comp
Create or update all 4 COMP files (CLAUDE.md, ORIENT.md, MEMORY.md, PLAN.md) for a directory
Skill by griffinhilly
Chalk Labs
Need help deploying comp?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
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/griffinhilly-comp
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/comp
What's inside
- Skill name
comp- Description
- Create or update all 4 COMP files (CLAUDE.md, ORIENT.md, MEMORY.md, PLAN.md) for a directory
- License
- MIT
- Instructions
- 51 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# COMP Update Create or update all 4 COMP files for a project directory. ## Input Directory to update: `$ARGUMENTS` If no arguments, use the current working directory. If ambiguous, ask. ## The 4 Files | File | Purpose | Audience | Change Frequency | |------|---------|----------|-----------------| | **C**LAUDE.md | Behavioral contract — how the AI should work here | Agent | Rare — only when conventions or architecture change | | **O**RIENT.md | Orientation — what this project is, how to work in it | Human | When project shape changes | | **M**EMORY.md | Accumulated knowledge — decisions, gotchas, cross-session context | Agent + Human | Most sessions | | **P**LAN.md | Direction — roadmap, phases, progress, next steps (with `## Current State` at top) | Human + Agent | Most sessions | ## Process ### Step 1: Read Existing Files Read all 4 COMP files if they exist. Also read the directory listing (ls). ### Step 2: Determine What Needs Updating Based on the current session's work and the current state of the directory: - Which files exist vs need creation? - Which existing files are stale or incomplete? - What new information needs recording? ### Step 3: Create or Update Each File **PLAN.md** — If creating: project roadmap, current phase, next steps, and a `## Current State` section at the top with active work, blockers, and what to do next session. If updating: mark progress, adjust priorities, refresh the Current State section. **MEMORY.md** — If creating: key decisions, known gotchas, cross-session context. If updating: add new decisions/gotchas. MEMORY is for durable knowledge that accumulates — don't put ephemeral session state here (that goes in PLAN.md's Current State section). **ORIENT.md** — If creating: one-paragraph project description, current codebase shape (mental model, not file index), most common operations, known weirdness, key links. If updating: refresh to reflect new capabilities or structural changes. Written for the human — answer "what do I need to know to work on this after two weeks away?" **CLAUDE.md** — If creating: project overview, key technical details, conventions, situational guides. If updating: add new conventions, update schema. Rare — don't add session-specific detail here. ### Step 4: Summary Report what was created/updated. ## Rules - Always READ existing files before editing — never overwrite blindly - Keep each file focused on its purpose — don't duplicate across files - CLAUDE.md is the behavioral contract (for the agent); ORIENT.md is orientation (for the human) - MEMORY.md is durable cross-session knowledge; PLAN.md's Current State section is the session-scoped snapshot - For sub-projects, reference the parent project's COMP files where relevant
From griffinhilly/claude-code-synthesis, MIT.
More from griffinhilly/claude-code-synthesis
- 0.575score
autoresearch
griffinhilly
Bounded autonomous optimization loop with rollback.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
check-resolvable
griffinhilly
Audit ~/.claude/ for orphans — skills, commands, guides, and tools that exist on disk but have no path from CLAUDE.md / RESOLVER.md / other reachable files.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
distill
griffinhilly
Diff an agent draft against the user-corrected final, extract patterns from the corrections, and propose candidate rules for skill files.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
wrapup
griffinhilly
End-of-session closer.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
verify
griffinhilly
Verify outputs after pipeline steps, implementations, or any multi-step work.
SkillCFreshness: C★ 67Claude Code +9 - 0.572score
implement
griffinhilly
Execute a pre-defined plan using implementation agents (no design changes)
SkillCFreshness: C★ 67Claude Code +9 - 0.572score
review
griffinhilly
Review completed work against success criteria and flag issues
SkillCFreshness: C★ 67Claude Code +9 - 0.510score
debug
griffinhilly
Structured debugging with mandatory evidence gathering before any fix attempt.
SkillDFreshness: D★ 67Claude Code +9 - 0.510score
ship
griffinhilly
Pre-commit readiness check.
SkillDFreshness: D★ 67Claude Code +9 - 0.507score
research
griffinhilly
Launch a research-only agent that investigates a topic and returns findings (no implementation)
SkillDFreshness: D★ 67Claude Code +9 - 0.507score
retro
griffinhilly
Periodic retrospective — assess what Claude did well, what went poorly, and what to improve
SkillDFreshness: D★ 67Claude Code +9 - 0.507score
finalize
griffinhilly
End-of-session COMP update — record decisions, progress, and state changes
SkillDFreshness: D★ 67Claude Code +9
Signals
- Stars
- 67
- Forks
- 6
- Open issues
- 1
- Last commit (this folder)
- Apr 1, 2026 (7 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from griffinhilly
- 0.575score
autoresearch
griffinhilly
Bounded autonomous optimization loop with rollback.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
check-resolvable
griffinhilly
Audit ~/.claude/ for orphans — skills, commands, guides, and tools that exist on disk but have no path from CLAUDE.md / RESOLVER.md / other reachable files.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
distill
griffinhilly
Diff an agent draft against the user-corrected final, extract patterns from the corrections, and propose candidate rules for skill files.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
wrapup
griffinhilly
End-of-session closer.
SkillCFreshness: C★ 67Claude Code +9 - 0.575score
verify
griffinhilly
Verify outputs after pipeline steps, implementations, or any multi-step work.
SkillCFreshness: C★ 67Claude Code +9 - 0.572score
implement
griffinhilly
Execute a pre-defined plan using implementation agents (no design changes)
SkillCFreshness: C★ 67Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get comp running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
comp skill. Agent Mart. https://agentmart.dev/skills/griffinhilly-comp (last updated 2026-10-24)