code-wiki-ru
Skill0 starsChanged 4 months agoMITView on GitHub
Builds and validates Russian Obsidian repository wikis, architecture docs, and code/docs/ADR drift reports.
- Reads credentials
We split this skill into the rules it gives an agent
25 rules. About 0 are common, 19 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈19
Distinctive · estimate
This author's own choices. Worth a look.
6
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
- Very specific: line 8
Use this skill to create, refresh, validate, and safely apply Russian Obsidian-compatible code wikis under docs/wiki or a user-specified …
- Distinctive: line 13
Default variables used below:
- Distinctive: line 24
Do not read, print, summarize, store, or commit secrets.
- Very specific: line 25
Treat .env, .env.*, private keys, certificates, credential stores, tokens, cookies, auth files, production dumps, and credential-bearing …
- Distinctive: line 26
Keep code identifiers, paths, commands, package names, API names, config keys, schema names, and ADR titles exactly as written.
- Distinctive: line 27
Prefer deterministic scripts before model analysis.
- Distinctive: line 29
Do not ask DeepSeek or OpenCode to modify files, stage files, commit files, run deployment commands, or execute destructive commands.
- Distinctive: line 35
Read repository-local instructions before generating wiki content, especially AGENTS.md, README.md, ADR indexes, docs indexes, and …
- Distinctive: line 42
Run the indexer before planning or drafting.
- Distinctive: line 52
Run the planner after the index exists.
- Distinctive: line 58
Use the generated plan to choose the smallest chunk set that satisfies the request.
- Very specific: line 62
Read references/wiki-taxonomy.md before designing page structure.
- Very specific: line 63
Read references/drift-policy.md before classifying mismatches.
- Distinctive: line 64
Prefer verified source references over broad summaries.
- Distinctive: line 65
Use Russian prose for wiki pages.
- Very specific: line 66
Use Obsidian [[wiki-links]] for internal links and relative Markdown links for source docs and ADRs.
- Distinctive: line 67
Use Mermaid only when the available source context supports the diagram.
- Distinctive: line 68
Keep pages focused: one subsystem, component, flow, API surface, data concept, integration, operation, decision, or glossary topic per page.
- Distinctive: line 78
Before invoking OpenCode, read:
- Very specific: line 91
Do not ask DeepSeek to apply patches.
- Distinctive: line 103
Every drift candidate must include:
- Distinctive: line 111
If evidence is insufficient, do not invent drift.
- Distinctive: line 115
Run validation before applying staged pages or reporting completion:
- Distinctive: line 135
Do not stage or commit repository changes unless the user explicitly requests it.
- Distinctive: line 139
When resuming an interrupted wiki update:
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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
Code Wiki RU
Use this skill to create, refresh, validate, and safely apply Russian Obsidian-compatible code wikis under docs/wiki or a user-specified wiki path.
For a read-only question about an existing wiki or ADR, use wiki-query without
running the indexing, planning, or drafting workflow below.
Default variables used below:
SKILL_DIR="${SKILL_DIR:-$HOME/.codex/skills/code-wiki-ru}"
REPO_ROOT="${REPO_ROOT:-$(pwd)}"
WIKI_PATH="${WIKI_PATH:-docs/wiki}"
META_PATH="${META_PATH:-docs/wiki/_meta}"Safety
- Do not read, print, summarize, store, or commit secrets.
- Treat
.env,.env.*, private keys, certificates, credential stores, tokens, cookies, auth files, production dumps, and credential-bearing logs as redacted paths. - Keep code identifiers, paths, commands, package names, API names, config keys, schema names, and ADR titles exactly as written.
- Prefer deterministic scripts before model analysis.
- DeepSeek output is draft material only. Do not apply it until it is reviewed and validated by the host agent.
- Do not ask DeepSeek or OpenCode to modify files, stage files, commit files, run deployment commands, or execute destructive commands.
- Preserve user changes. Do not revert unrelated work in the repository.
Repository Inspection
- Identify
REPO_ROOTfrom the user request or current working directory. - Read repository-local instructions before generating wiki content, especially
AGENTS.md,README.md, ADR indexes, docs indexes, and architecture notes when present. - Inspect dependency/configuration files relevant to the repository type before writing architecture claims.
- Identify existing wiki location. Use
docs/wikiby default unless the user or repository config clearly points elsewhere. - Identify source documents and ADRs, but treat code as authoritative for runtime behavior.
Init/Index
Run the indexer before planning or drafting. It initializes wiki metadata and records bounded, non-secret source inventory.
python3 "$SKILL_DIR/scripts/build_repo_index.py" --repo "$REPO_ROOT" --wiki-path "${WIKI_PATH:-docs/wiki}" --meta-path "${META_PATH:-docs/wiki/_meta}"Expected output is metadata under ${META_PATH:-docs/wiki/_meta}. If the command fails, fix the invocation or report the failure before proceeding.
Plan
Run the planner after the index exists. It creates update chunks for bounded analysis.
python3 "$SKILL_DIR/scripts/plan_wiki_update.py" --repo "$REPO_ROOT" --wiki-path "${WIKI_PATH:-docs/wiki}" --meta-path "${META_PATH:-docs/wiki/_meta}"Use the generated plan to choose the smallest chunk set that satisfies the request. Keep unrelated chunks out of scope.
