Refacil:explore

refacil:explore — Exploration Entrypoint

This skill is a thin wrapper that delegates the investigation to the refacil-investigator sub-agent. The sub-agent runs in an isolated context (does not saturate your main session with massive file reads) and returns a concise report with architecture, flows, and recommendations.

Prerequisites: sdd profile from refacil-prereqs/SKILL.md — use AGENTS.md as active context throughout the entire exploration.

Flow

Step 0: Validate question

• If $ARGUMENTS is empty, ask the user for the question or topic to explore BEFORE invoking the sub-agent.
• If there is a question, continue.

Step 0.1: Duplicate exploration guard (CA-11)

Before delegating, check the current session conversation context for a prior complete exploration report with overlapping scope (same modules, files, or question topic):

• If a prior complete exploration exists for the same or highly overlapping topic: summarize the already-known context in 2-3 sentences and ask:

`` I already explored [topic] earlier in this session. The key findings were: [summary]. Do you want me to run a targeted follow-up on a specific aspect, or proceed with a new full exploration? `` Wait for the user's answer before proceeding.

• If there is no prior exploration (or it is on a clearly different topic): continue to Step 1 without interruption.

Step 0.2: CodeGraph availability check

Run refacil-sdd-ai codegraph status --json and parse the output. Set codegraphAvailable:

• true if installed === true AND initialized === true
• false otherwise

If mode is enabled or per-repo but codegraphAvailable is false:

---

[GATE — STOP AND WAIT: CodeGraph not indexed]

Do NOT proceed to Step 1. Do NOT delegate to the sub-agent. STOP and ask the user exactly this, then wait for their reply:

CodeGraph is enabled for this repo but not yet indexed (~30s to build).
Without it, the exploration uses Grep/Read — slower and less precise for call-graph questions.

Do you want me to run /refacil:update first to build the index?
  yes → run /refacil:update then continue with the exploration
  no  → continue with standard exploration now

Only after the user replies:

• "yes" (or "si", "dale", "sí", "go", "ok"): immediately execute /refacil:update. Do not describe it in text.
• "no" (or "skip", "continue", "continúa", "así"): set codegraphAvailable: false and proceed to Step 1.

---

If mode is disabled or null: set codegraphAvailable: false and continue to Step 1 without asking.

Step 1: Delegate to the refacil-investigator sub-agent

Invoke the refacil-investigator sub-agent passing it:

• The user's question ($ARGUMENTS).
• codegraphAvailable: true | false (from Step 0.2).
• If the user explicitly requested detailed mode, indicate it (mode: detailed). Default: concise.

The sub-agent:

• Explores the codebase natively (Read/Grep with discipline — see sub-agent rules).
• Enriches with patterns and conventions from AGENTS.md.
• Detects cross-repo dependencies and, if appropriate, consults the bus according to refacil-prereqs/BUS-CROSS-REPO.md.
• Returns a report with architecture, flows, dependencies, and next-step recommendations.

Step 1.5: Log CodeGraph telemetry (silent)

After the sub-agent completes, run once (do not mention it to the user unless it fails):

refacil-sdd-ai compact log-codegraph-event --skill investigator --has-graph  --tool-calls  --tokens 

• --has-graph: the codegraphAvailable value from Step 0.2.
• --tool-calls: number of codegraph_* tool calls the sub-agent made (0 if it did not use the graph).
• --tokens: conservative estimate of tokens saved vs. reading whole files (~800–1500 per useful tool call; 0 if no graph or no calls).

If the command fails, ignore it; it must not block the flow.

Step 2: Present the report

Show the user the full report returned by the sub-agent. There are no artifacts to write — exploration is purely analytical.

If the sub-agent asked for clarification (because the prompt did not carry an explicit question), propagate the question to the user.

Step 3: Next step

The sub-agent already includes recommendations at the end of its report. Apply the natural continuity rule from METHODOLOGY-CONTRACT.md §5:

• If the report converges on one single next step (e.g. the finding clearly indicates a new feature or a bug), close with the single formula:
• "The next step is [description]. Do you want me to continue with /refacil:propose?" (or /refacil:bug, as appropriate).
• If there are multiple valid paths, list them numbered and ask the user to select.

Rules

• Always delegate to the sub-agent. Do not replicate the exploration logic here.
• Do not invoke without a question. If $ARGUMENTS is empty, ask for the question first.
• Do not write files. Exploration is read-only end-to-end.
• CodeGraph gate (Step 0.2): if mode is enabled/per-repo and codegraphAvailable is false, the GATE is mandatory — never skip it. If the user replies affirmatively ("yes", "si", "sí", "dale", "ok", "go"), immediately execute /refacil:update without any prior text. Do not describe it. Do not ask again. (See METHODOLOGY-CONTRACT.md §5.)
• Flow continuity: if the user confirms affirmatively ("yes", "ok", "go", "continue", etc.) the continuity question in Step 3, immediately execute the resolved /refacil: command. Deterministic resolution:

| Finding | Skill | |---------|-------| | New feature or improvement | refacil:propose | | Functional bug or production error | refacil:bug | | Missing initial repo configuration | refacil:setup | | Flow doubt or next command unclear | refacil:guide | | No clear match | Do NOT invoke — list numbered options and ask for explicit selection |

Do not describe it in text or wait for the user to type the command. (See METHODOLOGY-CONTRACT.md §5.)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Erikole21
• Source: Erikole21/refacil-sdd-ai
• License: MIT

Install and usage instructions live in the source repository linked above.