easyeda-mcp-pro
AGENTS.md67 starsChanged 10 days agoMITView on GitHub
These instructions apply to the entire repository. Prefer repository-owned commands and policy files over inferred conventions. - This root file applies repository-wide. - Nested AGENTS.md files add or narrow instructions for their subtree. - When instructions conflict, the closest applicable AGENTS.md wins. - Nested instructions must not weaken repository security, approval, evidence, release, or product claim constraints unless the underlying policy is intentionally changed in the same work. - Executable repository policy and tests remain authoritative if prose drifts. Nested…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository11 other files it gives its agents.
- easyeda-bridge-extension/AGENTS.mdAGENTS.md
- .github/AGENTS.mdAGENTS.md
- easyeda-component-reviewSkill
- easyeda-professional-layoutSkill
- easyeda-component-reviewSkill
- easyeda-professional-layoutSkill
- component-searchSkill
- design-validationSkill
- easyeda-component-reviewSkill
- easyeda-professional-layoutSkill
- easyeda-workflowSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
16 rules. About 0 are common, 13 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈13
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 10
When instructions conflict, the closest applicable AGENTS.md wins.
- Very specific: line 71
Before handoff or PR creation: run pnpm verify.
- Distinctive: line 80
Read-only inspection may proceed without mutation.
- Very specific: line 83
Never bypass confirmWrite, remote approval, scope checks, dependency-audit policy, release-age policy, secret hygiene, or branch protection …
- Distinctive: line 85
Treat timeouts during live writes as ambiguous state: read back before any retry.
- Very specific: line 86
Do not claim manufacturing approval, fabrication readiness, supplier approval, or electrical correctness solely from automated …
- Distinctive: line 98
update the canonical skill first.
- Distinctive: line 99
update or regenerate every applicable runtime distribution.
- Distinctive: line 101
update the relevant distribution tests and agent-runtime validation when needed.
- Distinctive: line 103
Do not infer product capability from a skill or environment-variable name alone.
- Distinctive: line 108
Do not hand-edit generated outputs when a repository generator/check command owns them.
- Distinctive: line 111
Keep tool/reference documentation, compatibility metadata, capability counts, runtime metadata, package artifacts, and extension artifacts …
- Distinctive: line 118
Keep changes scoped and add regression tests for bug fixes or behavior changes.
- Distinctive: line 120
Avoid moving security decisions into transport adapters or documentation-only groupings.
- Distinctive: line 123
Keep the working tree understandable: inspect git diff and git status before handoff.
- Distinctive: line 136
no security, approval, release, provenance, or manufacturing claim was silently weakened.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
Agent Repository Router
These instructions apply to the entire repository. Prefer repository-owned commands and policy files over inferred conventions.
Scope and precedence
- This root file applies repository-wide.
- Nested
AGENTS.mdfiles add or narrow instructions for their subtree. - When instructions conflict, the closest applicable
AGENTS.mdwins. - Nested instructions must not weaken repository security, approval, evidence, release, or product claim constraints unless the underlying policy is intentionally changed in the same work.
- Executable repository policy and tests remain authoritative if prose drifts.
Nested instruction boundaries:
src/tools/AGENTS.md— MCP tool metadata, schemas, mutation safety, profiles, scopes, and transaction behavior.src/remote/AGENTS.md— remote gateway, pairing, approval, session routing, scopes, replay protection, and audit behavior.easyeda-bridge-extension/AGENTS.md— EasyEDA browser extension runtime, local/remote bridge transport, mutation execution, hot-swap, packaging, and live verification..github/AGENTS.md— CI, branch-protection interfaces, security automation, release publication, provenance, and registry promotion.
Start here
README.mdfor product scope and local setup.CONTRIBUTING.mdfor coding, testing, branching, and review rules.SECURITY.mdanddocs/security-architecture.mdfor trust boundaries.docs/SAFETY_MODEL.mdfor EasyEDA mutation safety.docs/TOOL_APPROVAL_POLICY.mdfor remote approval semantics.docs/agent-runtime-config.mdfor MCP-capable agent runtimes.docs/benchmark-suite.mdfor mocked golden evals and their limits.docs/RELEASE_POLICY.mdfor release-channel, evidence, and recovery rules.
