devin-config
Skill0 starsChanged 3 months agoMITView on GitHub
Use when evolving the agent's capabilities with a new skill, custom subagent, plugin, hook, MCP server, or project rule — or when auditing/managing a project's `.devin/` configuration, producing deterministic read-only reports, detecting broken references/duplicates/divergences, and generating plans that persist under `.devin/` only after approval.
In this repository110 other files it gives its agents.
- AGENTS.mdAGENTS.md
- skills/project-bootstrap/templates/agents.mdAGENTS.md
- a11y-auditSkill
- agentic-ragSkill
- agent-reachSkill
- ai3d-genSkill
- ai-coding-dictionarySkill
- ai-toolsSkill
- animation-directionSkill
- api-specSkill
- architecture-diagramsSkill
- architectureSkill
- ascii-artSkill
- ask-bundleSkill
- binary-triageSkill
- bragSkill
- brag-slimSkill
- career-opsSkill
- code-reviewSkill
- computer-useSkill
- context7Skill
- context-hygieneSkill
- creative-engineeringSkill
- cu-realtimeSkill
- data-analystSkill
- databaseSkill
- dead-wireSkill
- debuggingSkill
- deploySkill
- dispatching-parallel-agentsSkill
- dmaSkill
- dockerSkill
- drift-reportSkill
- e2e-testingSkill
- edge-casesSkill
- executionSkill
- fact-checkSkill
- finishing-a-development-branchSkill
- firmware-probeSkill
- frontend-designSkill
- gatesSkill
- ghidraSkill
- ghSkill
- github-actionsSkill
- git-workflowsSkill
- grillingSkill
- handoffSkill
- humanizerSkill
- hyperframesSkill
- i18nSkill
- image-forensicsSkill
- impeccableSkill
- implement-layaSkill
- intakeSkill
- jiraSkill
- knowledge-modelingSkill
- laya-orchestratorSkill
- learn-modeSkill
- mcp-governanceSkill
- media-toolsSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
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
- Very specific: line 26
Before writing any skill/rule, follow writing-skills conventions (frontmatter, discovery keywords, completion criteria).
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Devin Config
Extend Devin CLI (devin-config) and audit .devin/ (devin-config). Files go
to the project (./.devin/) or global config (~/.config/devin/ /
%APPDATA%\devin\). Most changes load at next session/reload.
What to create for each need
| I want to... | Create | Location | Reload |
| Reusable invocable workflow | Skill | .devin/skills/<name>/SKILL.md or ~/.config/devin/skills/<name>/SKILL.md | next skill list |
| Specialized worker profile | Custom subagent | .devin/agents/<name>.md or <name>/AGENT.md (global: ~/.config/devin/agents/) | next session |
| Always-on context | Rules | .devin/global_rules.md, .devin/rules/*.md (global: ~/.config/devin/AGENTS.md) | session start |
| Lifecycle logic | Hooks | .devin/hooks.v1.json | next session |
| API/DB tools | MCP server | .devin/mcp_config.json or ~/.config/devin/mcp_config.json | next session |
| Distributable bundle | Plugin | .devin-plugin/plugin.json + skills/rules/hooks/mcp | install/reload |
Creating skills
Before writing any skill/rule, follow writing-skills conventions
(frontmatter, discovery keywords, completion criteria).
---
name: my-skill
description: Use when [specific triggering conditions]
---Optional frontmatter: allowed-tools, subagent: true, agent: <profile>,
model (free variants only — swe-2-max/swe-2-medium, never paid),
permissions (allow/deny), triggers (default [user, model]).
Discover: skill list --path ., skill search --path . --keywords "...".
Invoke: /my-skill.
Custom subagents
.devin/agents/reviewer.md or .devin/agents/reviewer/AGENT.md:
---
name: reviewer
---
You are a careful code reviewer. Report concrete issues with file:line only.Reference via skill frontmatter agent: reviewer, or dispatch a
subagent_general/researcher subagent (never subagent_explore on a free
parent — it may route to a paid model).
Rules
.devin/global_rules.md + .devin/rules/*.md load by default — keep them
small, detail lives in skills. .devin/rules/*.md support trigger
frontmatter (always_on, manual, model_decision, agent, glob).
Hooks
.devin/hooks.v1.json. Events: PreToolUse, PostToolUse,
PermissionRequest, UserPromptSubmit, Stop, SessionStart,
SessionEnd, PostCompaction. Command hooks run a shell command; prompt
hooks run an LLM evaluation per event.
MCP servers
.devin/mcp_config.json (project) / ~/.config/devin/mcp_config.json
(global). Audit cost and trust review first with mcp-governance.
Auditing .devin/ (devin-config mode)
Deterministic, read-only ops via scripts/devin-manager.py (this skill's
scripts/). Notes persist only under .devin/notes/devin-config/ and only
with --write --approve.
| Op | Purpose |
scan | Inventory .devin/ + agents/ + mcp_config.json with sha256, provenance, frontmatter, references |
explain | One artifact and its outgoing references |
diff | Compare two .devin/ states by hash |
doctor | Broken refs, duplicate content/skill names, config/hook divergences |
plan | Plan note from doctor findings |
python skills/devin-config/scripts/devin-manager.py scan [PROJECT]
python skills/devin-config/scripts/devin-manager.py doctor [PROJECT]
python skills/devin-config/scripts/devin-manager.py diff [A] [B]Approval rule: editing .devin/config.json, hooks, or memory requires
explicit user approval — reports are read-only by default.
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.