent code. Understand the whole. |
| "Semantic analysis is overkill" | Only available in Full mode. In Lite/Minimal, context is lightweight. |