drhalto/agentmako/mako-guide
Claude Code and agent skill by drhalto/agentmako
- community
Checking that the SKILL.md file is still in the repository.
Start here when working in a project with the mako-ai MCP server configured. Entry point that explains how mako's tool skills are organized and carries the policy for logging `agent_feedback` vs acknowledging static findings with `finding_ack`.
Install
npx vskill@latest install drhalto/agentmako/mako-guideSKILL.md
Mako Guide
Mako is a project-intelligence MCP server. Use it when the user asks structural, cross-surface, or evidence-backed engineering questions that are better answered from indexed code, schemas, routes, graph edges, prior sessions, or telemetry than from ad hoc file reads.
Do not use this skill as a schema reference. For current tool schemas, use MCP ToolSearch or the live tool definition loaded by Claude Code.
Skill Map
- Use
/mako-ai:mako-discoveryfor first-turn orientation, broad search, and
"which tool should I use?" questions.
- Use
/mako-ai:mako-tracefor targeted route, schema, file, table, RPC, edge,
auth, or error tracing.
- Use
/mako-ai:mako-neighborhoodsfor one-call table, route, or RPC context
bundles.
- Use
/mako-ai:mako-graphfor relationships, dependency paths, import impact,
symbol impact, flow mapping, and change planning.
- Use
/mako-ai:mako-databasefor direct database schema, RLS, foreign-key, or
RPC introspection.
- Use
/mako-ai:mako-code-intelfor structural AST pattern search and focused
lint diagnostics.
- Use
/mako-ai:mako-workflowfor investigation packets, artifacts, recall,
telemetry inspection, feedback, and finding acknowledgements.
General Routing Rules
- Read
_hintson every Mako tool result before deciding the next call. They
are generated from the actual returned result and may override the generic workflow expectation.
- Default to
reef_askfor project questions. It combines codebase, database,
findings, diagnostics, instructions, freshness, and quoted literal checks.
- Use
mako_helpwhen you need a task-specific Mako workflow recipe instead
of reading docs or guessing which specialist tool to load.
- Use
context_packet,cross_search,file_preflight, DB tools, and graph
tools as specialist follow-ups after reef_ask, mako_help, or tool_search points at a concrete need.
- Prefer Mako over built-in grep when the question is about relationships,
not literal text.
- Built-in text search is fine for exact strings inside known files.
- Start broad only when the target is unknown; switch to targeted tools once
you have a route, table, RPC, file, or symbol.
- Do not invent schemas, arguments, request IDs, finding fingerprints, or
telemetry events.
- If a tool name is uncertain, use
tool_searchbefore calling a Mako tool.
Feedback Logging Policy
Log agent_feedback selectively when a Mako result is notably useful, partially useful, noisy, stale, incomplete, wrong, or wasted the turn. Do not emit routine feedback after every tool call.
Before calling agent_feedback:
- Call
recall_tool_runs. - Filter by
toolNameand a recent ISO time window when possible. - Use a small
limitfirst. - Copy the recalled
requestIdintoreferencedRequestId. - Set
referencedToolNameto the tool being rated. - Use
grade: "full" | "partial" | "no". - Include concise
reasonCodesand a short human-readablereason.
If no requestId is available, do not fabricate one. Refine recall_tool_runs or skip feedback.
Use these starter reason codes exactly when they fit:
grade: "full":answer_complete,evidence_sufficient,
trust_matches
grade: "partial":partial_coverage,noisy,stale_evidence,
missing_known_caller, top_not_useful
grade: "no":answer_wrong,wasted_turn,tool_did_nothing,
schema_missing
The reason-code vocabulary is guidance, not a hard enum. New snake_case codes are acceptable when none of the starter codes fit.
Finding Acks Are Different
Use finding_ack for reviewed static findings, not for tool usefulness.
- Use
agent_feedbackto rate a specific Mako tool run. - Use
finding_ackto acknowledge a specific AST/lint/static finding as
reviewed, accepted-risk, false-positive, or otherwise handled.
- Do not use
finding_ackto rate search quality. - Do not use
agent_feedbackto suppress or accept a static finding.
Telemetry Inspection
Use agent_feedback_report and runtime_telemetry_report for inspection:
- confirm feedback capture worked;
- summarize tool quality over a window;
- debug whether a live session recorded events;
- prepare roadmap or implementation review notes.
Do not use telemetry reports as an automatic ranking system for tool selection unless Mako explicitly implements that behavior.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/codex/skills/mako-guide/SKILL.md
Popularity
- Repository stars
- 52
- Repository forks
- 4
- 7-day trend score
- 20
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/drhalto/agentmako/mako-guide)