Ss Troubleshooting Workflow

Troubleshooting Workflow

Chains the existing ss-* skills into a single diagnose-fix-deliver flow:

ss-inspect → [Gate: root-cause confirmation] → ss-create-branch
→ ss-plan → ss-coding (built-in review) → review-acceptance loop → ss-create-pr

Core principle: thin orchestration. This skill never writes code or drafts a plan itself. It calls the other skills in order, pauses at the gates below, passes each step's output to the next, and reports progress.

Review is not run twice. ss-coding already enforces a post-coding review internally and returns a verdict. This workflow never invokes ss-code-review separately — the review-acceptance loop below is driven entirely by that verdict.

Unlike ss-feature-workflow, the root-cause report from ss-inspect replaces the proposal, and root cause is confirmed before any branch is created — worth fixing is established before code is touched.

When to Run

This is heavy, end-to-end orchestration: it runs a root-cause investigation, creates a branch, spawns coding agents, runs reviews, and delivers the fix. Only run it when the user explicitly asks for this skill or clearly wants the full diagnose-to-delivery pipeline. For a plain question or a diagnosis-only request, do the work directly (or point to ss-inspect for diagnosis alone).

Inputs

• Problem source — an issue link, alert ID, or plain-text problem description. If missing, ask for one.
• Delivery mode — full (default) or lite. See "Delivery Mode" below.
• Investigation tier — auto (default), L0 Local, L1 Runtime, or L2 Systemic, forwarded to ss-inspect. An explicit higher tier is honored; an explicit lower tier is only a hint and must be promoted when its evidence conditions or risk boundary fail.
• Skip-gates — proceed through the root-cause confirmation gate without pausing (not recommended — see Iron Rule 2).
• Decide-autonomously — resolve ambiguous choices without asking.
• Worktree preference — forwarded to ss-create-branch.

Delivery Mode: full vs. lite

The quality gates — root-cause confirmation, TDD, multi-agent code review, verification — are identical in both modes. Only the start and finish of the fix change:

Mode resolution: an explicit mode input wins; otherwise ask once at the start and use that answer for the rest of the run. Never persist the choice to a project config file.

Multi-Repo Routing

Cross-service defects — service A errors, root cause lives in service B, or the fix spans both — are a frequent multi-repo case. This workflow checks at three points. Full rules in ../ss-references/multi-repo-detection.md (this table is a condensed copy; the reference file wins on conflict).

Judgment points:

1. Entrance, before pre-flight — scan the issue/alert text for the heuristic signal. On a hit, ask (or decide autonomously) and hand off with the original input.
2. Root-cause gate — the most reliable point: by the end of the investigation, affected repos are part of the evidence chain. Once the user confirms a complete report whose repository list has more than one entry, hand off with that report as the planning input. Single-repo continuation is forbidden after confirmation.
3. After build-plan — a safety net for initially single-repo reports: if ss-plan's scope check detects a cross-repo span and produces a master plan, hand off with it. If a previously confirmed multi-repo report reaches planning and the plan omits one of its repositories, stop and correct the plan instead of downgrading scope.

On hand-off: pass the original input, the root-cause report and any other artifacts, and the active automation flags to ss-multi-repo-workflow; print one line explaining the hand-off and terminate this workflow.

Iron Rules

Violating any of these means stopping and explaining to the user:

1. The orchestrator does no real work — only call other ss-* skills; never edit source or draft plans directly.
2. The root-cause-confirmation gate is on by default and critical — a wrong root cause causes full rework, so a human must confirm it (skip-gates mode can bypass this, but it isn't recommended).
3. Design understanding precedes repair design at every tier — before ss-inspect may write a permanent repair proposal, it must explain the complete relevant design and implementation for the evidence-derived scope: responsibilities/ownership, end-to-end flow, dependencies, invariants, existing mechanisms, healthy analogs, and all affected paths. Tiering may shorten the account but never skip this gate or authorize a patch-style fix.
4. Pre-flight checks ask, they don't abort — apply the autonomous default only when deciding autonomously.
5. No gate for review — run the loop instead — see "Review-Acceptance Loop". There is also no gate before delivery.
6. Resumable by identity, never recency alone — reuse only artifacts explicitly linked to the current **Problem ID:** and their exact upstream artifact.
7. Investigation completeness is non-skippable — skip-gates may bypass the human pause, but it never permits anything below Confidence: HIGH, an unjustified investigation tier/evidence budget, unresolved promotion signals, material evidence gaps, an incomplete design-understanding gate, missing canonical **Affected Paths:**/**Repositories Requiring Fix:** bullet lists, or an incomplete repair and verification plan (including canonical **Violated Invariant:**, **Owning Boundary:**, and **Patch-Style Alternative Rejected:**) to proceed.
8. Never relax safety guardrails — automation flags only affect questions and approvals; they never bypass the hard rules built into the underlying skills.

