From fellowship
Loads task-relevant context for non-trivial tasks by reading CLAUDE.md, checking checkpoints, identifying package scope in monorepos, and producing a structured Session Context block.
How this skill is triggered — by the user, by Claude, or both
Slash command
/fellowship:councilThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Loads focused, task-relevant context at the start of any non-trivial work session. Produces a structured Session Context block that serves as the foundation for all downstream work. This is the "onboarding the agent within every repo" pattern from context engineering.
Loads focused, task-relevant context at the start of any non-trivial work session. Produces a structured Session Context block that serves as the foundation for all downstream work. This is the "onboarding the agent within every repo" pattern from context engineering.
/questBefore doing anything else, check if .fellowship/checkpoint.md exists (in repo root). (.fellowship/ is the default data directory; users can override it via dataDir in ~/.claude/fellowship.json.)
If it does:
Read the root CLAUDE.md. Extract:
If no CLAUDE.md exists, note: "Consider running /chronicle to set up project context."
Also check for ~/.claude/fellowship.json (the user's personal Claude directory). If it exists, read it and note any non-default settings. These will be included in the Session Context block under Architecture Notes so downstream skills (quest, lembas) are aware of the active configuration.
Ask one focused question:
"What are you working on? (One sentence describing the task and which area of the codebase it touches.)"
If invoked by quest, the task description is passed in — skip the question.
Determine the scope of the task within the project:
Monorepo (has packages/, apps/, or a workspace config like pnpm-workspace.yaml):
packages/<name>/, apps/<name>/)ls top-level directories) and ask the userpackages/<name>/CLAUDE.md), read it and merge its conventions with the root CLAUDE.md. Package-level conventions override root conventions where they conflict.Single-package repo: Skip this step — the scope is the whole project. Proceed to Step 4.
This scope constrains all downstream scanning and verification.
Use the Explore agent (Task tool with subagent_type=Explore) to find files related to the task description. In a monorepo, scope the search to the identified package(s) — do not scan the entire repo.
Focus on:
Keep the scan targeted — 5-10 key files maximum, not an exhaustive listing.
Output a structured block in this exact format:
## Session Context
**Task:** [one-line description]
**Package(s):** [package name(s) and path(s)]
**Key Files:**
- [path/to/file:lines] — [why it's relevant]
- [path/to/file:lines] — [why it's relevant]
**Relevant Conventions:**
- [convention from root CLAUDE.md that applies]
- [convention from package CLAUDE.md that applies, if exists]
**Architecture Notes:**
- [constraints, patterns, or dependencies to be aware of]
**Out of Scope:**
- [things explicitly not to touch or change]
- [other packages not affected by this task]
Present the Session Context block and ask:
"Does this capture the right scope? Anything missing or out of bounds?"
Revise based on feedback.
npx claudepluginhub justinjdev/fellowshipScans codebase to auto-generate CLAUDE.md project config and .rune/ state files for full context in AI coding sessions. Use on new repos or missing/stale context.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.