project-context
Skill0 starsChanged 43 days agoMITView on GitHub
Use when setting up a project for agent work, or when any other skill needs facts about the product, users, stack, platforms, scale targets, architecture style, design system, brand, or constraints and none are written down. Creates and maintains .agents/project-context.md — a single versioned file every skill in this collection reads first — by auto-drafting from the codebase and asking only for what the code cannot reveal. Also use when the user says "remember our stack", "set up context", "onboard yourself to this repo", or when the context file is stale.
33 other projects use a version of this.
Adopted by
Most recently taken up by
- advait/effect-durable-agenttheir own version, first seen Nov 8, 2026
- shinpr/ai-coding-project-boilerplatetheir own version, first seen Nov 8, 2026
- upex-galaxy/agentic-qa-boilerplatetheir own version, first seen Nov 7, 2026
- egorfedorov/claude-context-optimizertheir own version, first seen Nov 1, 2026
- ritamgh/claude-skillstheir own version, first seen Nov 1, 2026
- OpenxIntelligence/Coding-Agent-Skillstheir own version, first seen Nov 1, 2026
Other versions33 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 33
- shinpr/ai-coding-project-boilerplate★ 236same name
- egorfedorov/claude-context-optimizer★ 127same name
- TenureAI/PhD-Zero★ 52same name
- upex-galaxy/agentic-qa-boilerplate★ 26same name
- sergeeey/Claude-cod-top-2026★ 9same name
- perhapsspy/project-context★ 7same name
- advait/effect-durable-agent★ 4same name
- aida-core/aida-marketplace★ 4same name
- I-Onlabs/claude-code-skills★ 4same name
- IgorBabikov/Flowcairn★ 4same name
23 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository45 other files it gives its agents.
- AGENTS.mdAGENTS.md
- accessibilitySkill
- ai-interface-designSkill
- ai-native-architectureSkill
- android-designSkill
- api-designSkill
- application-securitySkill
- app-screen-patternsSkill
- auth-flowsSkill
- clean-codeSkill
- codebase-organizationSkill
- code-reviewSkill
- conversion-uxSkill
- dashboard-designSkill
- data-dense-uiSkill
- data-infrastructureSkill
- data-modelingSkill
- dependency-managementSkill
- deployment-and-infrastructureSkill
- design-foundationsSkill
- design-patternsSkill
- design-resourcesSkill
- design-reviewSkill
- design-systemsSkill
- design-tasteSkill
- environments-and-configSkill
- frontend-architectureSkill
- how-to-build-softwareSkill
- interaction-designSkill
- ios-designSkill
- lessons-from-failuresSkill
- mobile-architectureSkill
- mobile-designSkill
- motion-designSkill
- object-oriented-designSkill
- onboarding-designSkill
- project-bootstrapSkill
- reliabilitySkill
- scalabilitySkill
- software-architectureSkill
- system-designSkill
- tech-stack-selectionSkill
- testing-strategySkill
- ux-principlesSkill
- web-platformSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 24 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 19
When running this skill itself: if the file exists, you are updating it (see Workflow step 6), not rewriting it.
- Distinctive: line 99
Recording secrets: Never copy API keys, connection strings, or internal hostnames into the file.
- Very specific: line 100
Inferring scale from nothing: If there is no data, write unknown — do not guess "10k users".
“Common” is measured: kept by at least 6 in 10 of the 24 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Project Context
Every skill in this collection makes better decisions when it knows the product, the users, the stack, the platforms, the scale, and the design system. Asking the same questions in every task wastes the user's time and produces inconsistent answers. This skill writes those facts once to .agents/project-context.md and keeps them current.
Before you start
If .agents/project-context.md exists (or .claude/project-context.md), read it first and treat it as ground truth for stack, platforms, scale, design system, and constraints. Only ask for what it does not cover. If it does not exist and the task is larger than a quick fix, suggest running the project-context skill — but never block on it.
When running this skill itself: if the file exists, you are updating it (see Workflow step 6), not rewriting it.
Core principles
- Draft from evidence, ask for the rest. The codebase answers stack, structure, and tokens; only humans can answer users, goals, and constraints. Never ask what a file already says.
