feat: add context-audit skill (Vesely/skills)#80
Merged
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Ports Vesely/skills/context-audit into the ulk skill ecosystem. Audits MCP servers, CLAUDE.md bloat, skills size, settings gaps, and missing .claudeignore — returns a 0-100 health score with prioritised fixes and optional auto-apply. - community-skills/context-audit/SKILL.md (5-step audit framework) - install.sh: _install_context_audit_skill() → ~/.claude/skills/context-audit/ - CLAUDE.md: listed under Community Skills Invoke with /context-audit after running /context. https://claude.ai/code/session_01GEeL7L13c3xJMNxGPFm6up
Ajoute l'agent 53-context-audit dans agents/session/ : 7 phases (prérequis /context, MCP, CLAUDE.md, skills, settings, housekeeping, score 0-100, auto-fix optionnel). Complète la skill /context-audit (community-skills/) — l'agent expose le même audit via /ulk:context-audit avec le protocole ulk (phases, scoring, rapport formaté, auto-fix guidé). - agents/session/53-context-audit.md - agents/registry.json + registry.md régénérés (74 agents) - CLAUDE.md : prochain numéro → 54 https://claude.ai/code/session_01GEeL7L13c3xJMNxGPFm6up
- README.md : ligne context-audit (1) dans la table Community Skills - install.sh : help · wizard · résumé · options summary mis à jour - site/install.astro : mention context-audit dans les skills bundlés https://claude.ai/code/session_01GEeL7L13c3xJMNxGPFm6up
Les agents 53 (routine) et 54 (ci-guard) étaient déjà pris dans main. - agents/session/53-context-audit.md → 55-context-audit.md - Titre interne mis à jour (53→55) - CLAUDE.md : Context Audit (55) + prochain → 56 - registry.json + registry.md régénérés (76 agents) https://claude.ai/code/session_01GEeL7L13c3xJMNxGPFm6up
izo
force-pushed
the
claude/audit-context-usage-UJw2a
branch
from
68e2b00 to
471b481
Compare
📝 WalkthroughWalkthroughThis PR introduces a new Changes
Sequence Diagram(s)sequenceDiagram
actor User
participant Agent as Context-Audit<br/>Agent
participant FS as File System
participant Settings as Settings<br/>System
participant Scorer as Score<br/>Calculator
User->>Agent: /ulk:context-audit<br/>(trigger audit)
Agent->>FS: Request /context output
FS-->>Agent: Context data
Agent->>Settings: Read ~/.claude/settings.json
Settings-->>Agent: MCP server config
Agent->>Agent: Classify servers<br/>(idle/redundant)
Agent->>FS: Scan CLAUDE.md files
FS-->>Agent: File contents & line counts
Agent->>Agent: Filter by rules<br/>(5 categories)
Agent->>FS: List ~/.claude/skills/
FS-->>Agent: Installed skills
Agent->>FS: Check SKILL.md sizes
FS-->>Agent: Skill metadata
Agent->>FS: Check housekeeping<br/>(.claudeignore, memory)
FS-->>Agent: Housekeeping status
Agent->>Scorer: Deductions & thresholds
Scorer-->>Agent: Health score (0–100)
Agent->>User: Structured report<br/>+ auto-fix options
User->>Agent: Confirm auto-fix
Agent->>FS: Apply safe changes
FS-->>Agent: Changes applied
Agent->>User: Results & diffs
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
community-skills/context-audit/SKILL.md (1)
55-59: Make fenced blocks markdownlint-compliant.Add code fence languages and keep blank lines around the health-label table to avoid lint warnings.