Analyze
- Read
references/wiki-taxonomy.mdbefore designing page structure. - Read
references/drift-policy.mdbefore classifying mismatches. - Prefer verified source references over broad summaries.
- Use Russian prose for wiki pages.
- Use Obsidian
[[wiki-links]]for internal links and relative Markdown links for source docs and ADRs. - Use Mermaid only when the available source context supports the diagram.
- Keep pages focused: one subsystem, component, flow, API surface, data concept, integration, operation, decision, or glossary topic per page.
Draft
For each selected CHUNK_ID, render a context pack:
python3 "$SKILL_DIR/scripts/render_context_pack.py" --repo "$REPO_ROOT" --wiki-path "${WIKI_PATH:-docs/wiki}" --meta-path "${META_PATH:-docs/wiki/_meta}" --chunk-id "$CHUNK_ID" --output "$CONTEXT_PACK"Before invoking OpenCode, read:
references/opencode-deepseek.mdreferences/prompt-contracts.md
Dispatch only through the verified local tests-docs worker route described in
the reference. Send the explicit export-authorized context through bounded
stdin or a supported private file, never a large prompt argument. Keep the
worker outside the repository with no source editing, shell, inherited MCP or
parent auth access. Select provider/model from private runtime configuration;
report BLOCKED when that exact route has not been verified. Preserve requested
and observed metadata, checks, and artifacts using subagent-result-merge.
Do not ask DeepSeek to apply patches. The parent reviews model-authored drafts
before staging them under ${META_PATH:-docs/wiki/_meta}/drafts/, validates
them, and owns application to the wiki.
Drift
Classify drift using references/drift-policy.md:
code-docs: docs disagree with verified runtime code.code-adr: code no longer follows accepted ADR intent.docs-adr: docs disagree with ADR intent.
Every drift candidate must include:
- drift class
- severity
- source file references
- short evidence summary
- recommended action
If evidence is insufficient, do not invent drift. State that the available context does not confirm a mismatch.
Validate
Run validation before applying staged pages or reporting completion:
python3 "$SKILL_DIR/scripts/validate_wiki.py" --repo "$REPO_ROOT" --wiki-path "${WIKI_PATH:-docs/wiki}" --meta-path "${META_PATH:-docs/wiki/_meta}"Validation should check structure, links, required metadata, and staged output consistency. Report exact command output or the failure reason.
Apply
Apply drafted wiki changes only after the user explicitly asks for application or the current task clearly requires it.
Before applying:
- Review staged drafts in
${META_PATH:-docs/wiki/_meta}/drafts/. - Confirm proposed page paths match
references/wiki-taxonomy.md. - Confirm drift findings include source references and recommended actions.
- Confirm no secret-bearing content is present.
- Re-run validation after applying.
Do not stage or commit repository changes unless the user explicitly requests it.
Resume
When resuming an interrupted wiki update:
- Re-read repository-local instructions.
- Re-run the index command to refresh metadata.
- Re-run the plan command and compare the current chunks with any existing drafts.
- Continue from the smallest unfinished chunk.
- Validate before applying or reporting completion.
Discard stale drafts only when they are proven obsolete or the user authorizes cleanup.
References
references/wiki-taxonomy.md: wiki hierarchy, page naming, links, diagrams, and page focus rules.references/drift-policy.md: source-of-truth order, drift classes, severity levels, and required finding fields.references/opencode-deepseek.md: safe OpenCode invocation and DeepSeek handling rules.references/prompt-contracts.md: required Markdown sections and staged-output contract for model drafts.
In this repository62 other files it gives its agents.
- AGENTS.mdAGENTS.md
- accessibility-ui-reviewSkill
- agent-retrospectiveSkill
- agents-md-retrospectiveSkill
- agent-squadSkill
- architecture-mapSkill
- architecture-normalizerSkill
- architecture-principlesSkill
- authz-boundary-reviewSkill
- c-cppSkill
- code-qualitySkill
- code-reviewSkill
- code-to-figmaSkill
- context-packSkill
- core-engineeringSkill
- debuggingSkill
- dependency-advisory-auditSkill
- deserialization-parser-reviewSkill
- design-system-extractorSkill
- documentationSkill
- engineering-standardsSkill
- esp32Skill
- esphomeSkill
- external-agent-pack-auditSkill
- fastSkill
- figma-to-codeSkill
- fix-security-findingSkill
- goSkill
- homeassistantSkill
- mcp-tool-routerSkill
- migration-plannerSkill
- mobile-qaSkill
- multi-agent-pr-reviewSkill
- performanceSkill
- playwright-visual-qaSkill
- pythonSkill
- quality-gatesSkill
- refactoringSkill
- responsive-breakpoint-checkSkill
- review-routerSkill
- rustSkill
- secrets-and-config-reviewSkill
- security-diff-reviewSkill
- security-routerSkill
- securitySkill
- session-memorySkill
- specialist-dispatchSkill
- subagent-result-mergeSkill
- supply-chain-reviewSkill
- tdd-guardSkill
- testing-tddSkill
- threat-modelSkill
- typescriptSkill
- ui-buildSkill
- ui-business-appsSkill
- ui-concept-firstSkill
- ui-figmaSkill
- ui-qaSkill
- ui-researchSkill
- ui-routerSkill
- The first 60 shown.
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.

