| name | project-dev-context |
| description | Route EmbodiChain development-context and codebase-navigation requests through agent_context/MAP.yaml. Use when asked to locate files, configs, defaults, entry points, registration paths, or change sites; explain a code or configuration resolution chain; reference project context; refresh or write context; register a context topic; or work with a named topic such as simulation-system, env-framework, rl-learning, manager-functor, ik-solvers, or atomic-actions. Chinese triggers include 文件在哪里、配置或默认值在哪里、 入口或注册逻辑在哪里、应该修改哪个文件、参考项目上下文、刷新项目上下文。
|
Project Development Context and Codebase Navigation
Start here
- Read
agent_context/MAP.yaml first
- Read
references/context-system.md for routing rules
- Read
agents/openai.yaml for the canonical agent metadata
- Read
agent_context/conventions/*.md when creating or updating context files
Select the operation
- navigate: locate a file, symbol, config, default, entry point,
registration path, or recommended change site.
- read: load the matched agent context without changing it.
- refresh: rebuild an existing topic from its current
source_of_truth.
- add: create and register a new topic from current source code.
An explicit refresh or add request determines the mode. If the user asks to
implement work covered by a specialized skill such as add-robot,
add-solver, or add-functor, let that skill own the implementation and
use this skill only for orientation or mapped context.
Route the request
- Resolve the topic through
agent_context/MAP.yaml.
- Match in this order: exact
id, then aliases, then keywords.
- For a matched read request, load only the Markdown files in
paths.
- For a matched navigation request, load the mapped topic and verify the
relevant path or behavior against the current
source_of_truth.
- For an unmatched navigation request:
- list candidate files with
rg --files;
- search symbols, flags, config keys, registries, and imports with
rg -n;
- inspect the closest package
__init__.py, config loader, registry,
test, and entry point as relevant;
- inspect
pyproject.toml and embodichain/__main__.py for CLI or
package-discovery questions.
- Do not add a topic merely because navigation did not match. Propose or add
one only when the user requests it or the missing area is recurring.
- Do not read
docs/source/ unless the user explicitly asks for Sphinx
documentation.
Never treat topic Markdown as a substitute for current code. For navigation,
report only paths and behavior verified in the working tree.
Navigation answer contract
Include the parts relevant to the request:
- the entry point or owning location;
- the call, registration, or configuration-resolution path;
- the file or symbol to change;
- the focused tests or documentation affected by that change.
Explicit refresh mode
Use explicit refresh mode when the request is phrased like:
refresh <topic> context
update <topic> context
根据当前实现刷新 <topic> 上下文
重写 <topic> 项目上下文
In refresh mode:
- Resolve the topic in
agent_context/MAP.yaml
- Re-read the files listed in
source_of_truth
- Rewrite the mapped topic Markdown from current implementation, not stale notes
- Update
aliases, keywords, paths, related_topics if needed
- Load and follow
agent_context/conventions/*.md
Add mode
- Choose a stable kebab-case topic id.
- Read the current source files that define the topic.
- Write one focused Markdown file under
agent_context/topics/<topic-id>/.
- Register the topic in
agent_context/MAP.yaml.
- Load and follow
agent_context/conventions/*.md.
Update contract
If code behavior changes a routed topic, update all relevant pieces in the same change:
- the matching file under
agent_context/topics/...
agent_context/MAP.yaml if topic metadata changed
AGENTS.md if routing guidance changed
.agents/skills/project-dev-context/references/context-system.md if routing behavior changed
.claude/skills/project-dev-context/SKILL.md if Claude adapter wording changed
.github/copilot/project-dev-context.md if Copilot adapter wording changed
Source-of-truth
This skill stores the routing procedure, not project facts. Canonical project
context lives in:
agent_context/MAP.yaml
agent_context/topics/**/*.md
agent_context/conventions/*.md
Map schema
agent_context/MAP.yaml topic entry fields:
id — stable kebab-case identifier
title — human-readable title
aliases — alternate names for matching
keywords — search terms for fuzzy matching
paths — Markdown files under agent_context/ to load
source_of_truth — source code files that define the behavior
related_topics — other topic ids for cross-reference
status — active or deprecated