- Facts, not aspirations. Record what is true now ("Postgres 16 on a single primary") separately from targets ("p95 < 300 ms at 200 RPS").
- Short beats complete. A 150-line file that is read beats a 900-line file that is skimmed. Link to deeper docs instead of copying them.
- Versioned and dated. Stale context is worse than none; every update bumps the version and adds a changelog line.
Workflow
- Scan the repository (read-only): README,
package.json/lockfiles,pyproject.toml,go.mod,Cargo.toml,build.gradle(.kts),Podfile/Package.swift,app.json/app.config.*(Expo),pubspec.yaml, framework configs (next.config.*,vite.config.*,tailwind.config.*or@themeblocks),DESIGN.md, token files,docker-compose*, IaC (terraform/,pulumi/), CI workflows,ios/,android/,docs/adr/. - Draft every section of the template below from what you found. Mark each inferred line with
(inferred). - Ask only the gaps, in one batch of at most 6 questions, highest-leverage first: users and their core job; the one metric that matters; scale targets; platforms and minimum OS; brand/visual direction; hard constraints (compliance, budget, deadlines).
- Write
.agents/project-context.md. Create the.agents/folder if needed. If the project already uses.claude/, you may also symlink.claude/project-context.mdto it. - Show a 5-line summary of what was recorded and what is still unknown.
- On update: edit only the sections that changed, bump the version, add a changelog entry naming the sections and the reason.
Template
# Project Context
**Version:** 1 · **Last updated:** YYYY-MM-DD
## Product
- What it is (one sentence):
- Primary users and the job they hire it for:
- The one metric that matters now:
- Stage: idea / MVP / early traction / scaling / mature
## Platforms
- Web: yes/no — browsers/devices that matter
- iOS: yes/no — minimum iOS version, iPad?
- Android: yes/no — minimum API level, tablets/foldables?
- Other: desktop, watch, CLI, API-only
## Stack
- Languages and frameworks (with versions):
- Data stores:
- Hosting/runtime (serverless, containers, edge):
- Key third-party services (auth, payments, email, analytics, LLM providers):
- Package manager, monorepo tool, CI:
## Architecture
- Style (modular monolith, services, serverless functions, local-first…):
- Main modules/bounded contexts and who owns them:
- Integration style (REST, GraphQL, tRPC, gRPC, events):
- ADRs: link to docs/adr/ if present
## Scale and reliability
- Current load (users, RPS, data size) and 12-month target:
- Latency/availability targets (SLOs):
- Regions, data residency:
## Design system
- Source of truth (DESIGN.md, Figma, token files, component library):
- Fonts, color strategy, radius/density, icon set:
- Brand voice (3 adjectives + 1 "never"):
- Accessibility target (default WCAG 2.2 AA):
- Visual references the team likes:
## Constraints
- Compliance (GDPR, HIPAA, SOC 2, PCI, LGPD…):
- Budget/cost ceilings:
- Team size and skills:
- Deadlines:
## Conventions
- Code style, testing approach, branch/commit conventions, anything agents keep getting wrong:
## Glossary
- Domain terms and their exact meaning:
## Changelog
- vN (YYYY-MM-DD): sections touched — reasonGotchas
- Asking before scanning. Users hate answering "what framework do you use?" when
package.jsonsays it. Scan first. - Recording secrets. Never copy API keys, connection strings, or internal hostnames into the file; it gets committed. Record the service name only.
- Inferring scale from nothing. If there is no data, write
unknown— do not guess "10k users". Downstream skills treat numbers here as real. - Writing a design system from scratch here. This file points to the design source of truth;
design-systemscreates it. - Overwriting user edits. On update, merge. Never regenerate the whole file.
- Monorepos. Record each app's platform/stack separately under Platforms/Stack rather than averaging them.
Output format
- The file
.agents/project-context.mdfollowing the template. - A reply:
Recorded: …(3–5 bullets) ·Unknown: …(the questions that remain) ·Next: …(the skill the user most likely needs next).
Related skills
how-to-build-software— once context exists, route the actual task.design-systems— when the Design system section is empty or contradictory.software-architecture— when the Architecture section reveals no clear boundaries.
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
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.