🧹 Proposed markdown cleanup
-Find all CLAUDE.md files: -``` +Find all CLAUDE.md files: +```text ~/.claude/CLAUDE.md (global) <project>/CLAUDE.md (project) <project>/.claude/rules/*.md (path-specific rules)@@
-Health labels:
-| Score | Label |
+Health labels:
+
+| Score | Label |
|-------|-------|
| 90–100 | ✅ CLEAN |
| 70–89 |⚠️ NEEDS WORK |
| 50–69 | 🔴 BLOATED |
| 0–49 | 💀 CRITICAL |
@@
-Output a structured report:-
+Output a structured report: + +text
CONTEXT AUDIT —
══════════════════════════════════════════
@@
-## Reference — Standard .claudeignore-
+## Reference — Standard .claudeignore + +gitignore
node_modules/
dist/
build/</details> Also applies to: 141-143, 154-184, 211-227 <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@community-skills/context-audit/SKILL.mdaround lines 55 - 59, The markdown
has unlanguaged fenced code blocks and missing blank lines around the
health-label table which triggers markdownlint warnings; update each unlanguaged
fence instances (e.g., the "Find all CLAUDE.md" block, the "Output a structured
report" block, and the "## Reference — Standard .claudeignore" block) to include
an appropriate language tag such astext orgitignore, and ensure there is
a blank line before and after the health labels table (the "| Score | Label |"
table) and any other tables mentioned (also apply the same fenced-language and
surrounding blank-line fixes to the other affected sections noted in the
comment).</details> </blockquote></details> <details> <summary>agents/session/55-context-audit.md (1)</summary><blockquote> `1-17`: **Frontmatter and placement follow agent conventions.** `model: sonnet`, `phase: review`, and the `agents/session/` location are consistent for this new agent. As per coding guidelines: Agent `model` field should be set to 'opus' only for a fixed allowlist; all other agents should use 'sonnet', and agent frontmatter must include a `phase:` field. <details> <summary>🤖 Prompt for AI Agents</summary> ``` Verify each finding against the current code and only fix it if needed. In `@agents/session/55-context-audit.md` around lines 1 - 17, The frontmatter must use the allowed model and include a phase: verify that the agent's frontmatter contains model: sonnet (not opus) and includes a phase: field; for this agent ensure name: context-audit and invocation remain unchanged and that model: sonnet and phase: review are present in the frontmatter so it complies with the agent allowlist rules and frontmatter requirements (update model or add phase only if missing). ``` </details> </blockquote></details> </blockquote></details> <details> <summary>🤖 Prompt for all review comments with AI agents</summary>Verify each finding against the current code and only fix it if needed.
Inline comments:
In@install.sh:
- Around line 607-616: install_context_audit_skill currently copies the skill
into "$SKILLS_DIR/context-audit" which doesn't match the prefixed glob patterns
used by cmd_uninstall and cmd_check, leaving the skill orphaned; update the
install to place the directory with the same prefix used by other community
skills (e.g., copy to "$SKILLS_DIR/community-context-audit" and adjust the
printf label), or alternatively update cmd_uninstall and cmd_check to include
the non-prefixed "context-audit" in their globs — refer to
install_context_audit_skill, cmd_uninstall, cmd_check, SKILL.md and the
"$SKILLS_DIR/context-audit" target when making the change.