Pre-flight Checks

Run the source/delivery checks once on start. Evaluate branch/worktree readiness only after a complete report is confirmed, immediately before Step 4. Any unmet check defaults to asking the user, not terminating.

Whether the investigation needs environment/observability access is decided by ss-inspect itself — pure code analysis needs none of that.

Resume Detection

Process

1. Pre-flight and resume.
2. ss-inspect — produce a report-only investigation: derive a stable **Problem ID:**, classify the evidence/risk boundary as L0 Local, L1 Runtime, or L2 Systemic, and gather only the tier-required discriminative evidence while promoting immediately when scope or risk expands. Verify the root cause with at least two independent sources, scope understanding by potential blast radius, reconstruct the relevant design/invariants, populate the canonical **Affected Paths:** and **Repositories Requiring Fix:** bullet lists, and design the permanent repair plus verification. It must not edit code/configuration, mutate runtime state, apply mitigation, create a branch, or invoke coding.
3. Design-understanding completeness check, then root-cause confirmation gate — first require a justified tier and evidence budget with no unresolved promotion signal, Confidence: HIGH, no material evidence gaps, and a passed design-understanding gate covering responsibilities, ownership, end-to-end flow, dependencies, invariants, existing mechanisms, healthy analogs, affected paths, and explicit rejection of patch-style repair. Missing evidence or impact-scoped understanding always returns to investigation, even under skip-gates. Then show the complete conclusion and evidence; pause for continue to fix / re-investigate / abort. Skip-gates may bypass this human pause only after completeness passes. Cap re-investigation at 2 rounds; if the report is still incomplete, stop with the evidence gaps and do not offer "continue anyway." Once a complete report is confirmed, its repository list is authoritative: more than one entry hands off immediately to ss-multi-repo-workflow before any branch is created.
4. ss-create-branch (full mode only) — branch from the Problem ID, prefix defaulting to fix/. Forward the worktree preference. Record the identity linkage, branch name, and worktree path.
5. ss-plan — draft the fix plan from the exact investigation report (including any spec delta). The plan must restore the violated invariant at the correct ownership boundary, cover every **Affected Paths:** bullet, and include compatibility/migration, rollback, observability, and regression verification. If it produces a master plan, hand off to ss-multi-repo-workflow and stop. If it drops a repository from a confirmed report, correct the plan before continuing.
6. ss-coding — execute the fix plan. It already covers TDD and test verification, and enforces its own post-coding review before returning a verdict.
7. Review-acceptance loop — adjudicate findings and feed valid fixes or invalid-finding reasons back into ss-coding until its fresh built-in review returns literal APPROVED, or escalate.
8. Delivery (full mode) — call ss-create-pr, referencing the original issue in the PR description; report the link, plus a reminder that ss-cleanup removes the worktree and branch after merge. (lite mode) — commit with a conventional-commits message referencing the issue (push optional), then print a summary.

Review-Acceptance Loop

After ss-coding returns its verdict, this workflow sets no gate and does not pause:

1. Read the verdict.
2. No unresolved findings (approved) → go to delivery.
3. Unresolved findings remain → judge each one: valid goes on the to-fix list; invalid (false positive, out of scope, by design, or suggestion-only) is recorded with a reason and left unfixed.
   • Scope-reduction findings get no leniency: a finding reporting unimplemented requirements, stubbed logic presented as complete, or scope silently deferred ("phase 2", "MVP", "simplified for now") must never be judged invalid as "by design" or "out of scope" unless the user explicitly approved that scope adjustment earlier — cite that approval.
4. All judged invalid → pass the complete adjudication record back to ss-coding and require a fresh built-in review of the current diff. Only literal APPROVED proceeds to delivery; if it cannot be obtained, escalate instead of rewriting the verdict.
5. Valid findings remain → call ss-coding again for just those findings → back to step 1.

Convergence protection:

• Cap workflow-level rounds at 3, independent of ss-coding's internal rounds; if exceeded, escalate with the latest verdict plus the fixed/unresolved lists.
• Track recurrence at the workflow level: the same finding judged valid across 2 consecutive workflow rounds after being "fixed" is a stall — escalate.
• Record every "invalid" judgment and its reason. Never force a finding to "invalid" just to pass.

Edge Cases & Error Handling

Examples

Run ss-troubleshooting-workflow on https://your-tracker.example/issues/PROJ-7150
Run ss-troubleshooting-workflow with "staging: user login returns 500", deciding autonomously
Run ss-troubleshooting-workflow on ALERT-20260626-001 in lite mode