Mapwrite
Codex and Claude Code skill for agent-maintained project wikis, layered code indexes, and low-token repo navigation.
npx -y skills add xts-x-xvxl-wxrld/MapWrite --skill mapwriteAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Navigate repo-specific wiki protocols, code-index maps, and agent update rules. Use when Codex needs to work in a repository that has `wiki/index.md`, `wiki/code-index/`, `AGENTS.md`, or `CLAUDE.md`; when creating, updating, or following wiki specs/plans; when creating or maintaining nested code indexes; when deciding which project files to read before code changes; or when updating agent-facing project navigation instructions.
SKILL.md
3.6 KB, 762 tokens by cl100k_base, as published. Nobody here has run it
Mapwrite
Purpose
Use this skill to keep project context discovery small, layered, and current. Treat wiki indexes and code indexes as one navigation workflow with separate artifacts.
First Pass
- Read the repo's root agent instructions, usually
AGENTS.mdand/orCLAUDE.md. - Read
wiki/index.mdwhen it exists. - Read
wiki/code-index/index.mdwhen it exists and the task touches code. - Follow only the relevant narrow spec, plan, or code-index shard.
- Use
rgorrg --filesto verify symbols and files before opening large code files.
Do not read all wiki files, all logs, generated mega-indexes, caches, env files, session data, or test scratch directories unless the user explicitly asks for that history or artifact.
Artifact Roles
Keep these responsibilities separate:
wiki/index.md: wiki-only map for specs, plans, shards, runbooks, and high-level knowledge documents.wiki/code-index/index.md: top-level human-readable code navigation map.wiki/code-index/**/*.md: narrow maps for subsystems, entrypoints, facades, boundaries, and related tests.wiki/spec/**/*.md: behavior and architecture source of truth.wiki/plan/**/*.md: implementation sequence and acceptance criteria.wiki/log.md: append-only change history; inspect recent headings or tail only unless a historical task requires more..agent-index/: optional generated symbol or call-graph helpers, not a substitute for targeted reads andrgverification.
Code-Index Protocol
When creating or updating code indexes:
- Prefer a nested map over one large file.
- Start with
wiki/code-index/index.md, then add subsystem shards only as needed. - Keep each shard concise and navigational; do not turn it into a behavior spec.
- Include subsystem purpose, read-first files, entrypoints/facades, boundaries, related tests, and common spec/plan links.
- Add more shards when a parent file becomes too broad.
Recommended shape:
wiki/code-index/
index.md
backend.md
bot.md
tests.md
scripts.md
alembic.md
backend/
api.md
services.md
workers.md
db.md
bot/
engagement.md
search.md
accounts.md
formatting.md
Generated Index Policy
Generated symbol indexes are optional helper artifacts. Keep them outside the human-authored wiki
maps, preferably under .agent-index/, and exclude:
wiki/llm-wiki.mdwiki/llm-wiki-md.txtwiki/log.md.claude/- env files
- sessions, data volumes, caches, and pytest scratch directories
Never rely on generated indexes alone before editing. Verify with rg and targeted file reads.
After Changes
Update only the artifacts whose responsibility changed:
- Update
wiki/index.mdwhen specs, plans, shard directories, runbooks, or code-index entrypoints change. - Update the relevant
wiki/code-index/shard when code files, ownership boundaries, facades, entrypoints, or related test maps change. - Update relevant specs only when behavior or architecture changed.
- Append
wiki/log.mdfor code or wiki changes. - Run the repo's local validation gates from
AGENTS.mdorCLAUDE.md; if a gate cannot run, note why in the handoff and, when required by the repo, inwiki/log.md.
Gives 0 of the 12 instructions most memory context skills give in 762 tokens
Counted across 674 of the 847 authors here whose files we hold, read 2026-08-06
- inform the user when setup is completein 21 of 674, across 6 files
- confirm the draft with the user before writingin 21 of 674, across 6 files
- update the agent skills block in place if it existsin 21 of 674, across 6 files
- present findings to the userin 20 of 674, across 5 files
- write the three docs files from seed templatesin 20 of 674, across 5 files
- ask the user about each decision one at a timein 19 of 674, across 4 files
- edit CLAUDE.md if it existsin 18 of 674, across 3 files
- explore current repo statein 18 of 674, across 3 files
- do not overwrite user edits to surrounding sectionsin 18 of 674, across 3 files
- back up the original file before overwritingin 16 of 674, across 8 files
- keep the memory index under 200 linesin 15 of 674
- Provide actionable steps and verificationin 13 of 674, across 2 files
Said here and by no other author read
- follow only the relevant narrow artifact
- verify symbols and files before opening large code
- prefer nested maps over single large files
- keep each code-index shard concise and navigable
- add subsystem shards only as needed
- run local validation gates after changes
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.