Nitpick comments:
In@agents/session/55-context-audit.md:
- Around line 1-17: The frontmatter must use the allowed model and include a
phase: verify that the agent's frontmatter contains model: sonnet (not opus) and
includes a phase: field; for this agent ensure name: context-audit and
invocation remain unchanged and that model: sonnet and phase: review are present
in the frontmatter so it complies with the agent allowlist rules and frontmatter
requirements (update model or add phase only if missing).In
@community-skills/context-audit/SKILL.md:
- Around line 55-59: The markdown has unlanguaged fenced code blocks and missing
blank lines around the health-label table which triggers markdownlint warnings;
update each unlanguaged fence instances (e.g., the "Find all CLAUDE.md" block,
the "Output a structured report" block, and the "## Reference — Standard
.claudeignore" block) to include an appropriate language tag such as ```text orlabels table (the "| Score | Label |" table) and any other tables mentioned (also apply the same fenced-language and surrounding blank-line fixes to the other affected sections noted in the comment).🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID:
1105e755-4f40-42e1-90c7-aac74d06dd4e📒 Files selected for processing (8)
CLAUDE.mdREADME.mdagents/registry.jsonagents/registry.mdagents/session/55-context-audit.mdcommunity-skills/context-audit/SKILL.mdinstall.shsite/src/pages/install.astro
Comment on lines
+607
to
+616
| _install_context_audit_skill() { | ||
| local src="$SCRIPT_DIR/community-skills/context-audit" | ||
| if [ ! -f "${src}/SKILL.md" ]; then | ||
| printf " ${WARN} %-14s ${D}community-skills/context-audit/SKILL.md not found${NC}\n" "context-audit" | ||
| return | ||
| fi | ||
| run rm -rf "$SKILLS_DIR/context-audit" | ||
| run cp -r "$src" "$SKILLS_DIR/context-audit" | ||
| printf " ${OK} %-14s ${D}/context-audit — token waste + health score${NC}\n" "context-audit" | ||
| } |
There was a problem hiding this comment.
context-audit gets installed, but uninstall/check flows don’t track it.
This installs to ~/.claude/skills/context-audit, but cmd_uninstall and cmd_check only handle prefixed globs, so this default skill becomes orphaned after uninstall and invisible in diagnostics.
🔧 Proposed fix
@@
# Flutter skills (community)
local n_flutter=0
for d in "$SKILLS_DIR"/flutter-*/; do
@@
if [ $n_flutter -gt 0 ]; then
printf " ${OK} Removed ${D}%s Flutter skills${NC}\n" "$n_flutter"
removed=$((removed + 1))
fi
+
+ # context-audit skill (community, default)
+ if [ -d "$SKILLS_DIR/context-audit" ]; then
+ run rm -rf "$SKILLS_DIR/context-audit"
+ printf " ${OK} Removed ${D}context-audit skill${NC}\n"
+ removed=$((removed + 1))
+ fi
@@
- for d in "$SKILLS_DIR"/ulk-*/ "$SKILLS_DIR"/addy-*/ "$SKILLS_DIR"/a11y-*/ "$SKILLS_DIR"/figma-*/ "$SKILLS_DIR"/kepano-*/ "$SKILLS_DIR"/nothing-*/ "$SKILLS_DIR"/swift-*/ "$SKILLS_DIR"/flutter-*/; do
+ for d in "$SKILLS_DIR"/ulk-*/ "$SKILLS_DIR"/addy-*/ "$SKILLS_DIR"/a11y-*/ "$SKILLS_DIR"/figma-*/ "$SKILLS_DIR"/kepano-*/ "$SKILLS_DIR"/nothing-*/ "$SKILLS_DIR"/swift-*/ "$SKILLS_DIR"/flutter-*/ "$SKILLS_DIR"/context-audit/; do🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@install.sh` around lines 607 - 616, install_context_audit_skill currently
copies the skill into "$SKILLS_DIR/context-audit" which doesn't match the
prefixed glob patterns used by cmd_uninstall and cmd_check, leaving the skill
orphaned; update the install to place the directory with the same prefix used by
other community skills (e.g., copy to "$SKILLS_DIR/community-context-audit" and
adjust the printf label), or alternatively update cmd_uninstall and cmd_check to
include the non-prefixed "context-audit" in their globs — refer to
install_context_audit_skill, cmd_uninstall, cmd_check, SKILL.md and the
"$SKILLS_DIR/context-audit" target when making the change.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Ports Vesely/skills/context-audit into the ulk skill ecosystem.
Audits MCP servers, CLAUDE.md bloat, skills size, settings gaps,
and missing .claudeignore — returns a 0-100 health score with
prioritised fixes and optional auto-apply.
Invoke with /context-audit after running /context.
https://claude.ai/code/session_01GEeL7L13c3xJMNxGPFm6up
Summary by CodeRabbit
context-auditagent and community skill for auditing Claude Code setup efficiency.