Repository map
src/: MCP server, tools, bridge client, remote transport, EDA analysis, and workflows.src/tools/: registered MCP tool definitions and execution-policy metadata.src/remote/: hosted/self-hosted remote gateway and approval/session boundary.easyeda-bridge-extension/: EasyEDA Pro browser/runtime extension workspace package.tests/: server, policy, security, integration, and eval coverage..github/workflows/: protected CI, security, release, and scheduled automation.scripts/: repository-owned verification, release, generation, and E2E helpers.skills/: canonical product-owned agent skills..agents/skills/,.claude/skills/,.codex/skills/, and.opencode/skills/: runtime-specific skill distributions or mirrors.
Runtime contract
The supported development runtime is intentionally pinned.
- Node.js:
24.22.1 - pnpm:
11.28.8
Start with:
node scripts/check-runtime.mjs --require-pnpmRuntime upgrades are atomic repository-wide changes. Do not update only packageManager, a
workflow, Docker metadata, .node-version, .nvmrc, or one documentation surface. Follow the
runtime parity procedure in CONTRIBUTING.md and update every surface reported by the repository
checkers in one focused change.
Verification contract
- During iteration: run targeted tests, then
pnpm verify:fast. - Before handoff or PR creation: run
pnpm verify. pnpm security:auditdepends on the package registry and may fail or hang for environmental reasons; do not classify that as a repository defect without advisory evidence.- Live EasyEDA tests are opt-in and require a connected disposable project; never substitute mocked golden evals for live-behaviour proof.
- For a narrow subtree change, also follow the closest nested
AGENTS.mdverification guidance.
Safety rules
- Read-only inspection may proceed without mutation.
- Require explicit user approval before the first live EasyEDA write, then preview/apply/read back and verify the result.
- Never bypass
confirmWrite, remote approval, scope checks, dependency-audit policy, release-age policy, secret hygiene, or branch protection to make a check pass. - Treat timeouts during live writes as ambiguous state: read back before any retry.
- Do not claim manufacturing approval, fabrication readiness, supplier approval, or electrical correctness solely from automated DRC/ERC/BOM/export results.
- Beta or version-sensitive EasyEDA APIs require explicit capability handling and must fail closed when the required runtime surface is unavailable.
Agent skills
skills/ is the canonical repository source for product-owned skill contracts.
Runtime-specific skill copies or wrappers under .agents/, .claude/, .codex/, and
.opencode/ must not become independent sources of truth. When a canonical skill changes:
- update the canonical skill first;
- update or regenerate every applicable runtime distribution;
- preserve equivalent safety invariants across distributions;
- update the relevant distribution tests and agent-runtime validation when needed.
Do not infer product capability from a skill or environment-variable name alone. Runtime capabilities, profiles, maturity, and feature flags remain authoritative.
Generated and policy-owned state
- Do not hand-edit generated outputs when a repository generator/check command owns them.
- Repository policy scripts are enforcement code, not disposable tooling. Do not weaken a checker merely to make CI pass.
- Keep tool/reference documentation, compatibility metadata, capability counts, runtime metadata, package artifacts, and extension artifacts synchronized through repository-owned commands.
- Preserve SHA-pinned GitHub Actions and dependency/runtime pins unless the change explicitly updates them.
Change discipline
- Keep changes scoped and add regression tests for bug fixes or behavior changes.
- Respect architecture boundaries enforced by
pnpm check:architecture. - Avoid moving security decisions into transport adapters or documentation-only groupings.
- Public MCP tool names, schemas, bridge protocol fields, environment variables, configuration keys, and CLI behavior are compatibility surfaces; follow deprecation policy before breaking them.
- Keep the working tree understandable: inspect
git diffandgit statusbefore handoff.
Definition of done
A change is complete only when:
- the implementation matches the documented trust and architecture boundaries;
- relevant focused tests pass;
- generated/policy-owned state is synchronized;
pnpm verify:fastpasses during iteration;pnpm verifypasses before handoff or PR creation unless a clearly identified environmental dependency prevents it;- live-behavior claims are backed by live evidence rather than mocked fixtures;
- no security, approval, release, provenance, or manufacturing claim was silently weakened.
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.

